Five Structural Edges, One Machine

A feature is a weekend of work. A structure is a rewrite.

Most of what a DeFi platform ships can be copied by a competent team in a few days. Gas sponsorship, a points programme, a forum, a referral code: all of it is copyable, and none of it is a reason to choose one venue over another for very long.

Five things about BASIS are not in that category. They are not screens or campaigns. They are decisions taken inside the contracts, and each one constrains the others, so copying any single one in isolation does not get you the behaviour.

Each section below names where it applies.

1. Payout architecture

Applies to: prediction markets

On an order book a prediction share pays $1 and not a cent more. That is the definition of the instrument. Every book in the world caps the payout at a dollar because a share is a claim on a dollar.

BASIS uses a modified pari-mutuel pool. Winners split the whole pot. The price you paid is not the price of a claim, it is your share of what everyone else put in, and there is no ceiling on what that share turns out to be worth. On a wide field where the crowd was wrong, the same correct call pays a multiple of what the book version pays, because the losing side's money is still in the pot rather than in the pockets of the traders who sold you the share.

This is not a fee schedule that can be changed. It is what the payout function does. A book that wanted this behaviour would have to stop being a book.

Nobody has to quote a price

On an order book, somebody is standing there carrying inventory risk against people who might know more than they do, and that service has to be paid for. Across a replay of 111 settled markets, that cost ran between 1.3% and 7.9% depending on field size, blending to 7.6%. BASIS charges 1.5% on a prediction trade, and no one is being paid to stand in the middle.

A second consequence: a pool price cannot move without money landing. You can cancel a quote on a book. You cannot cancel a bet in a pool.

Which markets exist at all

A book has to find someone willing to quote both sides of every outcome. On a hundred-way field that is a hundred separate quoting problems, so books either do not list the market or list a thin version of it. Structured bets are worse. An exacta, a superfecta, a ten-leg parlay, a survivor pool: each one is a correlated pricing problem that needs a team and a risk model.

In a pool there is nothing to quote, so building one of these is the same act as building any other market. An exacta with 360 combinations opens as though every combination had already been bought. A parlay reads the outcomes of our own markets on-chain, so there is no oracle in the loop and no counterparty to source. These bet types are not merely possible on BASIS, they are easy, which is why the platform can carry a class of market that books usually decline to offer.

Wide fields, longshots and structured bets are where the pool does its best work, and they are where the book is thinnest.

2. Token math

Applies to: the launchpad, and to Predict+ where it meets the markets

The token contracts expose a volatility dial that has no equivalent elsewhere.

At one end of the dial a token behaves like any other launch: price moves freely in both directions, which is what most memecoins want. At the other end sits Stable+, where the price can only rise. In between sit degrees of rising floor. A Floor+ token moves in both directions like anything else, but it carries a floor that ratchets up on buys and never comes down, so the price can fall, just never below that line. The founder picks a position on the dial at launch.

There is no separate pot of money holding the floor up. It falls out of the pricing formula. Every buy mints more tokens than the buyer receives, and every sell burns more than the seller returns. That asymmetry is the whole mechanism: it damps the move in both directions and ratchets the lower bound upward as volume goes through.

What it costs

The same formula that stops the price falling below the floor also slows it rising. Because a buy mints extra tokens alongside the ones the buyer gets, a dollar of buying moves the price less than the same dollar would move an ordinary curve. That is the trade: less upside velocity per dollar in exchange for a price with a floor under it, and collateral a lender can advance against.

What a floor makes possible

Collateral becomes trustworthy. A lender can advance against a price with a known lower bound in a way it cannot against a price that can go to zero.

Leverage stops needing a liquidation engine. BASIS offers 20 to 36x on floored tokens, and no position is ever closed on price. Loans expire on a date, checked against the block timestamp. A flash crash cannot take your position, because price is not an input to whether you still have one.

Predict+, where this meets the markets

A prediction market on BASIS carries two distinct bets on the same event. The first is the ordinary one: back an outcome, win a share of the whole pot if you are right, lose your stake if you are not.

The second is the volume play. Every market can issue a Predict+ token, a Stable+ token attached to the event, which trades on the activity around the market rather than on its result. It rises with the money moving through the market whether or not your side of the argument wins. A market that generates enormous trade and then resolves against you can still have been profitable.

The third possibility is the interesting one. Because credit runs at up to 100% loan-to-value against the token, a holder can put the same dollar into both: hold the Predict+ position, borrow against it, and place the outcome bet with the loan. Two exposures, one dollar, and the outcome bet is funded by an asset the holder still owns.

3. Virtual liquidity

Applies to: the whole platform

Depth on BASIS is arithmetic, not rented capital.

A conventional venue needs liquidity providers, and liquidity providers need to be paid, and they can leave. Every market has a cold-start problem: the first person to arrive finds nothing to trade against, so nobody is the first person to arrive.

BASIS prices from a virtual reserve. There is no counterparty supplying the other side and no position for anyone to withdraw. The practical effect is that depth exists before any trading does, in four places at once:

• A new token is deep in the block it is created. The founder chooses the depth at launch, and the token trades immediately rather than waiting for someone to seed a pool.
• Every outcome in a prediction market is buyable from the first dollar, including the ones nobody has touched.
• A combinatorial field opens as though all of its combinations had already been bought, which is what makes a 360-way exacta a normal product rather than an exotic one.
• The loan book lends against that depth, and no provider can pull it out from under a borrower.

There are no liquidity providers anywhere on the platform, and no LP positions to hold, farm or lose. That absence is the feature.

4. One machine

Applies to: the whole platform, including loans, leverage and vesting

None of the first three stands alone, which is the point of drawing this as a map rather than a list. The credit layer is only safe because the token math produces a floor. The floor is only reachable because virtual liquidity gives the token depth at launch. Predict+ only exists because the token math and the market math live in the same system. Remove any one and the others degrade.

A DEX, a token launchpad, prediction markets, loans, leverage, a yield vault and vesting sit behind one account and one SDK surface, across a suite of more than 20 smart contracts deployed in phases. A competitor would have to merge four products to match the surface, and their pieces still would not connect.

Credit is the connective tissue

There is no price oracle in the core contracts. Every other lender's liquidation engine is a price feed with a health factor attached to it. Ours has neither, because it does not need one: liquidation on BASIS is a date. When a loan expires the debt is settled from collateral and the remainder is claimable.

The cost is knowable before you sign. Borrowing $1,000 for a year costs 2% to open and 0.005% a day, which is $38, fixed at the moment you take the loan. Extending an existing loan costs a fraction of originating a new one. Conventional lenders float their rate with utilisation, which means the cost of a position you already hold changes underneath you.

Up to 100% LTV against tokens, vault positions and vested tokens is what turns locked allocations into working capital, and it is what links every product on the platform to every other one.

One dollar, several jobs, several fees

The same dollar does several jobs. Stake STASIS, earn fee yield on it, borrow against that position, deploy the loan into a market. Capital sitting in a long-dated prediction can earn while it waits, which turns markets that would otherwise freeze capital into places worth putting it.

The platform earns from multiple streams rather than one. Trades, bets, order fills, loans, extensions, leverage and settlement each carry a fee. Because the products are integrated rather than separate, the same dollar is earned on several times as it moves, which is what makes a 1.5% rate workable. Every stream lands in the same split: 48% to the BASIS staking contract, 20% to the market or token creator, 16% to STASIS staking, 12% operating team income, 4% to reward-phase holders. Creators keep their 20% permanently, which is distribution that cannot be forked away.

5. The evidence

Applies to: the payout claims above, and nothing else

The payout claims are not projections. They come from a full order-flow replay of 111 settled Polymarket markets: every trade on every outcome, $422m of notional, fields from 3 to 104 outcomes with a median of 13.

Run through the BASIS payout function, 61% of winning wallets and 71% of winning money would have finished ahead of the book. Across the 77 markets a longshot won, the winning side would have taken 1.33 times as much. Money that arrived in the first 5% of a market's life, in the markets the favourite went on to win, would have done 4.95 times better in the widest fields, and in fields of 31 outcomes or more it beat the book in every single market.

These are replays. They say what the same order flow would have paid under a different payout function, not what anyone has earned.

What is copyable

Gas sponsorship, points and the forum are growth mechanics, and any of them can be copied by anyone who wants to. The five above cannot, because they are not features sitting on top of the system. They are the shape of it.

For the mechanics behind each of these, see Why BASIS Chose Pools, Stable+: The Token That Can Only Go Up, Floor+: Speculation With a Safety Net, Zero-Liquidation Lending and The Predict+ Playbook.