# TopFlight — The Concept A complete account of TopFlight: what it is, what a person does on it, what pays them and why, the machinery that makes a market of this shape possible at all, and the choices still open inside it. It is built to be read at whatever depth the question needs, and the structure is the same at every level: a short version, then the parts it is made of, each of which opens with its own short version. Every claim sits at a stable address, every concept has exactly one canonical home, and nothing is restated on the way down; where a section relies on an argument made elsewhere, it points at the address that carries it rather than repeating it. **Two registers run through this document.** Most of it describes a design, and it is written in the present tense of that design. The mechanisms are decided; a number of the parameters inside them are still being tuned, and each of those is marked a **dial** and given with the range it lives in and what moving it costs, never with a preferred answer dressed up as a finding. The dials are collected in TF-11, which explains what tuning means here, and they are cross-referenced from wherever they bite. --- ## The Short Version **TopFlight is a price on every English football club, and holding a club pays you every day at a rate set by where that club stands in the Premier League table.** Six steps are the whole product. Buy tokens in a club. The tokens earn FOOTY every day from the moment they are yours. A club's place in the table sets what its tokens earn, each place up paying about 23 percent more than the place below, so first pays fifty times last. Results change the table and the table changes the rates. A club paying more FOOTY attracts buyers, so its price rises. Sell whenever you want, into a live price that is always there (TF-2). **What a club token is.** Every club is a slice of one closed set, priced together on a single curve, and the prices sum to exactly one dollar because a complete set of every name on the board merges back into exactly one dollar of collateral. The merge is worth pinning at first use, because it is not an action anybody takes: a Long and its matching Short arriving in one account collapse into collateral in the same transaction, with nothing to call and nothing to close. It is an accounting identity enforced before every state transition rather than a relationship anyone maintains. So a club's price is its share of the argument about where the season ends up, and the total those tokens divide between them is pinned by construction rather than arrived at by adding up opinions. Nothing settles, nothing expires, and no date exists on which anything falls due (TF-9.4, TF-8.4). **What decides the earning is a public fact nobody here owns.** The referent is the published Premier League table. All twenty places pay, on a fifty-times ladder from champion to twentieth, and a club outside the Premier League earns nothing without anyone writing a rule saying so, because it is not on the twenty places (TF-4). The table reaches the chain by bonded assertion: anyone may post the weekend's points under a bond, anyone may challenge inside a window, and an unchallenged post becomes the table in force. The referent is checked every weekend by millions of people who care intensely and are looking anyway, which is a stronger honest-assertion equilibrium than the ones optimistic oracles normally rely on (TF-5). **What the earning is worth is a market's answer, not a promise.** FOOTY is issued from a tank filled once at genesis, draining a fixed fraction of whatever remains every period, halving every four years. Nothing is ever promised in dollars, so there is no figure to renegotiate on holders when a season is quiet. The board's own income, trading fees plus the interest earned on the collateral sitting in the market, buys FOOTY back from its own pool; most of what is bought returns to the tank and comes out again through the ladder, which is what makes trading income something you collect by being right about football (TF-6, TF-7). **And a token is a membership as well as an asset.** Every club has a room, and holding enough of that club is the door. Everyone inside it is long the same club, has money on it, and is watching the same result, so the room is aligned by construction rather than by moderation, and selling below the threshold costs a member the room rather than costing them a penalty. One holding, two uses, and no second token for either (TF-2.8). **Why a market of this shape can exist at all.** Football money elsewhere is bets that expire and fantasy points that reset, because a market that pays you for standing, forever, needs machinery that had to be built. Five things had to be true, each standing on the one before it. Close the set: ask a comparative question over a fixed exhaustive list and one dollar buys the whole answer, because one club has to finish first and one has to finish twentieth. Then: **coherence**, a many-outcome market held consistent by accounting rather than by arbitrage, with positions stored as exposure rather than as tokens, and a short as a native first-class position (TF-9.2, TF-9.3). **One market maker across the whole partition**, pricing every outcome on one curve at a per-trade cost that does not grow with the number of outcomes (TF-9.7). **Perpetuity**, the settlement date removed entirely (TF-8.4). **Internal synthetic capacity**, which is what lets a full board open at its declared depth with no external liquidity provider and no funded market-maker inventory, and the solvency argument for it is exact (TF-9.8). And **the emissions machine**, which connects a market that never settles to the world through revenue instead of through settlement or a funding rate (TF-8.6). **The strongest of those is the fourth.** A board of a hundred and sixty-five names, coherent, perpetual, and liquid at full declared depth from the first second, with nobody having posted capital to make it so. The argument is short enough to state here: a Long and its Short are a complete set worth exactly one dollar; acquiring both sides through the curve costs one dollar before fees at any price and in any order; the ledger escrows one real dollar for every outstanding pair, and that dollar necessarily arrived as some trader's payment. So a profitable exit is funded, dollar for dollar, by the payments of the traders who moved the price, and the market can never owe more than it has taken in (TF-9.8). **And what is still being tuned.** The mechanisms above are decided; a handful of the numbers inside them are being calibrated, which is the work that comes after a design rather than before one. The fee level. The emission schedule and its half-life. The bond size and the challenge window. The room threshold. Whether what we take should be taken in FOOTY or in dollars. Two of the entries are genuinely a choice between two designs rather than a number moving: what the fee is for, and whether a member's idle cash earns anything or whether paying it nothing is what pushes money into positions and prices. And one quantity is not ours to set at all, being how much of the FOOTY that reaches people gets sold, which is an outcome rather than an input and is the single most consequential unknown in the model. TF-11 gives each one its range and what moving it costs. **One thing the design opens is deliberately not part of it.** These tokens are unusually good collateral, because every price is bounded between zero and a dollar by construction and a complete set is worth exactly a dollar by identity, so lending against them and looping the proceeds is an obvious thing for somebody to build. It would be a separate market and nothing described here depends on it: the board carries no obligation denominated in money and does not acquire one because a person borrows elsewhere (TF-13). --- ## The Map Every section carries a stable short address, **TF-n**, with subsections nesting: TF-6.2 is the second subsection of TF-6. Cross-references use those addresses throughout, so any claim can be followed to the section that carries it. The document is a reference map rather than an argument in sequence: each concept has exactly one canonical home and later mentions point at it, so a section read alone may send you elsewhere for the argument behind a claim. - **TF-1 — Orientation.** What the product is, and what is distinctive underneath it. - **TF-2 — The product in six steps.** Buy (2.1); earn every day (2.2); the place sets the rate (2.3); results move the rates (2.4); the yield moves the price (2.5); sell whenever (2.6); what a token actually is (2.7); the supporters rooms (2.8). - **TF-3 — The board.** The roster and Other (3.1); why no mechanism is needed (3.2); the roster against the screen (3.3); shorts, and the basket index (3.4). - **TF-4 — The referent and the ladder.** Why football (4.1); the twenty places, and why the ladder is steep (4.2); the rules that fall out of paying a published table (4.3); the general rule, and the two forms a rubric can take (4.4). - **TF-5 — Getting the table on chain.** Bonded assertion (5.1); what the log checks, and what a wrong table costs (5.2); what is still being tuned in it (5.3). - **TF-6 — FOOTY.** The water tower (6.1); the split, and when the cut is taken (6.2); the buyback loop (6.3); the pool, and the two exits (6.4); what the yield is actually backed by (6.5). - **TF-7 — The money.** The fee and its form (7.1); the interest on float, and which leg is larger (7.2); the unit of account (7.3); who loses (7.4). - **TF-8 — Why a market like this can exist.** The closed set as the opening move (8.1); coherence (8.2); one maker across the partition (8.3); perpetuity (8.4); internal synthetic capacity (8.5); the emissions machine as a general object (8.6); how they compound (8.7). - **TF-9 — The technical account.** The representation problem (9.1); exposure as the primitive (9.2); native shorts (9.3); the merge and the complete-set identity (9.4); the internal dollar (9.5); the ledger as counterparty (9.6); constant-cost pricing across the partition (9.7); the solvency argument in full (9.8); the contrast, in a table (9.9). - **TF-10 — Why it is not built on an existing conditional-token framework.** Three things, and the honest scope of the claim. - **TF-11 — The dials.** The fee (11.1); the emission schedule (11.2); the split, and whether we take FOOTY at all (11.3); selling pressure and locking (11.4); depth (11.5); the opening board and the price floor (11.6); the bond and the window (11.7); the pool (11.8); whether idle cash earns, a fork (11.9); the room threshold (11.10); and what it costs to leave a dial untuned (11.11). - **TF-12 — The precedents, and the questions no design settles.** - **TF-13 — What the design opens, and does not depend on.** Why the collateral is unusually good (13.1); the loop (13.2); where the risk sits (13.3); the questions it would raise (13.4). - **TF-14 — Key definitions.** --- ## TF-1 — Orientation **The product is TopFlight**: a price on every English football club, on one board that sums to a dollar, where holding a club earns FOOTY every day at a rate set by that club's place on the real Premier League table, and the board's own income buys that token back. Six steps are the whole of what a person does (TF-2). Everything else in this document is either what makes that possible (TF-8, TF-9, TF-10) or what has not been decided (TF-11, TF-12). **What is distinctive underneath it**, stated as facts and argued at the addresses given. Positions are held as exposure vectors on a ledger rather than as minted outcome tokens, which makes a short a first-class position from the moment a name exists and makes the complete-set identity an enforced invariant rather than an arbitrage relationship (TF-9.2 to TF-9.4). One market maker prices every outcome of the partition on a single curve, at a per-trade cost that does not grow with the number of outcomes, which is what makes a board the width of a football pyramid practical rather than theoretical (TF-9.7). The board opens at its declared depth with no external liquidity provider and no funded market-maker inventory, and that is the strongest claim here, argued step by step so it can be checked rather than believed (TF-9.8). And the emissions machine ties a market that never settles to the world through revenue rather than through settlement or a funding rate, which is the mechanism that lets "your team won" turn into money (TF-8.6). **One boundary, stated because it is deliberate.** The prices on this board are not a forecast of a settlement, because there is no settlement. They are the standing answer to a standing question, and what makes them mean something is that a place on the table pays: every position carries an earning rate anyone can compute from the published ladder, the published table and the token's price, and a price is what traders will pay for that earning (TF-2.5). --- ## TF-2 — The product in six steps The six steps are the product, and they are what a person is told. Each is a rule, and the rules are meant to be blunt. One worked example runs through all six, at one club: Brighton, tokens at 4.6 cents, Brighton tenth on the day of purchase. The steps carry no accounting. What a token of a club actually is sits at TF-2.7, and the machinery under that is TF-8 and TF-9. ### TF-2.1 — Buy tokens in a club **Every club has a live token price.** Put in a hundred dollars at 4.6 cents and you own about 2,174 Brighton tokens. The price is what the market currently pays for a token of that club, and trading is the only thing that moves it (TF-2.5). Those tokens were not taken from anybody: buying creates them, so there is no fixed pool of Brighton tokens being allocated among buyers and nobody is competing for a slice of a finite number (TF-2.7). There is no queue, no counterparty to find, and no minimum: the board's market maker quotes at every size at every state of the board, so a buy always fills, at a worse average price the larger it is, which is price impact rather than a limit on how much can be bought (TF-9.8). Prices are quoted in cents of a dollar because the board sums to one dollar. A club at 4.6 cents is a club the market currently gives 4.6 percent of the argument to. ### TF-2.2 — Earn FOOTY every day **The tokens earn FOOTY from the day they are yours.** Accrual runs continuously, second by second, out of tokens held, time held, and the published table, whether or not the holder does anything. Waiting costs nothing, because what has accrued is kept rather than forfeited; receiving the tokens takes a claim, which is a step anyone can submit and which needs nobody's permission and nobody's signature. Brighton are tenth on the day you buy, so the tokens earn at the rate for tenth. FOOTY is the board's own token. It has a price of its own on an open pool, and the board's income buys it back (TF-6.3). What a token earns per day in dollars is therefore the ladder, the table and the FOOTY price together. The board itself displays the earning rate without the FOOTY price in it, as FOOTY a day per $100 of a club at today's price, which orders the clubs identically and states a rate for today rather than a return over a year. ### TF-2.3 — Higher clubs pay more **A club's place in the table sets what its tokens earn, and each place up pays about 23 percent more than the place below.** Twenty places, a constant step of 1.2286 between them, so the champion's place pays fifty times what twentieth's pays. Brighton go top, and the rate on Brighton tokens is more than six times what tenth was paying. The ladder is a fixed published table of tokens summing to exactly one, and every figure quoting it has to name its base, because there are two. Each period's emission splits eighty percent to holders and ten percent each to two other streams (TF-6.2), and the ladder divides **the holders' block**. So the champion's place takes 18.9 percent of the holders' block, which is 15.1 percent of the period's whole emission, and twentieth takes 0.38 percent of the block and 0.30 percent of the emission. A relegation six-pointer moves real earning that weekend. Why the ladder is as steep as it is, and why it is the one authored number in the design, is TF-4.2. ### TF-2.4 — Results change the rates **When a club moves position, the rate its tokens earn moves with it.** Brighton win, the new table comes into force, and the rate on Brighton tokens goes up. The published standings are confirmed onto an append-only on-chain log, and accrual attributes earning from each table event to the next, so a rate change is a table change and nothing else (TF-5). This is the point where a real-world event becomes money in an account, and it is the answer to the question a market with no settlement usually cannot answer: what real thing happening makes me money. Your team won. ### TF-2.5 — Traders chase the yield **A club paying more FOOTY attracts buyers, and buying pushes the price up.** Brighton at first are the best yield on the board, buyers arrive, and the price moves from 4.6 cents to 9.2 cents. That is the whole connection between the earning and the price: the table decides what a place pays, and the market decides what being paid that is worth. Two consequences a reader should be given rather than left to discover. **The price moves continuously and the rate steps.** At three o'clock on a Saturday ten matches kick off, goals go in, and the board reprices the title race and the relegation places while the football is still being played, hours before any table is posted, driven by the same people watching the matches. What the table decides is the earning, and that steps when the table posts. **And a club that does not move on the table keeps the same earning while its price still moves**, because the price is a forecast of where the club will finish, and the forecast changes on a hamstring, a manager, a transfer window or a European run draining a squad. There is something on this board on a Tuesday in November, and in July. The misreading this invites, and careful readers make it every time, is that results are known on Saturday evening and the table posts later, so somebody buys the club about to climb before the price knows. That assumes the price is a quote sitting still until an oracle updates it. It is a curve that moves on every trade, so by the time a table is posted the market has been repricing for hours and a late arrival finds the standing opinion of everyone who acted sooner. What the maker does carry is that whoever trades first on new information trades against the price the last trade left behind, which is adverse selection, is real, and is paid by every automated market maker in existence. Where the money for that comes from is TF-9.8. ### TF-2.6 — Sell whenever you want **There is a live price on every club at all times, and a holder sells into it whenever they like.** Sell the 2,174 Brighton tokens at 9.2 cents, and the FOOTY those tokens earned along the way is yours as well. Nothing has to finish, nothing has to be waited out, and no counterparty has to be found. **Two exits exist and they are different objects**, which is the confusion this step is here to prevent. Leaving a club is not a pool trade: a position sells back to the board's own market maker, which quotes at every size at every state of the board, so the depth of the FOOTY pool has nothing to do with whether a holder can leave a club. The pool is where FOOTY is sold, so it governs what a holder gets for the emission they have earned, and nothing else. Read the pool's size as a limit on leaving a position and it looks alarming; read it as what it is, the market for the reward token, and it is the ordinary question of whether a young token's market is deep enough to absorb the selling that a young token's emission produces (TF-6.4). ### TF-2.7 — What a token actually is A token of a club is a slice of a closed set. The question the board asks is exhaustive, so a complete set of every name on the board merges back into exactly one dollar of collateral, and that merge is why the prices sum to a dollar: the total is pinned by construction rather than arrived at by adding up valuations. Merging is not something a holder does. A Long and its matching Short arriving in the same account become collateral in the same transaction, automatically, with no step to take and no position left over (TF-9.4). No club pays a prize and nothing settles, so there is no terminal value a holder is waiting for and no date on which anything falls due. What separates one club from another is therefore what it earns. A club's price is its share of the emission still to come, as traders currently price it, and the merge is what fixes the total those tokens divide between them. Because that total is fixed, buying a lot of the board is a purchase at fair value rather than a way of taking anything out of it (TF-8.1). Which is why the forecast that matters is where a club will sit: the place is what changes the earning. **There is no supply figure for a club, and there is not going to be one, because a token count is not a property a club has.** Tokens are minted on demand against collateral and destroyed again when a complete set merges: buying creates them, completing a set releases the dollar and takes them out of existence. So the number of a club's tokens outstanding at any instant is a consequence of what has been traded rather than a fact about the club. There is no genesis supply, no issued quantity, no cap and no allocation, and nobody chose the number, because nobody sets it. A reader arriving from equities or from a token with a fixed mint will reach for the token count as the thing that anchors everything else, and here there is nothing at that address: what anchors everything is the dollar a complete set is worth, which does not move. **One consequence, since it is the reason the question feels unanswered rather than dissolved.** Two people can ask how many Brighton tokens exist and get different true answers a minute apart, and neither of them learns anything about Brighton by asking. What a holder owns is a fraction, and the denominator is whatever trading has left. **And FOOTY is the opposite object, which is worth keeping apart from this.** Club tokens have no supply because they are claims minted against collateral. FOOTY has a hard cap, a single genesis fill and a published schedule, because it is the thing being paid out rather than a claim on anything deposited (TF-6.1). The two words look alike on a screen and are different in kind. Positions are held as tokens in a wallet and move between wallets like any other token. A user is one wallet, either an embedded one or one they hold the keys to, and nothing is ever summed across wallets: the identity is the wallet. Somebody who starts on an embedded wallet and later wants their own keys sends their tokens there and signs in with that address, and the identity follows, with no linking flow and no migration. One wrinkle travels with that: future accrual follows the tokens by itself, while FOOTY already accrued and unclaimed sits against the old address and is claimed from there. ### TF-2.8 — The supporters rooms **Every club has a room, and holding enough of that club is the door.** One room per club, entry gated on holding at least some threshold of that club's tokens, checked against the holding itself rather than against a list somebody maintains. Stop holding enough and the door closes again. How much is enough is a dial, being tuned rather than argued about (TF-11.10). **What that produces is a room aligned by construction rather than by moderation, and that is the whole point.** A general football forum is full of people arguing, because anyone can be in it and the only thing members have in common is the subject. Here every member of a club's room is long the same club, has money on it, and is watching the same result on the same afternoon. Nobody had to write a rule about what the room is for. The room is for what its members have all already paid to be exposed to, and the alignment is a property of the door rather than of the rules. **A token is therefore two things at once.** It is a yield-bearing asset, and it is a membership. One holding, both uses, and no second token issued for either: there is no membership pass to mint, no separate community token, no allowlist to run. The thing that pays you is the thing that lets you in. **That gives a position a reason to be held that is not financial**, which matters more than it sounds. Everything else in this design that discourages selling is monetary and therefore competes with the yield on its own terms. This does not. A member who sells below the threshold loses the room, and what they lose is social rather than financial: the conversation they were in, on the afternoon it matters. That is a holding incentive with no lockup, no vesting, no cliff and no penalty clause anywhere. Nothing is confiscated and nothing is forfeited. You simply are not in the room any more, and you can come back by buying back in. **And the door needs no identity.** It does not matter who a member is, only that they hold, so the room is resistant to somebody arriving a thousand times without anyone having to verify anything about anyone. The cost of entry is capital at risk in the thing the room is about. **The room's population is itself a live number worth reading.** The number of members is the number of people holding enough of that club to care, which is a different and in some ways better measure than price: price tells you what the marginal trader thinks, and the room tells you how many people are committed at size. It moves on results, and it is visible without anyone reporting it. **It also gives the product the moment it otherwise lacks.** Accrual runs continuously, second by second, which is the correct mechanism and a poor spectacle: nothing visibly happens at any particular instant. Full time on a Saturday is the spectacle. A match finishes, a club moves, and what every holder of that club earns changes together, at the same moment, for a reason they were all already watching. The room is where that lands. **One property falls out of the pricing and is worth noticing.** The threshold is denominated in the club's own tokens, so a room's entry cost is that threshold times that club's price. A cheap club's room is cheap to be in and a dear club's room is dear, without anybody setting a price for either. The door prices itself off the board. **The door is directional, and only Long holders pass it.** A Short holder does not enter a club's room, whatever quantity of that club's tokens the position technically contains. **The reason is the room's only property.** Everyone inside wants the same result. A Short holder is not acting in bad faith and is not a problem to be defended against; they simply want the opposite result. One of those in the room and the property is gone. So the door filters on DIRECTION, and it does not have to be clever about anything else: not conviction, not size beyond the threshold, not who anybody is. **And the rule has to be stated rather than left to fall out of the mechanics, because the mechanics would let them in.** A Short on a club is one position consisting of one token of every OTHER club (TF-3.4), so a large enough short clears the threshold on many clubs at once, and it passes every economic test the door would otherwise apply: those tokens are really held, they really earn those clubs' rungs, and the holder really loses money if those clubs fall. It is not even a cheap way in, because a Short costs the sum of every other club's price. The position is honest. What it is not is a person who wants that club to win. The room's purpose and the room's arithmetic disagree here, and the purpose is what the door is for. **On where the room physically lives, this document deliberately says nothing.** A room can be a surface inside the product or a channel on a chat platform, and the mechanism above is identical either way, because the only thing doing work is the holding check at the door. The venue is a product decision with real consequences for reach and for who has to run it, and it is not a property of the design. --- ## TF-3 — The board The board is the whole English professional pyramid priced as one object: **a hundred and sixty-four clubs plus Other, a hundred and sixty-five positions**, six tiers deep, all of them listed from the start. That shape is what makes promotion a price move rather than an event anyone has to handle, and it is only affordable because a board of a hundred and sixty-five names costs no more to open than a board of twenty (TF-8.5). ### TF-3.1 — The roster, and Other **The roster is a hundred and sixty-four clubs, fixed at genesis, and it never changes.** Six tiers of the English pyramid, which is every club with a plausible route to the top flight, all of them listed on day one. A club that goes up changes which rung it sits on; a club that goes down keeps trading and stops earning. Nothing is ever listed later, because there is no club worth listing that is not already listed, so promotion is a price rising on a club that was already on the board and relegation is a club leaving the paying twenty. There is no carve, no credit, no migration and no special case anywhere in the system. The depth is set by the only precedent that exists. Luton Town were in the fifth tier in 2009 and in the Premier League in 2023, the first English club to make that climb. Six tiers is one level deeper than that. **Other is the hundred and sixty-fifth position, and it is permanent.** It is the bucket for everything below the listed clubs, held and priced in aggregate, and it is what lets a finite list still sum to a dollar. It is a position like any other: it can be bought and sold at curve prices, and it earns nothing while no unlisted club sits in the Premier League. If such a club ever did reach the top flight, Other holds that place's rung and earns it exactly as a listed club would, and nobody has to do anything about it. **Other never splits.** It does not carve into a promoted club, it does not fold anything back, and there is no path anywhere in the design by which a name comes out of it. It is not a placeholder for a future listing and not a staging area for anything; it is the residual, and it stays the residual. **And that is the argument for the whole shape, so it is worth stating rather than leaving implicit.** Every mechanism ever proposed at this point in the system, the carve, the split, the fold-back, existed to compensate for a board that was too narrow to contain the thing that was about to happen. A wide board plus a permanent residual needs none of them. Simplicity is not a side effect of the decision here; it is what the decision buys. **And listing the pyramid is what makes the cheap end interesting.** A club four divisions down earns nothing today and costs a hundredth of a cent. If it climbs it lands on a rung, and the rungs at the top are fifty times the rungs at the bottom. So the lower divisions are a board of long-dated options on a rise through the leagues, priced in tenths of a cent, which is a reason to hold a club that has nothing to do with yield and everything to do with football. **Other is also the general closure device**, and this is the property that makes the machine point at things other than football. A closed set is the precondition for everything (TF-8.1), and most published rankings are not closed: a world ranking runs to a thousand names and has no last place. Other closes it in one move. Name as many places as are interesting and bundle the remainder into a single residual, and "the top ten, plus Other" is exactly as closed, exactly as exhaustive, and exactly as much a complete set worth one dollar as "twenty clubs, plus Other" is. Nothing has to be invented to make the set closed, and nothing has to be excluded to keep it closed. ### TF-3.2 — Why no mechanism is needed **Count the things that do not exist here, because the absences are the design.** There is no listing process, no admission criterion and nobody deciding which club deserves a market. There is no carve, so no club is ever created out of the residual and no price ever has to be assigned to a new name. There is no migration, so nobody's position is ever moved from one instrument to another. There is no promotion event and no relegation event: a club going up is a rung changing hands, and a club going down is a club that stops being on the twenty places (TF-4.2). And there is no season rollover to administer, because the ladder pays the table in force and accrual never stops. **Each of those absences replaces a mechanism that was designed and then thrown away**, and they were all thrown away for the same reason. A narrow board has to do something when reality produces a name it does not contain, and everything it can do is complicated: carve the name out of a residual and you must assign it a price, which means somebody sets a price after the opening, which is the one thing the design is built never to do (TF-8.6). Split a residual and you have to decide what happens to everyone holding it. Fold a name back and you have to decide what its holders get. A wide board never meets the situation, so it never needs the answer. **What that leaves is a board that is authored once and then simply runs.** Two numbers are set at genesis, the steepness of the ladder and the opening prices, and after that every price on the board is the market's and no human decision is pending on any date. That is what makes the machine pointable at something other than football: an instance is a roster, a referent and a rubric, and none of the three needs an operator standing behind it (TF-8.6). The one thing that is genuinely still open in this area is not the list. It is how the tail is PRICED at the opening, and how low the price floor under a listed club should sit, which is TF-11.6. ### TF-3.3 — The roster and the screen are different objects The roster has to be exhaustive for the accounting to work at all, and most of it earns nothing while it sits outside the top flight. The interface is curated: the Premier League and a handful of others, with the rest reachable rather than displayed. A board full of rows that pay nothing is plumbing, and showing it is not a product decision anyone is obliged to make. ### TF-3.4 — Shorts, and the basket index **Shorts are native, and the word carries a different object here than it does elsewhere, so it is worth pinning before anything leans on it.** A Short on a club is ONE position, bought outright and owned outright, and what that position IS is one token of every other club on the board. Nothing is borrowed to hold it, so nothing is financed and nothing accrues against it. There is no matching long side to fund and no second position to close. And the club you shorted pays you nothing rather than costing you something: its rung goes to the people who hold it, and you are simply not one of them (TF-9.3). Everything else follows from that. You earn each of the other clubs' places through the ladder, alongside every other holder of those clubs. There is no liquidation, no funding rate and no expiry, because there is no borrowed thing to call back and no date on which anything is due. **Which is how you buy the whole board in one trade.** Shorting the club that pays LEAST is very nearly owning the entire board: every club except the one on the bottom rung, all of them held in a single position. It costs a dollar minus that club's price, paid in full at purchase and never topped up. And it earns almost all of the emission the board pays out, because the only club left out is the one paying least of all. **So what the position earns is not a spread.** There is nothing to subtract from it. No financing is being paid, because nothing was borrowed. No short-side rate is being paid, because the club you did not buy pays its own holders and you are not among them. What arrives is simply the emission of the clubs you hold, which is nineteen of the twenty paying places. **What that gives a holder is a yield on the whole board without picking anybody.** Take no view, buy one position, earn roughly what the board earns on average per dollar. That is an index, and it is the baseline anyone can have for declining to have an opinion. It also gives every opinion something to be measured against, because a club is only worth holding instead if it pays more per dollar than this does. **And nothing had to be built for it.** There is no basket product, no wrapper, no fund, no rebalancing and no second contract. The index position falls out of two properties that are already there for other reasons: the board is conserved, so the complement of one club is exactly the sum of all the others (TF-9.4), and a short is a single first-class state entry rather than a portfolio somebody assembles and maintains (TF-9.3). Put those two together and the index is simply one of the positions the board already offers. Everywhere else an index is a product somebody has to construct and charge for; here it is a consequence, and it belongs in the argument for both of those properties rather than beside it as a bonus. It also carries more weight than it looks: it is what makes a board on which idle cash earns nothing a reasonable place to be, which is TF-11.9. **Here is the arithmetic, because a reader who is not shown the working supplies their own.** Three steps set what the position earns. *What it costs.* A Short on a club costs one dollar minus that club's price, paid once, in full, at purchase. By the coherence property that figure is exactly the sum of every other name's price (TF-9.4), so the basket costs the sum of its parts and the discount a reader imagines is not there. There is no ongoing cost of any kind, which is what makes this a price rather than a margin. *What it buys.* It pays one token of every other name, so per token acquired the buyer has paid that token's full market price. Nothing on the board is acquired cheaply by buying it this way. *What it earns.* A holder's token of a club's rung is their tokens of that club divided by every token of that club outstanding, the market maker's own inventory included. That denominator is not a fixed supply and never was: it is whatever the trading has left, since buying mints tokens and completing a set destroys them (TF-2.7). It moves, and a holder's fraction moves with it. So a token held through the basket earns exactly what the same token held directly earns, and the 99.62 percent of the holders' block that the other nineteen places collect between them is what all the holders of those clubs token. The basket is one of those holders, holding one token of each. Put the three together. Emission per dollar, for a club, is its ladder share divided by its market value. A position that buys every club at market price earns the average of those figures weighted by the dollars it puts into each, which is what an average is: the basket is the definition of the board's average rather than an exception to it. It also pays market price for every leg that earns nothing, the clubs below the Premier League and Other, because all twenty rungs belong to clubs on the current top-flight table. So the basket is the board index, and that is a normal thing for a market to contain. It takes no view, pays full price, and earns the board's average emission per dollar in one leg, which is what somebody who wants the board rather than an opinion should be able to buy. It behaves like an average: every club whose ladder share over market value sits above it beats it, every club below it does worse. **Which is what the ladder is for.** Under a flat rubric the index would be close to optimal, because every club would pay about the same per dollar and there would be little to be right about. At fifty times, the champion's place pays about fifty times twentieth's, so being right pays a multiple of the average rather than a margin over it. The stratification is what makes taking a view worth taking, and the index is the baseline carry a person earns for declining to (TF-4.2). --- ## TF-4 — The referent and the ladder Two objects decide who earns and how much: a **referent**, one public reality read on a cadence and turned into an ordering, and a **rubric**, a published table mapping that ordering to tokens of each period's emission. On TopFlight the referent is the Premier League table and the rubric is a ladder over its twenty places. The referent is the only part of the machine that touches the outside world, and choosing one well is what decides whether a board of this kind is buildable at all (TF-4.4). ### TF-4.1 — Why football Football is the right first board because the referent is naturally perpetual and the question is native. A league table is public, free, argued about by millions, updated on a cadence the whole audience already watches, and never finished: who stands where renews every season without the market ever closing. No metric has to be invented and no scoring rule has to be authored by anyone here, which is a test most candidate domains fail. **The referent is owned rather than rented.** League tables are published, free and confirmable from several independent sources, so no single provider can revoke the input, change its definition, or price it. A board whose one connection to reality is a number somebody else computes and could switch off is a different and worse product. **Fixtures supply the forcing function.** A market that never settles has, in every other form it has taken, lacked a reason to look at it on any particular day. Football supplies one every weekend, and a second supply of news through the week: a hamstring, a manager sacked, deadline day, a takeover, a European run. The table changes weekly and the forecast changes daily. **And the answer to the oldest question about a market like this is legible.** Asked what real-world event makes a holder money, the answer is: your team won. ### TF-4.2 — The twenty places, and why the ladder is steep **The ladder pays the twenty Premier League places and nothing else.** All twenty rungs belong to clubs on the current table, on one geometric step of 1.2286 from champion to twentieth, which is fifty times end to end. There are no special cases inside it: no separate treatment for promoted clubs, no carve-out for relegated ones, and no band that pays zero. A club outside the Premier League earns nothing because it is not on the twenty places, so relegation is a club leaving the paying set and promotion is a club joining it, and the cliff at the bottom of the table survives as an effect of the rule rather than as a rule of its own. Other earns nothing on the same principle, and holds its value as the option on promotion instead (TF-3.1). Ties are settled by the dead-heat rule: clubs level on all asserted criteria pool their rungs equally, which is deterministic from the asserted data alone and needs nobody's judgement. **Why the ladder is steep, which is the one authored number in the design and therefore the one that has to be justified.** A league table publishes an order, not a magnitude. It says who is third, not how much third is worth, so something has to say what a place is worth. The obvious candidate is the table's own number, and it is the wrong one: points are compressed, because every club in the division wins matches, while the value of finishing first rather than eighteenth is not compressed at all. The sport's published money says the same thing twice over. Premier League central payments are almost flat by design: the equal token every club receives is 68.5 percent of the pot, so in 2024-25 the champion's total central payment of £174.9m was 1.60 times the bottom club's £109.2m. That flatness exists to keep twenty businesses solvent rather than to reward finishing, and the component that does reward finishing, the merit payment, spans about 20 to 1 across the table on its own, £53.1m for the champion against £2.6m for twentieth. Where a sport pays for finishing rather than for existing, the spread is steeper still: Wimbledon paid its 2025 singles champion £3,000,000 against £66,000 for a first-round loser, 45.5 times. So a rubric paid pro rata on the published number would understate what a place is worth by an order of magnitude and would leave a title race moving nothing. Fifty times is calibration to that gap rather than a taste decision, and it sits where the sports that pay for finishing already sit. **The ladder is fixed for the life of a board.** It is published state, a table of tokens summing to exactly one, and it does not move, because changing it would change what every existing holding earns and break the thing the board is for. The commitment that makes that worth more than a sentence is a one-way lock on the setter (TF-12.2). ### TF-4.3 — The rules that fall out of paying a published table **The close season pays.** Between the last match and the next kickoff the table does not move and the emission does not stop: the final standings keep paying through the summer. That is deliberate. Winning the league is rewarded through the close season rather than ending on the last afternoon of it, and a holder who was right about the season collects on being right for as long as that season's table is the table. The same is true at every rung, so a survival on the final day pays through the summer too. It follows that a season is the gap between one kickoff and the next rather than the span of its own fixtures, and that season length is not a fixed number: it is whatever that gap turns out to be, moving a day or two each year as the fixture list does. **The published table on the day is the table**, and banked emission is never recomputed. Two consequences follow, and both are stated as rules rather than discovered by a holder. A retroactive points deduction restates the table from the day it is applied and does not claw back FOOTY already earned under the old standings; the Premier League has applied deductions twice in three seasons, so this is a live case rather than a hypothetical. And a club with games in hand sits lower than its eventual position until it plays them, and its holders earn at that lower rung for as long as the backlog lasts. Both are permanent when they happen. What the rule buys is a referent anyone can check against a public table on a given date, with nothing to argue about and no re-accrual to trust anyone with. **And at the start of a season there is a brief genuine dead heat**, which is briefer than it looks. Between the first fixture and the end of that matchday most of the board is level on zero, so nearly everyone tokens pooled rungs. Once a full round is in, all twenty have played, ten have won and ten have not, and goal difference separates them into clusters of two or three rather than one pile. ### TF-4.4 — The general rule, and the two forms a rubric can take The referent is chosen by a rule rather than by preference, and the rule decides which boards are buildable. **A referent has to be published, ordered and uncontested.** Published means someone outside this company produces it on a schedule. Ordered means it already arrives as a ranking rather than as raw material somebody has to score. Uncontested means two people reading it get the same answer and neither has to accept anyone's judgement to do so. A league table is all three. Where a candidate fails those tests, the operator has to author the ordering itself, and an authored ordering is a rule the operator owns and eventually has to change, which is the thing this design is built to avoid. **The rubric then follows what the referent publishes, and there are two cases.** Where the referent publishes only an ORDERING, the rubric has to be a ladder over places, which is authored and therefore has to be justified rather than asserted, as football's is (TF-4.2). Where the referent publishes a MAGNITUDE, ranking points in tennis and golf, or a rolling prize-money ranking, the rubric is pro rata on that magnitude: a competitor's share of the published total is their share of the period's emission. That authors nothing at all, so there is no ladder to defend and no scoring rule anyone here can be asked to revise, and it dissolves the tie problem, because ties in a continuous published quantity are vanishingly rare. The trade-off is in the feel rather than in the soundness: a magnitude moves smoothly, so the yield does not lurch the way it does when a club crosses a place, and in those sports the ranking race carries the drama instead. **The rule as it stands: pro rata where the published magnitude already reflects what a position is worth, a calibrated ladder where the published number understates it.** Between that and Other as the closure device (TF-3.1), any published, ordered, uncontested ranking becomes a board without anything being invented. --- ## TF-5 — Getting the table on chain The board has exactly one input from the outside world, and everything else is arithmetic. That input is the league table, and how it arrives is the whole of the product's trust surface. ### TF-5.1 — Bonded assertion **Anyone may post the weekend's result, putting up a bond and submitting just the points.** If nobody disputes it inside the challenge window, it becomes the table in force and the poster recovers the bond plus a fee. A wrong assertion is disputed by anyone, and the bond goes to the disputer. There is no privileged writer, no permission to obtain, and no key whose holder decides what the league table says. **Why this referent fits an optimistic oracle better than the things one is normally used for.** The honest-assertion equilibrium such oracles depend on is the assumption that a false claim will be noticed and challenged in time. Here the referent is checked every weekend by millions of people who care intensely about it and are looking anyway. A false Premier League table does not survive the afternoon. Most optimistic-oracle deployments have to hope somebody is watching; this one is asserting a fact that a large, motivated, entirely independent audience has already verified for its own reasons. ### TF-5.2 — What the log checks, and what a wrong table costs The log checks the SHAPE of a table on the way in: the season must match, there must be exactly twenty clubs with no repeats, and the asserted points, goal difference and goals scored must be consistent with the order posted. What it does not check is truth, so an internally consistent false table passes the shape check. That is what the bond and the challenge window are for, and a challenged post never takes effect at all. The log is append-only and every entry is stamped with its own block time, so nothing amends an entry: a posted table is the table in force from its posting until the next post replaces it, and accrual cuts at every table event and never rewrites banked history. **So the cost of a table nobody challenges is bounded by the posting cadence**: one posting period's worth of the holders' block, misdirected between clubs, with everything banked before it untouched. There is no scenario in which a bad post reaches back into what a holder has already earned. ### TF-5.3 — What is still being tuned in it The mechanism is settled; five of its parameters are being tuned, and they trade against each other rather than being independently choosable: the size of the bond, the length of the challenge window, what the fee to an honest poster is and where it comes from, whether the dispute path escalates to an external arbitration system or to one of our own, and what happens to accrual while a dispute is live (TF-11.7). --- ## TF-6 — FOOTY FOOTY is the board's emission token: the thing a club's tokens earn, at a rate the table sets. Its issuance is fixed and public, its value floats, and the board's own income buys it back. The design's central choice is there: nothing is ever promised in dollars, so there is no liability denominated in money that can outrun what the board earns, and no figure that has to be renegotiated on holders when a season is quiet. ### TF-6.1 — The water tower **Issuance is a tank that is filled once and drains a fixed fraction of whatever remains in it every period**, to whoever the referent says is earning, with the fraction set so that the rate halves every four years. Creating supply and issuing it are two different things here, and the distinction is what the tank is for. Nothing is ever created beyond that one fill. What comes back to the tank afterwards is supply that already exists coming round again, so cumulative issuance can exceed the fill by exactly the amount that has been through the loop, while the amount ever brought into existence cannot. The halving is a halving of the FRACTION, not of the tokens. What a period actually emits is that fraction of whatever is in the tank, so it falls as the tank drains and rises as the loop refills it. The water tower is the right picture: because the drain is a fraction of what remains rather than a fixed quantity, recycled tokens lift the rate, and the lift then drains at the same fractional rate as everything else. So this period's income never sets this period's emission. A fat month lifts later emission gently for years, a lean month barely dents it, and in steady state the emission tracks smoothed income with no per-period arithmetic anyone has to do. That smoothing is what the design is for. The rate at which the tank drains and the length of the half-life are dials on that shape rather than questions about it (TF-11.2). ### TF-6.2 — The split, and when the cut is taken Of each period's drain, **eighty percent goes to holders through the ladder, and ten percent each to a team stream and a treasury stream**, on the same schedule as everybody else, with no cliffs. The ladder then divides the holders' eighty percent block, which is the base every ladder figure in this document names (TF-2.3). **The cut is taken when FOOTY reaches a person, not when it leaves the tank**, and the difference is larger than it sounds. A large part of what leaves the tank is never paid to anyone: the market maker holds whatever share of the board nobody has bought yet, and its token of every rung returns to the tank to be emitted again later. Taxing at the moment FOOTY leaves the tank taxes that same FOOTY again on every pass, so a headline of twenty percent compounds into something like thirty-one percent while the maker's inventory is large, and nobody would choose that number. Taxing once, at the moment FOOTY is credited to a real holder, makes the headline and the reality the same figure. It costs something to do it that way and the cost is worth stating: near-term income in FOOTY is materially smaller, because the maker's half is deferred rather than skimmed in passing. It is closer to a deferral than a reduction, since recycled FOOTY is taxed later when it does reach someone. **Two more effects of an early board belong here rather than in a footnote**, because a reader who recomputes the split will find them. While the maker's inventory is large, real holders capture appreciably less of a period's emission than the nominal eighty percent and the two streams appreciably more. And the schedule stretches: the tank halves every four years by construction, while what a holder actually receives halves more slowly than that at a high maker token. Both gaps close as the board fills, and both are properties of an early board rather than of the design. Whether the two streams are paid in FOOTY or in dollars is the one entry in this area where the choice is between two designs rather than a number, and it is TF-11.3. ### TF-6.3 — The buyback loop **Every dollar of the board's income buys FOOTY from the board's own pool.** That is the loop that makes the token something other than an inflation schedule, and it runs in both directions by construction: thicker trading means more fees and more collateral earning interest, a larger buyback, a firmer token, a higher yield in dollars at the same emission, and more reason to hold a club; thinner trading means the reverse, in the same proportion. Nothing in the design promises a floor under the token, and nothing needs to be changed on holders when a season is quiet, because the schedule is fixed and the value floats. **Of what the buyback buys, the larger part goes back into the tank**, so it comes out again through the ladder to whoever is earning then. That is the tie between trading income and the sport: income buys the token, the token returns to the tank, and the ladder distributes it by league position, so what a club earns is a function of how much the board is traded as well as of where the club sits. Remove that leg and income stops rewarding being right about football at all, and the token's connection to the sport narrows to the base schedule alone. **The smaller part goes into the pool as liquidity the protocol owns**, and it replaces a burn. A constant-product pool needs both sides, so that leg is not "buy FOOTY and deposit it" but "buy FOOTY with half of it and deposit both halves": a partial buyback that also deepens the pool, where the buy pressure and the depth come out of the same money and a pure burn gives only the first. **The two are different goods and neither substitutes for the other**, which is why the choice was a real one. **Burning** buys scarcity: supply permanently lower, no cap on the upside, and nothing left to catch a fall, because the money is spent. **The pool** buys liquidity: less upside, because the pool's FOOTY sells into every rally, less downside, because the pool's dollars bid every dip, and the position keeps earning swap fees. What decides it is the state of the market rather than the principle. The pool is small against what a season emits, so nobody can transact at size, and depth is the shortage while scarcity is not. **And the order is decided even where the principle is arguable**: a pool position is reversible and a burn is not. A position can be withdrawn later and burned if scarcity turns out to be what was wanted, and nothing can be un-burned, so the reversible one goes first. Three things about that choice read backwards until they are stated. **Burning does not create value, it moves it**: whoever holds FOOTY at that moment gains in proportion to holdings, and whoever earns later loses, because those tokens would have gone back to the tank and out through the ladder. Sum over everyone for all time and the total is identical, which is what makes it look free. It is a balance rather than a fault: burn nothing and holding does nothing for you, so everybody earns and sells and nobody who turned up early has a reason to stay; burn everything and the league stops mattering. **A locked pool position still leaves the FOOTY on sale**, because pool FOOTY is not out of circulation even if the position is never withdrawn; if the price runs, arbitrageurs buy it out. Locking stops the issuer removing the depth, not the market taking the supply. And **a pool position makes the protocol the market's main counterparty**, so a falling FOOTY price leaves the treasury holding more FOOTY at lower prices, which is the same property as the depth being there when people need it. **One correction that has to travel with any buyback claim.** A token buyback retires every token it buys. This does not. What permanently retires supply is only the smaller leg, so to somebody who holds FOOTY and does nothing, the accretion is that leg and not the whole buyback. The recycled part is a transfer through time from today's holders to whoever earns next, which is what makes the emission self-sustaining rather than a fixed drain. Never reach for a stock-buyback analogy that implies the whole of the money retires supply. ### TF-6.4 — The pool, and the two exits FOOTY trades against dollars on an open constant-product pool. That pool is where FOOTY is sold, and it is not where clubs are traded (TF-2.6). **It is also the one place in the whole design where money has to come from the issuer**: the FOOTY side of the pool is minted, and the dollar side is cash. Two things follow. The pool's depth decides how fast the opening price converges to whatever the market thinks the token is worth, which is really a question of how long the issuer's opening opinion survives contact. And **the first weeks prove nothing**, because nobody holds FOOTY at launch: it accrues over time, so there are buyers and no sellers, the pool cannot be tested, and the price is close to whatever it was set to. That period should not be read as evidence that any number was right. Pool depth and the starting price are both dials, and TF-11.8 gives the range each sits in. ### TF-6.5 — What the yield is actually backed by There are two yields, and the difference between them is the most important thing in the economics. **Yield today** is what the board displays: the holders' share, times the FOOTY emitted per year, times a club's ladder share, times the FOOTY price, divided by the money on the board and the club's price. It is a real number and it is a number about the current token price. **Backed yield** is a different quantity. Ask what FOOTY would be worth if it were priced on income alone, which is the income chasing the FOOTY people sell: price equals income divided by the fraction sold times the FOOTY emitted per year. Put that into the first expression and the FOOTY price and the emission rate cancel each other out, because emitting twice as much makes each one worth half as much. What is left is: > backed yield = (holders' share ÷ fraction of FOOTY sold) × income rate on money held × (club's ladder share ÷ club's price) Worked, for the top club, at a four percent income rate on money on the board and half the FOOTY sold: about sixteen percent a year. **Backed yield is a floor, not a forecast.** Nothing forces the price down to it, and it may never arrive. FOOTY cannot be shorted, and the reason is worth saying because shorts are native everywhere else in this design: a Short exists for a name on the board, where it is the complement of a closed set, and FOOTY is not a name on the board. It is a token trading against dollars on a pool, with no complement to buy. So no arbitrageur is paid to push an overvalued token down, and overvaluation can persist indefinitely. What the buyback gives is the floor, because below it the same income buys more tokens per dollar and pushes back. Anything on screen above the floor is belief, which is fine and may last years, and is not backed. **What is missing from that expression is the point of writing it down.** The emission rate is not in it. The money on the board is not in it. The tank is not in it. Which leg of the buyback goes where is not in it. At the floor, none of them move the yield. **Three things are left and they are the whole game**: how much of the FOOTY that reaches people gets sold, the income rate on money held, and whether the yield is good enough to attract deposits at all (TF-11.4). It is not a feedback loop. The money on the board cancels out of both sides, so there is no runaway. It is stable under every setting but one: if people sell LESS as the price rises, hard enough, it can oscillate, and the concrete test is whether a ten percent price rise cuts selling by less than ten percent. It also collapses if the income rate falls under about three percent, because the yield stops being worth anyone's while and deposits sit at the floor. --- ## TF-7 — The money The board's income has two sources, both of them the venue's own, and the second is the larger one at every scale this product can realistically reach. That is stated first, because describing the engine wrongly is the risk worth avoiding: this is a fee-taking venue sitting on an interest-bearing float, and at achievable size the float leg is the engine. ### TF-7.1 — The fee, and its form **The form is the min-side one: a percentage of the smaller of a position's price and its complement, with a floor of one cent per trade.** Worked through the buy-only path, that form has a simple identity. It is the same percentage of the cash a trader puts in and of the cash they take out, so a round trip costs twice the rate, flat at every price on the board, because the price terms cancel. At thirty basis points that is 0.30 percent in, 0.30 percent out, 0.60 percent of the capital deployed on a round trip: a thousand dollars into a club at 4.4 cents pays three dollars in, three dollars out, six dollars all in. **The form matters more than the level and it is settled for a structural reason.** A flat fee on notional, the face value of the tokens held, makes cheap clubs impossible to leave. Leaving a club is mechanically a purchase of its complement, which then merges against what the holder already has and returns the collateral, so exiting a club priced at half a cent is a buy at 99.5 cents. A notional fee therefore charges that exit as if the trader were moving a dollar per token when the position is worth half a cent. The min-side form charges the economic size of the bet rather than the collateral it locks, which is why entry and exit cost the same and why the cheap end of the board stays exitable. Buying a Short is cheaper again for the same reason. **Two quoting conventions exist and only one of them is honest to lead with.** Against notional, the same round trip reads as 0.02 to 0.10 percent, which is arithmetically true and roughly twenty-five times flattering. The capital figure is the one that leaves the trader's pocket, so it is the one to lead with, and the notional figure survives only as the named contrast. For a sourced comparison at the other end of the range, Football Index charged 2 percent commission on sales. **The fee level is a dial, and the question underneath it, what the fee is for, is a fork rather than a dial** (TF-11.1). No fee is charged inside the ledger itself: an ad-valorem fee needs a price and the ledger holds none, prices living in the market maker, so the fee is charged where the price is. ### TF-7.2 — The interest on float, and which leg is larger Deposits do not sit idle. They are supplied to a lending venue and the ledger credits the interest-bearing receipt it measures back, so the float earns while it is escrowed and that interest is the second input to the buyback. The rate is the whole question and it is the market's rather than ours: USDC on Aave V3 Ethereum paid a median supply rate of 3.85 percent over the 1,292 days from February 2023 to August 2026, with a tenth percentile of 2.28 percent and a ninetieth of 8.52 percent. **Put the two legs together and the crossover is computable.** Fees overtake interest when the board turns over its escrowed collateral more times a year than the interest rate divided by the fee on a single leg. At a four percent rate: about 13.3 times a year at thirty basis points a leg, about 8.0 times at fifty, about 4.0 times at one percent. At a 3.7 percent rate those become 12.3, 7.4 and 3.7. The shape does not change. Turnover on a board of this kind is not a measured quantity anywhere, and the honest working range is somewhere around five to fifteen times a year, with judgement rather than measurement behind it. **Nothing in that range turns over fast enough to make the fee the larger leg at a low rate**, so at achievable scale this is primarily a float business with a football front end. That is a legitimate shape and it is stated rather than dressed up, because the mistake worth avoiding is describing a float-driven buyback as a fee-driven one. It is also exactly why the fee level is a calibration rather than a maximisation, since the leg it feeds is not the one carrying the business (TF-11.1). Two qualifications belong with it. The interest accrues on all deposits rather than only on escrowed collateral, so whether a member's idle cash earns any of it is a decision to be made deliberately rather than inherited, and it is TF-11.9. And the yield venue introduces a third party's credit risk into the backing, which is disclosed below as a property of the unit rather than tucked away. ### TF-7.3 — The unit of account **The collateral is held as the interest-bearing receipt rather than as the stablecoin behind it**, and here is how that squares with a complete set being worth exactly a dollar. A receipt of this kind holds its unit value at one to one with the stablecoin and pays interest by increasing the BALANCE, not the price: a depositor ends the year with more of the receipt, each unit still redeemable for one dollar. So the invariant is exact in both currencies at once. One complete set escrows one internal dollar, the internal dollar is a one-to-one claim on deposited collateral, and the collateral behind it is held as the receipt at one to one (TF-9.5). That is also how principal and interest separate, and the separation is mechanical rather than a matter of bookkeeping intent. The ledger records what it owes as its total value locked, and the interest shows up as the receipt balance rising above that figure. The surplus, and only the surplus, is what is withdrawn to the buyback; the recorded liability is never touched. The same comparison is what a permissionless circuit breaker watches, so a balance that ever fell below the liability would freeze deposits and withdrawals rather than quietly running under-collateralised, and no key can trip or untrip it. Two consequences follow. **The lending venue's withdrawal queue is not in the solvency path**, because what the escrow holds and what it owes are the same instrument, and the in-kind exit pays the receipt out directly without touching the stablecoin at all. And a holder who prefers to deal in the receipt can, straight into and out of the internal dollar, with the stablecoin hop at the edges a convenience for people arriving and leaving rather than a step the accounting depends on. What that leaves is the receipt's own credit risk: the claim is on a lending protocol rather than on a bank, and that is a property of the unit, disclosed as one. ### TF-7.4 — Who loses The venue's income is the fee and the interest on float, and neither depends on which way a position goes: the same fee is taken from a trade that turns out well and a trade that turns out badly. The people who lose are the ones an exchange has. Somebody buys a club high and sells it low, and the trader on the other side of those two trades gains. Somebody holds for the yield and finds that the same money left in a lending protocol would have earned more, which is a return that underperformed rather than a loss to a counterparty. **A bookmaker's loser is a different object**, and the distinction is worth being precise about because the product sits near a shape people will assume it has. There the operator's own book is the other side of the bet and the operator's revenue IS the customer's loss. Here the market maker is the standing counterparty at published prices on one curve, its exposure is bounded by the complete set being worth exactly a dollar, and it is solvency-checked on every state transition like every other account (TF-9.6). Trader profit and loss is zero-sum against other traders, minus fees, because the market cannot pay out more than it has taken in (TF-9.8). --- ## TF-8 — Why a market like this can exist A market that pays you for standing, forever, on a question nobody ever settles, is not a thing that was sitting there waiting to be configured. Five things had to be true, each standing on the one before it, and the whole of TF-9 is the argument for the ones that are argued rather than asserted. This section is the chain, in one page, in dependency order. **They are not five separate achievements and should not be counted as such.** Perpetuity is what the coherence representation already permits once the settlement date is removed, rather than a separate mechanism. Novelty is claimed narrowly and only where it is argued: the constant-cost market maker across a whole partition (TF-9.7), and a coherent many-name partition that is liquid at full declared depth with no external capital (TF-9.8). The rest is a chain of consequences, and the chain is the asset rather than the count. ### TF-8.1 — The closed set is the opening move Everything below rests on one move. **Ask a comparative question over a fixed, exhaustive list, and one dollar buys the whole answer.** Where does each of these twenty clubs finish? One of them has to come first and one of them has to come twentieth, so the set of answers is complete and a complete set is worth exactly one dollar, by construction rather than by agreement. That single fact is what bounds the market maker's liability, which is what makes synthetic capacity safe, which is what makes a board liquid from the first trade with no outside capital, which is what makes a board standable for any published table at all. Remove the closed set and none of the layers exist. **It also answers a question a reader reaches at about this point, which is what stops one person buying the whole board.** Nothing does, and nothing needs to. Buying the whole board means buying complete sets, and a complete set costs a dollar and is worth a dollar, so somebody who bought every name would hold exactly what they paid for. There is nothing for them to do with it. The total cannot be pushed up, because the total is a dollar by construction rather than by anyone's restraint, and the way out is the same curve they came in on. They would collect the whole of the holders' block while they held it, which is what owning every share of everything means and which they paid a dollar a set to get. On an ordinary exchange, cornering a book is an attack: the buyer moves the price against everybody else and extracts the difference. On a conserved board there is no total to move and nothing to extract, so a cap on ownership would be machinery standing guard over a non-event. What a large buyer does move is the RELATIVE prices, which is what every buyer moves, and they pay the curve to do it. It is also what makes the zero-sum reading of the board honest: the question sums to a hundred percent because the question is comparative and declared that way in its construction, and the accounting delivers exactly that. Nothing is being asserted about any external metric. ### TF-8.2 — Coherence: a multi-outcome market held consistent by accounting **A board is one object, not a collection of related markets.** The ledger stores a position as exposure across the outcomes of a partition, fully collateralised, and enforces one invariant before every state transition: a complete set of every outcome is worth exactly one dollar, and no transition that would breach solvency ever executes. Two rules fall out of the representation and they are what make the board coherent. **If two positions pay out identically in every possible world, they are the same ledger state.** And **if two actions cause the same change in exposure, they are the same state transition.** So a Long and a Short of the same club arriving in one account merge in the same transaction, releasing the dollar, and the merge is the accounting itself rather than a function anyone calls (TF-9.4). A short is a single first-class state entry, the complement of the board held natively, with no borrowing, no liquidation and no funding rate (TF-9.3). And the prices of a board sum to a dollar as a structural property, checked on every trade. **What that replaces is worth being concrete about.** In the multi-outcome venues that exist elsewhere, a race with twenty outcomes is twenty separate binary markets, each with its own token pair and its own book, stitched together by a conversion adapter. The prices hanging together is a relationship participants maintain: when the sum drifts, the adapter creates a riskless trade, arbitrageurs collect it, and their payment comes out of the market. Consistency costs money and holds only as well as each book is deep. In the exposure representation, consistency is definitional and costs nobody anything. The full argument is TF-9.1 and TF-9.2. ### TF-8.3 — One market maker across the whole partition **A single automated market maker prices every outcome of the partition simultaneously on one curve, and the cost of a trade does not grow with the number of outcomes.** One pool of declared risk quotes the whole board: a hundred and sixty-five names in one market, at flat per-trade cost. Depth and worst-case liability are the same declared number, and that worst case grows only logarithmically in the outcome count, so wide boards are cheap to hold open. What that replaces is a hard ceiling. Every on-chain logarithmic market scoring rule that has been surveyed recomputes its full pricing state on every trade, at a cost linear in the outcome count, which caps a market in the tens of outcomes. Gnosis shipped a linear-cost multi-outcome LMSR and moved production to a constant-product maker instead; Augur v1 shipped an order book. The chained-binary construction has the same problem from the other side: each of twenty books needs its own liquidity, so keeping twenty outcomes consistent gets more expensive as the number grows, and wide markets bleed out before they get interesting. The basis for the constant-cost claim is TF-9.7. **What it unlocks here is the whole shape of the board.** Six tiers of English football is a partition of a hundred and sixty-five names, and listing all of it is a product decision only because it is a cheap one. ### TF-8.4 — Perpetuity: the settlement date removed **The coherent board with the settlement date removed.** Nothing resolves, no oracle ever declares an answer, nothing expires. A position is held for as long as the holder likes, and what a holder experiences as selling is mechanically a purchase: they buy the complement, the two sides merge on arrival in the same account, and the collateral is released. Nothing is borrowed at any point and no inventory changes hands, so the system is buy-only underneath, with the market maker always the counterparty. The product says buy and sell because those are the member's words (TF-2.6); the accounting only ever adds. The price is not a forecast of an event with a date. It is the standing answer to a standing question, and trading is the only thing that moves it. **What that replaces is the two existing shapes for a market through time.** A prediction market ends: its price converges to an outcome and the market is over, so a standing question needs an endless relay of expiring markets. A perpetual future never ends but needs a funding rate to tether its price to an external spot reference, paid continuously between the two sides. A perpetual coherent market has neither an expiry nor a funding rate, because the complement is fully collateralised and settles by construction, so there is no derivative and no external reference for a tether to attach to. **And it unlocks the layer above it directly.** Removing the settlement event is the exact mechanical fact the zero-capital solvency argument rests on: with no resolution there is no future cash call (TF-9.8). The closest familiar shape is a perpetual future, and where the resemblance stops is the whole point. A perp has no expiry, the position persists, long and short are both native and fully collateralised, and a continuous payment holds its price against something outside itself. A club position tokens every one of those. What it cannot token is what the payment is anchored to: a perp tethers to a spot price, and the tether holds because anyone can trade the spot against the perp and collect the difference, while a league position has no spot, because a table is an ordering rather than a price. That is exactly why the emission has to exist. A tether that cannot be traded has to be paid, and paying it by rank is what makes a place on a table worth money. The comparison then runs toward the holder: a perp's funding is a cost of carry that can run against a position every eight hours, while this payment only ever flows toward holders, out of a published schedule, with nobody paying a counterparty for the privilege of holding. ### TF-8.5 — Internal synthetic capacity: a board that starts liquid **The headline property of the whole design: a board that is instantly liquid at full declared depth with zero external capital.** At creation the market maker's solvency floor is relaxed by a fixed internal synthetic amount, so it quotes every outcome from the first second without holding a dollar. **What that replaces is capital before quoting, everywhere.** An order book opens empty until makers arrive and fund it. A pooled automated market maker holds locked liquidity-provider capital. A classical market maker of this family has a sponsor who funds its worst-case loss with real collateral before the first trade. In every case somebody raises, locks or risks money to make a market exist, which is why new markets are scarce and thin ones stay thin. **Stated at its defensible width**, because the claim is easy to overstate and worth having narrow: a simple bonding curve is trivially zero-capital. The thing that has not been done elsewhere, to our knowledge, is a COHERENT MANY-NAME PARTITION, at the width the previous layer allows, running perpetually, liquid from the first second at zero external capital. The full solvency argument is TF-9.8, and the honest deflated statement of it is that the design does not finance the market maker's bounded loss, it removes the settlement event that made that loss a cash obligation. **What it unlocks is boards as configuration.** A new board costs a configuration file and a deployment rather than a fundraise, which is what lets the layer above treat whole products as instances of one machine. ### TF-8.6 — The emissions machine **The mechanism that ties a perpetual board to the world without an oracle ever settling anything.** Reality decides who earns, income backs what the earning is worth, and a position's price is what traders pay for that earning. One machine, five modules. The **board** is everything above. A **referent adapter** reads one public reality and turns it into an ordering on a cadence, and it is the only module that touches the outside world. A **rubric** maps the ordering to tokens of each period's emission (TF-4.4). An **accumulator** meters token-seconds, who held how much of each position for how long, and splits each position's token among its holders in proportion. An **emission token** carries the value. An instance of the machine is a roster, a referent and a rubric. **Why this is hard, and what the two existing answers cost.** The two ways of connecting a market to reality each pay a price this one does not. **Settlement** connects price to the world by ending the market, which gives up everything perpetuity built. **A metric-anchored funding rate** connects price to an external index continuously, and hands traders a standing arbitrage against the venue's own maker wherever the metric has a foreseeable trajectory, because a funding design pays the drift out as it happens. **The emissions machine routes reality to the EARNING instead.** The referent sets who earns and at what rate; the price of a position is set by trading, and the earning is what gives that price something to be valued against, because every position carries a yield anyone can compute from the published ladder, the published table and the token's price. That is the difference from a funding design on foreseeability: a freely traded price prices a foreseeable path IN ahead of time rather than paying it out, and a trader who is right about the season is paid by the traders who take the other side rather than by a formula. **And the shape the machine reduces to is one input and arithmetic**, which is worth stating because it is the point of the design. An instance has exactly one input, the published ranking, arriving through an oracle anyone can post to and anyone can challenge (TF-5). Everything else is arithmetic: the drain is a closed-form curve, the ladder is fixed, the accrual is a running index that every referent event advances, and a holder's balance is that index rather than a payment somebody makes. **No price is ever set after the opening.** The board is authored once, at genesis, with opening priors that put a champion above a sixth-tier side rather than starting everything level, and from the first trade onward every price is the market's. Because the roster is complete from that moment, there is no later occasion on which a position has to be given a value by anybody. **Two numbers are authored in the whole design**, the steepness of the ladder and the opening board, and both are set once. **And nobody manages a season**, because the ladder pays the place a club occupies today and accrual does not stop between seasons, so the calendar is something the board reads about rather than something it needs. ### TF-8.7 — How they compound One machine. Coherence makes a many-outcome market a single object. Constant-cost pricing makes that object arbitrarily wide. Perpetuity makes it permanent. Internal synthetic capacity opens it at its declared depth on the day it is declared. The emissions machine connects it to reality through revenue. Each is the reason the one above it works at all. The emissions machine is only sound because the board it pays over cannot be settled against and cannot be made incoherent. A second board costs configuration rather than capital. And boards are only worth having at product scale because the partition can be as wide as the question. What exists elsewhere are single layers of this: coherent boards that resolve, perpetuals that fund against an index, emissions that are pure inflation. TopFlight is the place where all five pull at once: a wide board, never finished, open and deep from day one, earning by a table the whole world already keeps, on accounting that cannot come apart. --- ## TF-9 — The technical account This section carries the arguments the sections above rely on. It describes the architecture and the basis for each claim rather than the implementation; the update mathematics and the data-structure procedures are implementation detail that publishes with the source. Where a claim can be checked by behaviour rather than by reading code, the check is stated. ### TF-9.1 — The representation problem, and where it comes from The argument in one paragraph, then the steps. Centralised betting operators already use the correct representation: they record each user's contingent exposure directly, with no tokens and no duplication. Their problems are structural rather than representational. Token-based prediction markets fixed the structural problems by CHANGING THE REPRESENTATION, and inherited a new class of problem by doing so. The path taken here fixes the structure while keeping the representation. **Start with a simple centralised betting site.** It maintains a ledger for each user and prices trades with an internal mechanism. When a user trades, the mechanism prices a change in contingent exposure and the ledger updates accordingly. At the accounting layer, all the system needs to record is each user's contingent exposure. There is no tokenisation, no portfolio of outcome tokens, no splitting or merging, and no duplication of representation. The system records exposure state directly, and the representation is correct. **Its drawbacks are structural.** The operator controls custody, settlement, pricing and the ledger, so users must trust the counterparty. A single pricing mechanism operates inside the system. Users cannot freely transfer positions, because positions exist only as entries in the operator's ledger. And positions cannot be used as collateral, traded externally, or composed with anything else. Every one of those arises because the system is centralised and closed, not because it records exposure directly. **The wrong turn.** When on-chain prediction markets tried to solve those structural problems, they changed the representation: from direct exposure to outcome tokens, minted, held in wallets, transferred freely. Tokenisation buys transferability and composability, and it introduces representational duplication. **The first duplication: economically identical actions treated as distinct.** In a binary market, "buy YES" and "sell NO" have the same economic effect, and both increase exposure to the positive outcome by the same amount. But buying YES requires somebody to sell a YES token out of inventory, and selling NO requires somebody to buy a NO token into inventory. The actions touch different balances and need differently funded inventory. Capital sits doubled up, holding tokens on both sides of a position that would net against each other if the system recognised they were the same state. **The second duplication: economically identical positions in different states.** On a three-name board, "Green does not win" and "Red wins plus Blue wins" settle identically in every world. They are the same bet. In a token system they are different on-chain states: one is a NO-Green token, the other a portfolio of YES-Red and YES-Blue. Reconciling them takes a routing contract or an arbitrageur. **The consequence is that liquidity gets stuck in representation.** Because identical actions are different transactions and identical positions are different states, capital cannot automatically net back into collateral. Liquidity is constrained by funded inventory rather than by willingness to take risk. That is not an implementation problem better engineering solves; it is a structural consequence of representing exposure as tokens, and the fix requires a different primitive. **The path taken instead has three steps and keeps the representation.** *Decouple pricing from the ledger*: the market maker becomes a separate participant with its own account, balances and solvency constraints, rather than a component fused to the ledger. It quotes prices and agrees to trades; the ledger maintains the exposure state of both sides. *Make the ledger economically neutral*: once pricing is decoupled, the ledger does not distinguish between a user, an automated maker, a professional market maker or an order book. They are all accounts with balances, contingent exposure and solvency constraints. Multiple makers can coexist, and an order book can coordinate trades and settle through the same ledger. *Decentralise the ledger itself*: balances, exposure, solvency enforcement and settlement all operate on chain, non-custodially, without trusting any single entity. What comes out is a decentralised counterparty: the correct representation, with the structural limitations gone. ### TF-9.2 — Exposure as the primitive **Instead of tokens, the ledger tracks exposure directly.** For each market, each account holds a vector of numbers representing how much it gains under each outcome. Collateral is committed upfront, and the ledger enforces one invariant: no account can ever be owed more than it has committed. That is the solvency constraint, and it is checked before every state transition rather than reconciled afterwards. The two rules of TF-8.2 are consequences of that representation rather than features added on top of it. If two positions always pay out the same way, there is one number, and any trade that moves it is the same trade regardless of which name the user invokes. If two actions cause the same change in exposure, the ledger applies the same update: no routing, no separate inventory. **Balances are derived rather than stored.** What the ledger holds is the exposure vector; a Long balance, a Short balance and every combination are views computed from it. That is what makes the equivalences structural rather than maintained: there is no second book that could disagree with the first, because there is no second book. ### TF-9.3 — Native shorts The concrete consequence is a change in what a short IS. In a token system, a short is a portfolio. Holding Short(Brighton) means holding a Long of every other club, assembled by buying each one individually. The basket is a derived property of the holder's token balances, and maintaining the equivalence between "short Brighton" and "long everything else" requires routing infrastructure and relies on arbitrage to close gaps when individual pools drift. **Here, Short(Brighton) is a single state entry.** It is a tilt in the exposure vector: positive exposure to every outcome except Brighton. The basket IS the ledger state. Buying it updates one entry. Transferring it moves the whole basket atomically. There is no routing, no arbitrage and no portfolio to rebalance. That shift is what makes shorts first-class rather than derivative. They do not need a separate platform, a funding rate or a maturity date; they exist in the same accounting system as longs, at the same layer, with the same guarantees, from the moment a name exists. ### TF-9.4 — The merge, and the complete-set identity **When a Long and a Short of the same name arrive in one account, the ledger recognises that they are the same state and merges them on arrival**, releasing the collateral. No user action, no routing contract, no coordination with a market maker. The merge is a consequence of the accounting, and it is emergent from the single arrival path every trade, transfer and wrapper movement passes through, rather than being a function somebody calls. **And a complete set of all positions in a market is equivalent to exactly one internal dollar at the ledger level**, enforced by the solvency accounting, independent of any pricing mechanism. That is the identity the board's prices summing to a dollar expresses at the pricing layer. The two are not the same statement: a different pricing mechanism could quote prices that drifted from the identity, and the ledger truth would still hold, with the divergence appearing as an arbitrage opportunity rather than as broken accounting. Consistency is not something the market maintains here; it is something the accounting is. Two further properties fall out without separate engineering. **Pricing is pluggable**: because the ledger handles all exposure accounting, any pricing mechanism can quote and the ledger will execute the corresponding exposure update, and several can coexist on one market. And **the exposure event layer is complete**: every balance in the system can be reconstructed from the canonical exposure event alone. ### TF-9.5 — The internal dollar The ledger's unit is an internal dollar, minted one to one against deposited collateral and burned one to one on withdrawal. Trading never mints or burns it; supply changes only at the boundary. It is exposed as a standard token so that wallets and external tooling can see it. **It is a projection rather than a second balance sheet**, and that is the load-bearing design choice. It holds no balance mapping of its own: every read delegates into ledger storage, and every transfer event is emitted by the ledger. If it stored its own balances there would be two books of the same money, and the merge, which moves collateral from locked to free inside a settlement, would have to be mirrored across them on every trade or drift apart. Projection makes drift impossible by construction, because there is one number and the token is a view of it. The price paid is a strict obligation in the other direction: every change to free collateral anywhere in the ledger must emit a matching transfer event, or the token's balance stops equalling the sum of an account's events. That obligation is discharged at a single choke point, which emits one net transfer per distinct touched account per settlement. ### TF-9.6 — The ledger as counterparty The traditional counterparty holds inventory and bears open-ended risk that somebody has to capitalise. This design changes what the counterparty IS, not whether one exists. When you trade on the board, the market maker's account is the other side of the trade, and it does carry exposure: bounded, provisioned at market creation by its declared liability, and solvency-checked on every trade like any other account. What the design removes is unbounded, discretionary counterparty risk and the need for external capital to bear it. The ledger applies an accounting update, your exposure vector changes, the maker's exposure vector changes, the ledger verifies that neither account violates the solvency constraint, and the trade clears. The ledger is indifferent to which mechanism quoted the price. No entity can go insolvent, because the solvency constraint is checked before every state transition. No entity can move a price without capital at risk. The counterparty function, settlement, consistency and solvency, is discharged entirely by accounting. **And the check is not what people assume it is.** The ledger does not trust the market maker: it grants a defined amount of room at creation and tests the maker against that room on every trade, so a maker that has gone wrong is contained inside the ledger rather than able to take the ledger with it. What the check governs is the maker's own solvency, not a holder's order. It is not a size limit, not a queue and not a rationing rule, and it is not what decides whether a trade prices. If it ever fired, what that would mean is that the market maker had failed, not that the market had run out of something. And it cannot reach a holder's collateral either way, because the ledger is the thing enforcing solvency and the maker is a participant inside it. ### TF-9.7 — Constant-cost pricing across the whole partition The logarithmic market scoring rule is the correct pricing mechanism for a multi-outcome market: it prices every outcome simultaneously, guarantees a continuous quote at all times, and has bounded loss. **No live venue runs it for multi-outcome markets at meaningful scale, because gas cost scales linearly with the number of positions.** That is the survey result behind the claim: Gnosis shipped a multi-outcome on-chain LMSR whose cost grows linearly with the outcome count and moved production to a constant-product maker instead, its own forum discussion citing the gas cost of logarithmic mathematics over many outcomes; Augur v1 shipped an on-chain order book rather than an automated maker; and platforms wanting multi-outcome markets use an order book with a routing adapter on top. **Here the trade cost is identical whether a market has five positions or five hundred.** To our knowledge no published on-chain implementation of this rule achieves that, and the platform history above is the basis for the claim rather than an assertion of it. **The enabler is the separation between pricing and state.** In a token system the maker manages token inventory, so every trade changes global state and requires a full recomputation. Here the maker only quotes prices and the ledger holds the state, which makes it possible to cache the pricing state and update it incrementally rather than recompute it. That is the architectural shape of the property; the exact update mathematics publish with the source. **"Exact" carries a declared bound.** Fixed-point error is limited by a rounding policy that rounds dust toward the protocol, on the order of two millionths of a dollar per trade. Exact up to declared dust, with the bound itself an audit item. **And the claim is checkable behaviourally without the implementation.** Every trade is an on-chain transaction and per-trade gas is readable from the chain by anyone, so constant cost across roster sizes is a measurable property rather than a document's assertion. ### TF-9.8 — The solvency argument, in full The short version first: **the ledger guarantees a dollar backing the full set; the market maker charges at least a dollar for a full set and puts that dollar in the ledger.** That is the whole answer. The rest of this section is the elaboration. **These markets never resolve, and that single fact changes what the maker's worst case means.** In a resolving market the maker carries a real future liability: at settlement every winning share is paid a dollar in cash, at once, which is why such markets need seeded capital and why "who bears the worst-case loss" is the right question to ask them. Here there is no settlement day. The board is one multi-dimensional bonding curve over all the names: every entry is a cash purchase into the curve, every exit is a sale back into it, and prices move as the curve is traded. If you are the only trader, you get out what you put in, minus fees. **Money can leave the system exactly one way.** Positions come in Long and Short pairs that are, by construction, a complete set worth exactly one dollar. Both sides are bought outright and held outright; neither is a borrowed or financed leg, so there is no third party to whom anything is owed and no cost running against either side while it is held. The two prices of a name sum to one dollar at every instant, and acquiring both sides through the curve costs exactly one dollar before fees, whatever the price and whatever the order. That pair-cost identity holds at zero fee, so **solvency does not depend on the fee at all**: fees add margin and make round trips strictly losing, and they are never a necessity. When a Long and a Short of the same name meet in one account they merge and a dollar of cash is released, in that same transaction and without anyone calling anything (TF-9.4). The ledger enforces, on every single trade as a precondition of settlement, that one real dollar sits escrowed for every such pair outstanding, and that dollar necessarily arrived as some trader's payment. A trade that would violate this cannot execute. **So a profitable exit is funded, dollar for dollar, by the payments of the traders who moved the price, and the market can never owe more cash than it has taken in.** **The synthetic capacity is what lets the maker quote a deep, full board from block one with no seeded capital, and it is not a cash obligation.** It appears in exactly one place: the maker's own worst-case solvency check, which is a bound on a resolution-shaped liability that has no trigger on a market that cannot resolve. It is never credited to a balance, never transferred, never withdrawn, and the separate constraint protecting user-redeemable cash deliberately excludes it. Nobody bears it, because there is no state in which it converts into a claim on anyone's dollars. **And it is a grant made once, not a budget that trading consumes.** The synthetic amount is written at market creation and never changes; there is no mutator for it anywhere. What MOVES is a solvency figure the ledger computes against that grant on every trade, rising as the board concentrates on one club and falling again as prices spread back out. That figure is a reading rather than a withdrawal. Nothing leaves the grant and nothing accumulates against it, so there is no exhaustion event and no question of what happens when the budget runs out, because it is not a budget, a pot, a float or a line. Describing it as one invites the reader's next question, and the design cannot answer that question because the premise is wrong. **So the maker cannot go dark.** The pricing function is defined at every state, so a quote exists at every size, and the cash to buy a position back has already arrived, because selling to the maker walks the same curve down and the collateral was paid in by the traders who walked it up. A seller of any size is filled along that curve, at a worse average price the larger they are, which is price impact rather than a limit on how much can leave. A holder is never waiting for a counterparty and is never told the venue has run out. **What the venue does pay for is adverse selection**, and that is a running cost rather than a capital one. Whoever trades first after new information trades against the price the last trade left behind, and the maker is on the other side of that trade. What the venue actually pays is the difference between what informed flow takes and what the fee earns against it, met out of fees (TF-7.1). **The honest deflated statement, and it is still the strong one: the design does not finance the market maker's bounded loss, it removes the settlement event that made that loss a cash obligation.** **One edge worth stating for early traders.** No external pool subsidises liquidity, so there is no subsidy paying informed traders to show up in a young market, and an early trader's PROFITABLE exit is funded only by other traders' subsequent inflows. The distinction in that sentence is load-bearing: exiting at all never requires another trader, because the maker is always the counterparty. What later inflows determine is the price at which the exit happens, not whether it is available. Depth is real from block one; flow is whatever the market brings. **Two known rough edges sit inside audit scope and belong here rather than in a footnote.** Rounding dust exists at the smallest denomination, and the guarantee is enforced by revert. Together they define the one boundary case worth stating precisely: a sell can revert only on a rounding-dust boundary, bounded at the smallest denomination and vanishingly rare because the fee margin dwarfs the dust, and a revert is the system failing safe rather than overpaying. And one consequence sits beside the guarantee: because the market cannot pay out more than it takes in, trader profit and loss is zero-sum against other traders, minus fees. **The system's promise is not yield. It is that the cashbox is closed.** ### TF-9.9 — The contrast, in a table | Property | Token-based conditional frameworks | The exposure ledger here | |---|---|---| | Position representation | Token balances, held as portfolios | Exposure vectors | | Counterparty | Funded inventory plus arbitrage | Ledger accounting | | Short mechanism | Synthetic basket assembled by a routing adapter | Native ledger state | | Sum-to-one enforcement | Arbitrage-maintained, approximate | Ledger invariant, exact | | Multi-outcome pricing | Order book plus routing adapter | Constant-cost LMSR | | Capital required for a new board | Pool seeding required | Zero | | Pricing mechanism | Fixed | Pluggable | --- ## TF-10 — Why it is not built on an existing conditional-token framework The obvious question about all of TF-9 is why any of it needed writing, given that a general conditional-token framework already exists, is battle-tested, and genuinely does support a board that never resolves. That last part is true and worth conceding first: splitting and merging positions in such a framework requires only that a condition has been PREPARED, never that it has been resolved, and nothing times out. The complete-set accounting is substantially the same idea in a different state representation, and that is the easy half. **The answer is three things, and the case should never be argued on anything else.** **Native shorts.** A short here is a first-class position (TF-9.3). In a conditional-token framework the only short is the complement, held either as a union position with its own token identifier, which the framework's own maker cannot price because it indexes atomic positions only, or as twenty separate longs that somebody has to assemble and keep assembled. **Auto netting.** Buying one side and selling the other collapse into a single settlement and net against each other, because one arrival path handles every trade, transfer and token movement, and the complete-set merge is emergent from it rather than being a function anyone calls (TF-9.4). In a conditional-token framework, merging is an explicit user call, which is a different object with different behaviour and different failure modes. **A market maker that needs no capital.** The maker carries internal synthetic capacity, and the whole zero-external-capital claim is that plus the absence of a settlement event (TF-9.8). The framework's split operation performs a real transfer of collateral in, so an equivalent board needs somebody to post the full declared depth before the first trade. The only route to the property inside such a framework is to make its collateral token something we can issue, which means keeping this ledger underneath as the issuer, at which point the framework is a layer on top of the thing it was supposed to replace. **And the honest scope of the claim**, because a case argued too wide is a case that loses on its weakest limb. Most of the other distinctive things this stack can do are convenience rather than necessity for this product: a peer-to-peer intent path, an external venue hook, a time-weighted price feed, several makers on one market, and the circuit breaker are all either unused here or defences rather than capabilities. Even the constant-cost pricing is not load-bearing at twenty outcomes; it becomes load-bearing at the width of the wide board (TF-8.3). The case rests on the three above. **One trap for anyone who revisits this**, recorded because it would be a silent product-ending mistake: in a conditional-token framework, the annual league table must never be reported as a payout, because the payout denominator is write-once and doing so would open redemption at fixed values and kill the board permanently. --- ## TF-11 — The dials **The shape of this product is decided. What follows is the tuning inside it.** Every mechanism described above exists and works the way it is described. What is being calibrated is where a small number of parameters sit within the ranges they live in: how steep, how deep, how much, how long. That is tuning, and tuning is the last thing done rather than the first, because most of these numbers cannot be set sensibly until the thing they are a parameter of exists. A threshold needs a door. A fee needs to know what the fee is for. **Naming them is a claim about the design rather than an admission about its state.** Every working system in this category has the same numbers in it; the difference is whether anybody wrote down which ones they are, what range each lives in, and what moving one costs. A design that publishes its own tuning surface is further along than one that hard-codes the same values and hopes nobody asks. Somebody chose them either way, and the only question is whether the choice was visible. So each entry below names a parameter, the range it sits in, and what is gained and given up at each end of that range. **None is answered here, and that is the point of the section**: an answer written into this document would be a decision that had escaped from somewhere it belonged. Every settable number in the design has been enumerated and the enumeration runs to several dozen; what follows is the subset where the calibration is live rather than a value nobody would contest. **Two entries are a different object and are marked as such.** Whether trading is the right place to charge (TF-11.1) and whether idle cash earns at all (TF-11.9) are choices between two designs rather than a number moving inside one. They are the small minority here, and they are called forks so that nothing in this section dresses a dial as a fork or a fork as a dial. Two of these interact strongly enough to be worth flagging before the list: the fee and the depth schedule are the same number seen twice (TF-11.1, TF-11.5), and the split and the buyback destination are two ways of asking who the income belongs to (TF-11.3, TF-11.8). One set of questions is deliberately NOT here: the ones about lending against these tokens live at TF-13.4, because they concern a separate market rather than this one, and listing them among the tuning would imply they sit on the path. ### TF-11.1 — The fee: a dial on the level, a fork underneath it **The current setting is thirty basis points, min-side.** The form is settled for the structural reason in TF-7.1 and is not in play. What is in play nests: a fork about where to charge, a fork about what the charge is for, and then a dial on the level. **First, the fork: whether trading is the right place to charge at all.** This is a choice between two designs rather than a number moving, which is why it is not a dial. The founder's current inclination is that it probably is, and that it can be small while still belonging there. That is a lean rather than a decision, and the difference matters, because the alternative was examined seriously rather than dismissed. The alternative is a boundary fee, charged on money entering or leaving rather than on trades. It was put to a three-seat panel and lost three to nil, on arithmetic rather than on taste, and the reasons are worth carrying because they are what makes the lean sit where it does. **The amortisation argument is zero-sum**: a boundary fee spreads the same take over fewer events, so whoever trades less pays more per trade, and in practice that means retail pays whatever the high-frequency participants stop paying. **The base is not what it looks like**: the internal dollar is transferable, so people entering and leaving net against each other peer to peer and the charge lands on net flow rather than on gross flow. **And charging on the exit door falsifies a published guarantee**: the design's exit promise is that a holder can always leave, and a fee on the emergency exit is a fee on the thing that is promised to be free of obstruction. **Second, and this is the fork that re-bases everything else: what job the fee is actually doing.** The fee is not obviously the company's income line. It has two other jobs. It **indexes the depth schedule**: depth expands on synthetic collateral at a rate indexed to the fee charged on each trade, so the fee is the meter rather than the funding (TF-11.5). And income buys FOOTY, most of which returns to the tank and is paid out again through the ladder, so income reaches people as emission rather than as profit (TF-6.3). Both of those change what "too low" means. **Which puts a genuinely small fee on the table.** If the fee's job is to be a meter and a buyback input rather than a revenue line, then something in the region of two to five basis points may do the same work as thirty while barely touching honest flow. The depth argument supports that directly: raising the fee and raising the depth multiplier are arithmetically the same for the safety bound that limits how fast depth can grow, and only one of the two taxes trading. **Zero does not work**, and the reason is mechanical rather than commercial: a zero fee is a meter that measures nothing, and because the pricing curve is path-independent, a notional meter with no cost attached can be washed by trading in circles. **Against that, the fee is what makes the emission worth something.** Income buys the token that holders earn. Cut the fee to nothing and you have not removed the cost from the user, you have moved it onto the yield, and the yield is the product. That is the tension, stated at full strength on both sides, and it is not resolved here. **Third, and this one IS a dial: the level, if trading is where it is charged.** Four independent panels have examined this question, and every one of them scored the fee as company income and crossed it against the interest earned on deposits, which is exactly the objective the section above puts in doubt. **What survives that re-basing** is the evidence base rather than the conclusions: comparable venues charge several times this rate once their charges are converted to a cash denominator; no published elasticity of trading volume with respect to fees exists anywhere near these rates, in any market; and the sourced precedent is mostly one-sided rather than symmetric, with Football Index charging on sales and UK stamp duty charging on purchases. **What does not survive** is every conclusion of the form "this rate raises more money than that one", because raising money may not be the objective. **What the level costs in either direction.** Too high and the board tracks the league table worse, because tracking is bought with trades: every correction to a mispriced club is a trade somebody has to want to make. Too low and the buyback is thinner, the token is cheaper and the yield in dollars is smaller at the same emission. There is no arithmetic that supplies the ceiling; it is set by what a rate sounds like said out loud to somebody holding one club. ### TF-11.2 — The emission schedule, and the half-life **Set to a four-year half-life on a continuously draining tank, with the total that will ever exist falling out of the drain rate and the half-life rather than being chosen directly.** The four years is a borrowing from a well-known issuance schedule, and the borrowing is issuance mechanics only. It carries none of that schedule's value story and should never be presented as if it did. **What a longer half-life buys** is a gentler dilution curve for somebody who holds through the first years, and a schedule that stays meaningful for longer. **What a shorter one buys** is a larger present-day yield, which is what a new board has to offer to be worth anyone's attention at all. The choice is really about who the board is for in its first two years, and it interacts with TF-11.4: the more a schedule front-loads, the more it matters how much of what it pays out gets sold. **One property of the current shape is worth knowing rather than rediscovering.** In the first year, circulating supply grows several times over as the tank drains, so a holder who does nothing is diluted in the near term and accretes only once issuance slows against the buyback. That is a property of the schedule rather than of the market, and it is visible in the arithmetic from the start. ### TF-11.3 — The split, and whether we take FOOTY at all **The split is eighty to holders and ten each to two streams**, and the honest question underneath it is not the percentages. It is whether the company should be paid in FOOTY at all. **For taking FOOTY:** dollar income is near zero until there is money on the board, so taking FOOTY is a way of being paid before there is anything to be paid from. **Against:** FOOTY paid to us is a claim on the token we are trying to support, and if we ever sell it we are the sell pressure on our own buyback. The alternative is that holders take a hundred percent of the FOOTY and the company takes a share of the DOLLAR income instead. That is real money, it dilutes nobody, and it removes the entire question of when the cut is taken (TF-6.2) rather than answering it. **So the question is: paid early in a token we are supporting, or later in real money?** Everything else in this area is mechanics that follow from that answer. ### TF-11.4 — Selling pressure, and locking **How much of the FOOTY that reaches people gets sold is the single most consequential quantity in the whole model**, and it is an OUTCOME rather than an input: it is the one thing in the backed-yield expression that is neither ours to set nor observable in advance (TF-6.5). Nothing is currently set to influence it. **The shape that would is lock-to-earn**: extra FOOTY for locking rather than selling. Four things make it the right shape if anything is done here at all. It transfers from SELLERS to LOCKERS rather than from the table to token holders, which is why it survives arguments that a plain holder yield does not: an earner who locks is unaffected, one who sells pays for it. It self-selects, because locking costs something real and not everyone will. Unlocks stagger themselves, because people lock when they choose. And it is voluntary. **One shape is ruled out on its own arithmetic rather than on preference**: a lock that releases everyone on one date synchronises the selling, which crashes the token on a schedule. The problem there is synchronisation, not locking. **What is being tuned** is whether to run it at all, at what rate, for how long, and where the reward comes from. That last one is not a detail: new emission funds it by diluting sellers, while taking it from the ladder's own block funds it by taking from the table, and in code both are just FOOTY leaving the tank, so the distinction has to be held deliberately. **And there is a tension with what a watched chart looks like.** Destroying units is the only thing that mechanically raises price per unit, so every other destination leaves the chart flat, which for a product people watch reads as failure. Locking cuts circulating supply, which supports the price without destroying anything, and whether that nets out turns on how much actually locks per unit of reward. ### TF-11.5 — Depth: what a trade should feel like **Depth expands on synthetic collateral, at a rate indexed to the fee charged on each trade, and it stops when the allowance is spent.** Three properties of that shape are settled and worth stating before the dial. It is indexed rather than on a clock, deliberately, because depth that arrives whether or not anyone traded is a subsidy and subsidised depth is not information. It **stops**, because by then the board is mature and depth belongs elsewhere, and a maker that stops growing while the market grows past it is doing its job rather than running out. And the ceiling is knowable at deployment, because the maximum depth a board can ever reach is a number stated on day one. **Three numbers are in the tuning: the multiplier, the size of the allowance, and where on the range the board should sit.** Depth is taste rather than arithmetic, and the way to have the argument is what a trade feels like. A working target: a hundred dollars should nudge a club and not swing it, a thousand should be clearly visible, five thousand should be an event and not vandalism. **Two things make that harder than picking a number.** **One depth setting cannot serve both spectacle and size**: making a thousand-dollar trade civilised takes roughly eight times more depth, at which point a hundred-dollar trade moves nothing at all. That is a property of having one liquidity parameter, and it restates the standing position that the automated maker is the coherent quote of last resort while real depth comes from people. And **percentages mislead across the curve**: the same hundred dollars moves a very cheap club by hundreds of percent and a very expensive one by a tenth of a percent, while in percentage POINTS it is nearly flat. Because prices must sum to a dollar, nothing on a football board is ever near fifty percent, so the whole board sits in the steep region. **There is a safety limit on the growth rate and it is state-dependent.** Growing depth hands value to whoever already holds a position, out of the maker's float, because the rescale changes the book without changing anyone's token balance, and that transfer must not exceed what the trading that triggered it paid in. The bound is roughly twice as permissive on a flat board as on one with a heavy favourite, so it has to be computed from the board's state rather than fixed. It is also derived assuming inventory is bought through the curve at full slippage; acquiring it peer to peer is free, so the true limit is lower than the derivation says and needs deriving properly. ### TF-11.6 — The opening board, and the price floor **The list is decided. How the tail is PRICED at the opening is the parameter** (TF-3.2). The board is authored once, so this is a number set at genesis and never revisited, which makes it cheap to leave open now and impossible to change later. **The problem is that the Premier League's own pricing method does not reach the tail.** Opening prices for the top flight come from a finishing distribution mapped to expected ladder share, which is a real derivation from real market prices. It returns zero for every club that is not in the Premier League, so it cannot price the other hundred and forty-four at all. Something else has to. **A flat tiny tail** is the honest ignorance prior: give every club below the paying set the same small opening price and let the market do the rest. It is cheap, it needs no source, and it claims nothing. **Its cost is real and should be stated**: a flat tail says two very differently placed clubs are equally likely to come up, which is false and known to be false, so the first informed trader takes that money. That is a transfer from the opening board to whoever reads a promotion market, and it happens once. **A derived tail** prices each club off its promotion odds, discounted for the seasons it spends earning nothing before it can earn anything. It is better where the odds exist, and complete books thin out fast below the second tier, so most of the roster would end up on a judgement anyway. It also imports a hand-set discount factor, which is exactly the kind of authored number the rest of the design works to avoid. **And the floor is the real lever, not the list.** Every listed position carries a minimum price, and with a tail this long a high floor pins a large part of the board at the same pinned number, which reads as a wall of identical prices rather than as information. Dropping the floor lets each club carry a distinct price. It costs sensitivity, because the depth parameter is derived from the smallest price on the board, so a much lower minimum makes the whole board move more per dollar traded, and it costs a little at the top, because every position takes some of the dollar. **What is being tuned is how low to set it**, and that is the knob to reach for if the bottom of the board reads badly. ### TF-11.7 — The bond and the challenge window Five parameters of the bonded assertion are being tuned, and they trade against each other rather than being independently choosable (TF-5.3). **The bond size** trades honesty against participation: too small and a false table is cheap to try, too large and nobody posts. **The window length** trades safety against latency: a longer window is more time for a challenge and more delay before results reach the rates, and the whole product's promise is that a result changes what a club earns quickly. **The fee to an honest poster** has to be enough to pay for the gas and the attention, and it has to come from somewhere, which is a small but real claim on income. **The dispute path** is a choice between escalating to an established external arbitration system, which imports its own assumptions and its own token, and building an escalation of our own, which is more code to be wrong about. **And what happens to accrual during a live dispute** is the one here with two real answers rather than a range: freezing it is safest and stops the board paying anything while somebody argues, while continuing on the last confirmed table keeps the product running and accepts that a successful challenge means a window paid on standings that turned out wrong. **What bounds the whole question** is TF-5.2: the exposure is one posting period's worth of the holders' block, and banked history is never rewritten. These parameters set the odds on losing that, not the size of it. ### TF-11.8 — The pool: how deep, and what FOOTY opens at Where the buyback's money goes is settled, both legs (TF-6.3). **What is being tuned is the pool it acts on**: how deep it is, and what price FOOTY opens at. **Pool depth is the one place in the whole design where money has to come from the issuer**, because the FOOTY side is minted and the dollar side is cash. Deeper means the opening price is harder to move and the pool absorbs more of a young token's selling; shallower means the issuer commits less and the market sets the price sooner. There is no arithmetic that picks the number; it is a judgement about how much of the early price the issuer wants to own. **And the opening price.** The price floats, so it arbitrages toward whatever the income supports and the opening number is a starting point rather than a setting, but two things still make it matter. What it normalises to is a bet, since fair value is discounted future income and at launch there is none. And **direction matters more than level**: open above what the market bears and it falls, so everyone who bought in the first week is underwater, which is a bad start whatever the theory says. Open below and it rises. The two dials are the same dial from opposite ends, because depth is also what decides how long the opening opinion survives contact (TF-6.4). ### TF-11.9 — Whether idle cash earns, which is a fork A member's money sits in one of two places: in a club, or in the ledger as internal dollars between positions. **Whether that second state earns anything is a fork rather than a dial**, because the two answers are two different designs with different behaviour rather than two settings of one, and it is genuinely live: the person who decides it has not. **The case for paying it.** There is something to pay with, and it is the depositor's own money that generates it: collateral is supplied to a lending venue and the interest on it is the larger of the venue's two income legs at any realistic scale (TF-7.2). A member sitting between views is therefore financing the buyback while earning nothing themselves, which is a cost for doing nothing. It is also the shape people arrive expecting, because cash in almost every venue of this kind earns something. **The case against, which is the sharper one, is that not paying it forces information out of the system.** The board's entire output is a set of prices, and a price only improves when somebody takes a position that expresses a view. If idle cash earns, a member can collect a yield without ever having an opinion about a club, and money that could have been an opinion sits as money instead. If idle cash earns nothing, then the cheapest safe thing to hold is still a position, and every position carries information into the price. That is a real product argument rather than a way of saving money on interest: the mechanism is being asked to convert passive capital into price discovery. **The obvious objection to paying nothing is that cash becomes dead money, and the answer to it is the index.** Nobody has to sit in cash in order to be safe. Shorting the club that pays least is one trade and one position, bought outright with nothing borrowed and nothing running against it, and it takes no view about anybody while earning roughly the board's average yield per dollar (TF-3.4). So the choice a member faces when cash pays nothing is not "risk or nothing"; it is "a view, or the board". That is exactly the choice the design would like people to be facing, and it is why this dial sits next to the basket carry rather than in the fee section. **What it actually turns on is the balance of fees**, and that is the honest condition rather than a hedge. A round trip has a cost (TF-7.1). If that cost is large relative to the yield a member would forgo by sitting out for the period they intend to sit out, then paying nothing on cash stops being an invitation to take a position and becomes a penalty on somebody who is genuinely between views and cannot afford to move twice. So this is not independent of TF-11.1: a low fee makes the forcing function humane and a high one makes it a toll. Set them together or not at all. ### TF-11.10 — The room threshold Every club's room is gated on holding enough of that club (TF-2.8). **How much is enough is being tuned**, and three questions sit inside it. Who is eligible at all is not one of them: the door is directional and only Long holders pass it (TF-2.8). **The level.** A hundred, a thousand, ten thousand tokens have all been named as candidates and none chosen. The trade is the ordinary one for any door: low and the room is populous and cheap to be in, so it fills with people who are barely exposed; high and the room is small, committed and possibly empty. What makes it harder than the usual version is that the threshold is denominated in the club's own tokens, so the same number is a different amount of money at every club on the board (TF-2.8), and a level that reads as serious commitment at the top of the table is pocket change four divisions down. **Whether it is an absolute count or a share of supply**, and these behave very differently as the board grows. **An absolute count** is legible, easy to say and easy to check: a thousand tokens is a thousand tokens. But tokens are minted on demand rather than issued in a fixed quantity (TF-2.7), so the number in existence rises as collateral arrives, an absolute threshold gets easier to clear over time, and a room that meant something in the first season admits nearly everyone by the fourth. **A share of supply** holds the room's exclusivity constant forever, and pays for it twice: it is harder to explain, and the bar moves under a member who has not traded, so somebody can be shown the door by other people buying. **And what happens the moment a holding drops below the line.** Immediate is honest and unkind: sell a few tokens and the door shuts mid-conversation. A grace period is kinder and means the room contains, for a while, people who no longer hold, which is the one thing the door exists to prevent. There is a related question of whether the board itself has a room that anyone holding anything can enter, alongside the per-club ones. **What it costs to leave open** is small in build terms and not nothing in product terms: the threshold is what decides whether the rooms are a reason to hold or a formality, and it is the kind of number that is very hard to raise later, because raising it evicts people. ### TF-11.11 — What it costs to leave a dial untuned Tuning late is right for most of these and wrong for a few, and the difference is specific rather than a matter of temperament. **The fee.** The fork underneath it is the expensive part rather than the level: until it is settled, every model built either way is thrown away when the other objective wins, and the rate keeps being argued on two incompatible scoreboards. It also means the first change to the rate, whenever it comes, costs whatever credibility the number had, so a rate shipped as provisional is a different object from a rate shipped as settled. **The emission schedule and the split.** These are the ones that harden fastest. A schedule is a promise the moment somebody holds a position under it, and changing it afterwards is the exact move that made the closest precedent to this product notorious (TF-12.1). The window in which they are genuinely adjustable closes before the window in which the fee is. **Depth.** Cheap to leave open, because it is changed by configuration rather than by breaking a commitment, and it has a designed schedule and an ordinary way to move. **The opening board is the opposite.** It is authored once and then never touched again, so it is cheap to leave open right up to the moment the board is created and impossible to revisit afterwards. Whatever the tail opens at is what the tail opened at, permanently. **The bond and the window.** Cheap to leave open in the sense that the exposure is bounded and known, expensive in the sense that they are the last thing standing between the product and a single writer with a key, and nothing else in the design is a substitute for them. **Whether idle cash earns.** Cheap to leave open in the sense that paying nothing is the state it is already in and paying something later is an addition rather than a reversal. Expensive in one specific way: it shapes what the first cohort of members does with their money, and habits formed in the first season are what the board's price quality is made of. **The room threshold.** Cheap to tune now, expensive later, and asymmetric: lowering it is a welcome and raising it is an eviction. **And the one that is not tuning at all.** How much FOOTY gets sold is not a parameter anybody sets. It is measured, and it can only be measured with real users, real money and a real token. Every model here carries it as an assumption, and the honest position is that a season is what answers it. --- ## TF-12 — The precedents, and the questions no design settles ### TF-12.1 — The precedents Three things have been tried in the same space, each of them instructive, and each of them is handled here by conceding what is unflattering rather than distinguishing our way out of it. **Football Index** is the closest precedent and the mechanical distinction is the whole lesson. They promised a CASH dividend: a liability denominated in money, which outran revenue and was corrected by changing the rubric on holders. Reported losses were around £90 million. TopFlight emits a fixed-schedule TOKEN that is nobody's liability, whose value floats on the buyback, so payouts cannot outrun income and a bad season is a cheaper token rather than an unpayable debt (TF-6.1). There is a second structural difference and it is the more important one: they priced thousands of individual PLAYER assets with no closed set and no total, which forces an authored rubric chasing transfers, form and media noise, on assets that decay as players move, age and retire. This prices ONE BOUNDED, COMPLETE, PERMANENT SET, which is the enabling constraint rather than a smaller ambition. Their commission was 2 percent on sales, which is the sourced comparison in TF-7.1. And their 500,000 traders are a ceiling on category interest rather than evidence of appetite here, because those traders came for a subsidised operator-funded dividend. **Sorare** is the category's reach and capital high-water mark: a $680m Series B at a $4.3B valuation, 378,000 cardholders. The collectible treadmill is not inherited here, and the part that matters is the other one: they face a UK Gambling Commission prosecution, with trial listed for 7 June 2027, and Sorare denies it. A fantasy game with a skill defence still drew a charge, so a club-price board with a yield does not get to assume it sits further from that line. **Fan tokens** show what a per-club token with no cash flow and no comparative structure drifts to: CHZ around 95 percent off its high, PSG down 73 percent in four months, and the Commons Culture, Media and Sport Committee recommending exclusion from fan-engagement measures. The concession is explicit: the 2022 criticism that the only buyers are crypto enthusiasts lands on any on-chain football product until an audience shows otherwise, and a thin buyback bid would make FOOTY a fan token with extra steps. ### TF-12.2 — The regulatory shape **The substantive question is the characterisation of the instrument**, and it turns on taking money. A product whose earning follows sporting results sits near gambling's regulatory shape, and the Sorare charge above is the reason not to assume otherwise. Conversations are underway with crypto counsel and Gibraltar is the leading route; a formal opinion is funded before real money rather than after. This is neither a dial nor a fork, because the answer is not ours to choose at all. **Two commitments that belong beside it**, because they are what makes the question answerable rather than deferrable. A third-party security audit is passed before any real-money deployment rather than after one. And the ladder setter takes a one-way lock before real money, which is what turns "the rubric is fixed for the life of a board" from a sentence into a property (TF-4.2). No round size, valuation or instrument appears in this document, by design. ### TF-12.3 — The question no design settles **Whether an audience trades a football market that pays every day is not answerable from inside this document, and nothing in it should be read as claiming otherwise.** The mechanism can be argued and it has been. The precedents bound the category and do not evidence the demand. A season is what answers it. Two things follow from that and are stated rather than left implicit. **A practice board measures the wrong thing**, because where the money is free, activity measures belief in whatever might be recognised later at least as much as it measures appetite for a football market. And **there is no channel plan, no acquisition cost and no partnership in this document**: the audience is described and nothing is claimed about reach that has not happened. **Who the first users are**, stated as a view rather than as a finding: traders taking a position on the table. A club is bought because the market is underrating it, held while it climbs, and sold when everybody else has caught up. The emission is the machinery that makes being right pay rather than the reason to be there: a club's earning rises with its place, so a holder early to a climb collects more every day on the way up AND re-rates the position as the rest of the board arrives at the same view. Two edges, one position. Early users are likely to arrive for the emission itself, which is worth saying plainly rather than dressing up, and it is subordinate to that: nobody holds a club to collect a percentage, and the percentage is what a view is worth when it turns out to be right. The distinction between a football audience and a crypto audience is not one the product needs to resolve, because football crosses over with everything. --- ## TF-13 — What the design opens, and does not depend on **Nothing in this section is part of the product described above.** Nothing else in this document depends on it, nothing is being built toward it, and the design is complete without it. It is here because a reader who has understood what a club token is will think of it within a minute of finishing TF-3, and a document that has not got there first looks as though it had not noticed. The observation is that these tokens are an unusually good thing to lend against, and that once lending exists the obvious loop follows. One other thing sits here and needs a sentence rather than a section: the same directional door that fills a club's room with people who want that club to win would just as readily fill a room with people who want it to lose (TF-2.8), and no such room is part of this or is being built. ### TF-13.1 — Why the collateral is unusually good A club token is already close to the profile a lending market wants: it is yield-bearing, it has a live price at every instant, and there is a market maker on the other side of it at every size and at every state of the board, so it can always be valued and always be sold (TF-9.8). Two further properties are ones most collateral does not have. **The price is bounded by construction.** Every token sits between zero and a dollar, and the whole board sums to a dollar (TF-8.1). Collateral that cannot go to a surprising number is a different object from collateral that can, and it is worth being blunt about how different: essentially every liquidation engine ever written exists to handle the case where a price does something nobody sized for. Here the range is not a risk parameter somebody chose. It is the arithmetic of the board. **And a complete set is worth exactly a dollar, always, by identity rather than by market** (TF-9.4). So a full-board position is not merely stable collateral, it is collateral whose value is an accounting fact that holds at every price and in every state. The basket carry in TF-3.4 is a near-complete set which also earns the board's average emission, which makes it close to the ideal object to lend against: a dollar-ish thing that pays a yield, whose dollar-ish-ness is enforced rather than observed. ### TF-13.2 — The loop, described rather than sold Hold tokens, borrow against them, buy more tokens, repeat. It is the same loop that appears wherever collateral yields, and it produces two exposures rather than one: **leverage on a club's league position**, and **leverage on the emission**. The first is the one with no existing instrument anywhere, because there has not previously been a way to be levered long a team's place in a table. That is a description rather than a recommendation. What follows belongs in the same breath rather than at the end. ### TF-13.3 — Where the risk sits, stated precisely **The base design carries no dollar liability, and that does not change.** Nothing here is ever promised in money: the emission is a fixed schedule of a token whose value floats, so no obligation can outrun what the board earns and there is no figure that has to be renegotiated on holders when a season is quiet (TF-6.1). That property is what separates this from the precedent that promised a cash dividend and had to change the rubric on its holders when the dividend outran revenue (TF-12.1). It is load-bearing, and a lending market built next to the board does not touch it. **Lending against these tokens would be a separate mechanic.** The tokens are transferable, so nobody needs permission to lend against them and such a market could be built by anyone, or by nobody. Whoever borrows takes on a dollar obligation, by choice, in that market. The board did not make that obligation, does not owe it, and does not become party to it. What the board owes anybody, before and after, is nothing. **The exact analogy is margin on a stock.** A company does not become riskier because somebody borrowed against its tokens. The leverage is a fact about that person's position, not about the asset and not about its issuer. That is the correct way to read every sentence in this section. **The caution is real and it belongs to the borrower.** Recursive leverage against a yield-bearing token is a well-known way for a position to unwind violently, because the yield supports the price, the price supports the borrowing, and the borrowing buys more of the yield, so the same move runs through all three at once on the way back down. Anyone who knows the space will have thought that before reaching this paragraph and they are right to. It is a caution about leverage, and about whatever market provides it, rather than a hole in the design described here. **One thing genuinely does reach back, and it should not be glossed.** Forced selling is still selling. Liquidations in a lending market would arrive at the board as trades, and the board's prices are its output, so a large enough cascade would move them. Two things bound that rather than removing it: the maker quotes at every size at every state of the board, so there is no gap and no failure to fill (TF-9.8), and the sum is pinned at a dollar, so the board cannot fall as a whole and what happens instead is a redistribution across names. That is a real and specific effect, it is different from the usual one, and it is the last question in TF-13.4. ### TF-13.4 — The questions it would raise, and they are somebody else's These are questions about a market that does not exist rather than tuning on the one that does, which is why they sit here rather than in TF-11. **Whether lending is native or left to third parties.** The tokens are transferable, so third-party lending needs nothing from us and is the default outcome of doing nothing at all. Native lending would be a second mechanism to design, secure and answer for, and it would put the board's operator on the other side of an obligation the board is currently free of, which is precisely the property TF-13.3 protects. **Whether the protocol would ever lend against a position itself**, which is the sharper form of the same question and the one where the answer most plausibly is no. **Whether emission continues to accrue on a token posted as collateral, and to whom.** This is a real design fork rather than a detail. If it accrues to the borrower, the yield offsets the interest and leverage becomes materially more attractive, which is the same as saying the loop in TF-13.2 runs faster. If it accrues to the lender, loans get cheaper to offer and the borrower gives up the thing they bought the token for. If it accrues to neither while the token is escrowed, emission stops reaching a person during that time, which changes the accrual accounting rather than only the economics. **And what a liquidation does to a conserved board.** Selling one club moves every other price, because the prices sum to a dollar. So a liquidation in one name is a repricing of the whole board, and the shock lands on holders who had nothing to do with the loan. Whether that is better or worse than the ordinary case is not obvious and nobody has worked it out. Better, in that the board cannot gap to zero and the aggregate is fixed, so there is a floor underneath that unconstrained collateral does not have. Worse, in that the disturbance is socialised across names rather than contained in one. It is a genuinely novel question and it is stated as one. --- ## TF-14 — Key definitions **Board.** One market: a partition of named outcomes priced together on one curve, with prices summing to one dollar by construction (TF-8.1). **Long / Short.** Positive exposure to a name, and its complement. A Short is a native single state entry, the rest of the board held as one position, with no funding rate, no expiry, no borrowing and no liquidation (TF-9.3). **Merge.** A Long and its matching Short arriving in the same account merge in the same transaction, releasing the collateral. The merge is the accounting rather than a feature (TF-9.4). **Complete set.** One of every outcome. Worth exactly one dollar, as an enforced ledger invariant (TF-9.4). **Other.** The board's residual: everything not individually listed, held and priced in aggregate, and a permanent first-class position rather than a placeholder. On TopFlight it is every club outside the listed hundred and sixty-four; it takes a top-flight rung and earns it if one of those clubs ever occupies one. **It never splits**, and no carve, fold-back or migration path out of it exists anywhere in the design (TF-3.1). Generally it is the closure device that turns any open-ended published ranking into a closed exhaustive set, which is what lets a finite list still sum to a dollar. **Internal synthetic capacity.** The fixed synthetic amount the ledger grants a market maker at creation, relaxing its solvency floor by that much, so a board opens at its declared depth with no external liquidity-provider or market-maker capital required. The grant is made once and never changes, and nothing is ever withdrawn from it; what moves is a solvency figure computed against it on every trade, which is a reading rather than a withdrawal. Its purpose is to bound a maker that has gone wrong, so if the check ever fires the maker has failed rather than the market having run out of anything (TF-9.8). **Referent.** The one public reality an instance reads, turned into an ordering on a cadence. TopFlight's is the published Premier League table (TF-4). **Rubric.** The published table mapping the referent to tokens of each period's emission. It takes one of two forms, decided by the referent: a ladder over places where the referent publishes only an ordering, which is authored and therefore has to be justified, or pro rata on a published magnitude, which authors nothing (TF-4.4). **The ladder.** TopFlight's rubric: twenty rungs over the twenty Premier League places, one geometric step of 1.2286 between them, fifty times end to end, summing to exactly one. It divides the holders' block, and every figure quoting it names that base (TF-2.3, TF-4.2). **Dead-heat rule.** Clubs tied on all asserted table criteria pool the tied rungs' emission equally (TF-4.2). **Token-seconds.** The accounting primitive of the accumulator: tokens held multiplied by time held, per holder per position (TF-8.6). **The water tower.** The emission token's issuance model. A tank starts full at the cap and drains a fixed fraction per period to current earners, on a four-year half-life; the buyback returns most of what it buys to the tank; nothing beyond the initial fill is ever created (TF-6.1). **FOOTY.** TopFlight's emission token. One emission token per instance (TF-6). **Backed yield.** What the income actually pays for, as distinct from what the board displays: the holders' share over the fraction of FOOTY sold, times the income rate on money held, times a club's ladder share over its price. A floor rather than a forecast (TF-6.5). **The earning rate, as displayed.** FOOTY a day for $100 spent on a club at its current place and token price (TF-2.2). It is deliberately not an annualised percentage: annualising a rate the table resets every week states a year's return the design does not offer, and multiplying by the FOOTY price would move the figure when nothing about the club had changed. **The internal dollar.** The ledger's unit of account, minted one to one against deposited collateral and burned one to one on withdrawal, exposed as a standard token and holding no balances of its own (TF-9.5). **The interest-bearing receipt.** What collateral is actually held as, rather than the stablecoin behind it. It holds its unit value at one to one and pays interest by increasing the balance, which is how the complete-set identity stays exact in both currencies at once. Its credit risk is a disclosed property of the unit (TF-7.3). **The fee.** A percentage of the smaller of a position's price and its complement, per trade, with a one-cent minimum. The min-side form (TF-7.1). **The basket carry.** Shorting the club that pays least, which means holding one token of every other club on the board. It costs a dollar minus that club's price, and it earns almost all of the emission the board pays, so it is a yield on the whole board without picking anybody: the index. Nobody built it, and there is no product behind it; it falls out of a conserved board plus native shorts. Beaten by any club paying more per dollar than the average, missed by any club paying less, and beaten by more by anyone who is right about which clubs climb (TF-3.4). **Supporters room.** A club's room, entered by holding at least a threshold of that club's tokens as a Long and left by ceasing to. The door is directional: a Short holder does not pass it, whatever quantity of the club's tokens the position contains, because the room's only property is that everyone inside wants the same result. A token is therefore a membership as well as an asset, the alignment of the room is a property of the door rather than of any rule, and the threshold is unset (TF-2.8, TF-11.10). **Dial.** A parameter of a mechanism that exists, whose value is still being tuned. A dial is a knob with a range and a cost at each end, never a gap where a design should be. Dials are collected in TF-11, which explains what tuning means here, and are never written as decisions. **Fork.** The rarer case, where the choice is between two designs rather than between two settings of one. Two are named in this document, at TF-11.1 and TF-11.9, and they are marked so that nothing dresses a dial as a fork or a fork as a dial. Questions about markets that could be built alongside this one, rather than about this one, are neither, and sit at TF-13.4.