Every losing outcome pays into one pool, and winners split all of it. In a full order-flow replay of 111 settled Polymarket markets ($422M of real trading), 61% of winning wallets and 71% of winning money would have finished ahead of the order book, and early money on wide fields would have been paid up to 4.95× more. BASIS is a full-stack, agent-native DeFi platform on BNB Chain: two products at the front, the uncapped prediction venue and the token launchpad, and three systems underneath that power both, the exchange, the lending desk, and the savings vault.
Ernst & Young
Legal Counsel
Hashlock
Audit Partner
(Pending)
BNB Chain
Launch Network
Veilon AG
Switzerland
The moment
The parent of the New York Stock Exchange put $2 billion into Polymarket. Kalshi raised at a $22 billion valuation. And on June 10 the CFTC issued its first-ever framework for the asset class. The capital, the volume, and now the federal rulebook are all here. Every incumbent shares the same cap.
And the market is only one of two front doors
The launchpad and the three systems under it are priced against DeFi itself, and BNB Chain is where that flow already sits.
The Structural Edge
Polymarket, Kalshi, and Hyperliquid all run the same architecture: a binary order book where every share is collateralized at exactly $1. That cap is structural, not a choice. So the most exciting bet, the longshot, pays the least it possibly can. BASIS runs the alternative: an uncapped parimutuel pool where winners split the entire pot. And it opens the markets order books can't: a 100-way field needs 100 quoted books on an order-book venue. Here one transaction opens every outcome, priced, live from the first dollar.
Others · Capped order book
Maximum payout per share, fixed in advance. Welded to being a regulated, fully collateralized venue, and to the $15B–$22B valuations built on it.
BASIS · Parimutuel pool
Winners split the entire pot, so a winning longshot pays from all of it. In a full order-flow replay of 111 settled Polymarket markets ($422M of real trading), 61% of winning wallets and 71% of winning money would have finished ahead of the order book, and across the 77 longshot-won markets the winning side would have taken 1.33× more out of the pool. Every figure is the pool alone; the market's own token sits on top.
Method: 111 settled Polymarket markets, every trade on every outcome, $422M of notional, fields of 3 to 104 outcomes, replayed under pool rules on the venue's own published order flow with unfavourable results included. Figures are replays of what would have been paid, not live results, and they are reported by field size: on a three-to-six outcome market the book's embedded cost is about 1.3% against our 1.5% fee, close to par. The edge concentrates in early entry, wide fields, and outsiders.
In the replay, money arriving in the first 5% of a market's life would have been paid 1.85× what the order book paid it on the typical field, and 4.95× on fields of 31 or more outcomes, where the pool beat the book in every single market. The measured cost of trading the book rises from 1.3% on small fields to 7.9% on wide ones, against a flat 1.5% pool fee. The pool beats the book up to 70.5 cents, and the window sits at 71 to 78 percent of the pot.
Every outcome opens at 1/N, so the favorite is the mispriced thing. The first wave into a fresh pool is priced favorite arbitrage, not faith.
That inflow fattens the pool and cheapens the underdog hedge against a growing prize.
Hedge volume pushes the favorites cheaper again, reopening the cross-platform gap.
Closing the gap lands the arbitrage capital in the pool: a deeper pool, a bigger prize, an even better hedge. The engine monetizes its own arbitrage. Round again.
Additive, not competitive
BASIS doesn't ask you to leave the platforms you already use. Hold your position on the incumbent, then take the opposite side here. The hedge pays most in exactly the scenario it exists for: a crowd that piled onto the favorite and got it wrong.
We grow when they grow. Our market isn't their market share. It's the entire flow we sit on top of, half a trillion dollars a year and climbing.
What only BASIS can do
Every market mints a Predict+ token on our Stable+ design, a floor that only ratchets up. Take a position on the event's volume itself. Exits are peer to peer, so the pool never drains. It only grows. In replay, early money on wide fields was paid up to 4.95× what a $1-capped venue paid, before the token is even counted. Two engines, one bet.
Because every market is a set of parimutuel pools, one event can carry several markets at once. Live today: winner-take-all fields of up to 150 outcomes, up-and-down markets, and private markets. Built and deploying at Phase 2: podium and top-N, exactas and combinations, parlays, and survivor pools. An order book can't price an exact finishing order at all.
Launch a market on anything with an outcome, then broadcast the event live to the same audience trading it. The creator earns 20% of everything the market does: bets, token trades, leverage, and loans, for as long as it lives. No incumbent does this.
For creators
Launch a market on anything with an outcome (no code, live instantly), then stream the event to the same crowd trading it. You don't just host the moment; you own the entire economy it spins up.
Start creatingNot just the bets. Every fee your market generates flows back to you, for as long as it lives. You built it, so you own its economy.
AI native, not AI assisted
An agent connects a wallet and starts trading in minutes, no scraping, no reverse engineering. Onchain identity via ERC-8004, sponsored gas for registered agents, and first-class access to every market. BNB Chain is the #1 agent chain, and BASIS is the venue built for it.
Explore the SDKResolution
No external oracle. No third-party token whose whales can flip a result. Anyone can propose an outcome by posting a bond they lose if they're wrong.
The rest of the machine
Two products sit at the front: the prediction venue and the token launchpad. Three systems sit underneath and power both: the exchange, the lending desk, and the savings vault. One token, one fee stream, and one enforced floor securing the tokens, the loans, and the leverage. Launching a token costs nothing and requires no liquidity. There is no price liquidation anywhere, on anything. A challenger could fork the pool in a weekend. Catching the machine would take years.
Borrow the full value of your collateral with no lenders. A flash crash can't liquidate you. Only the loan's expiry date can.
No price liquidation, only a time-based expiry. Leverage sits on the protected floor, not the volatile price, so the constraint is the spot-to-floor ratio. 36× is the theoretical maximum, not a constant.
Wrap STASIS into wSTASIS (ERC-4626). Yield from activity, not emissions.
Cliff or gradual schedules. Borrow against locked tokens before they vest.
An up-only base asset whose floor can't move down. The stable asset agents hold to operate.
A rising floor with a free market on top. Selling raises the floor, so the death spiral is impossible.
Our DEX lists only factory tokens: no scam listings, no rug mechanics. The tokens themselves are standard ERC20s and travel anywhere.
One dollar, three jobs
Buy STASIS and stake it. You now earn a share of every fee the platform generates.
Borrow against your staked position at 100% LTV. No price liquidation, only time.
Place your prediction with the borrowed funds. You get the pool position and the market's Predict+ token.
One deposit, three streams: vault yield, token appreciation, and the bet itself. Deadliest on long-dated markets, where competitors' capital sits dead for months.
Real revenue to BASIS stakers, not inflationary emissions. At TGE, 48% of fees - four fifths of the company's current 60% - route to the BASIS staking contract. Creators keep 20% of everything their own markets generate (bets, trades, leverage, loans), forever. The more the platform does, the more holders are paid.
Today
At TGE
Two hard contract limits sit behind the split: the BASIS staking share can never fall below 30%, and every fee setter is capped at 200bps. At the 1.5% headline rate, BASIS stakers take 0.720% of volume, or $7,200 per $1M traded.
Where it's built
Veilon AG, a Swiss company, develops and owns the software, with legal counsel from Ernst & Young. Veilon does not operate the venue. The venue runs offshore across roughly 190 countries and geo-blocks restricted jurisdictions at the interface until the license is in hand, the same offshore-to-onshore arc the category leader walked. Parimutuel is not a novel structure: it is a mechanism regulators have supervised for a century, and Polymarket is the recent precedent.
The contracts are self-audited today, with the public AI-generated review reports on our GitHub. A full third-party audit by Hashlock is the first use of funds and gates real-money settlement. A suite of more than 20 smart contracts, deployed in phases.
FAQ
Calling BASIS a prediction market is like calling Amazon a bookstore. BASIS is a full-stack, agent-native DeFi platform on BNB Chain. Two primary products sit at the front: the uncapped prediction venue and the token launchpad. Three systems sit underneath and serve both: the exchange, the lending desk, and the savings vault. One token, one fee stream, live in public beta. Winners split the entire pool instead of a capped dollar, and it's additive: keep your bet on Polymarket, Kalshi, or Hyperliquid and hedge it here.
All three run a binary order book where every share is collateralized at exactly $1, so payout is capped by design. BASIS is parimutuel: winners split the whole pot, so a winning longshot pays from all of it. In a full order-flow replay of 111 settled Polymarket markets ($422M of real trading), 61% of winning wallets and 71% of winning money would have finished ahead of the order book, and on the 77 markets a longshot won the winning side would have taken 1.33× more out of the pool. Small fields are close to par: the edge concentrates in early entry, wide fields, and outsiders, where early money would have been paid up to 4.95× more. And it's additive: you don't leave the venue you use, you hedge on top of it.
On a capped venue you buy underdog NO, safe and capped. On BASIS you buy the same underdog's YES. If the underdog loses, your capped NO pays and your BASIS cost is small. If the underdog wins, your BASIS YES pays from the entire pool, uncapped. The hedge pays most in exactly the scenario it exists for.
Yes. Every market is a set of parimutuel pools, so one event can carry several markets at once. Live today: winner-take-all fields of up to 150 outcomes, up-and-down markets on five assets, and private markets. Built and deploying at Phase 2: podium and top-N, exactas and combinations, parlays, and survivor pools. An order book can only list yes or no contracts.
Launch a market on any event and broadcast it live in the same place, to the same audience trading it. The creator earns 20% of everything that market generates: bets, token trades, leverage, and loans, for as long as it lives. No other prediction market lets the creator host the event and become the house.
AI does the work, bonds keep it honest, humans are only an appeals layer. Anyone proposes an outcome by posting a bond. An AI resolver calls markets within hours; a second, independent AI audits every call. Disputes go to a bonded panel vote where the outcome with the most votes wins, the public can veto a ruling, and the team is the final appeals layer. No oracle, no third-party token to capture.
Agents connect a wallet and use the Basis SDK (Python or TypeScript), with 595 methods across two SDKs plus an MCP server with 205 tools. Onchain identity comes from ERC-8004 and registered agents get sponsored gas. Passing agent: true auto-registers onchain. Three API calls from zero to earning.
No code. Connect a wallet, open Create New Token, choose Stable+ or Floor+, set the details and starting liquidity, pay the BNB gas (about $0.14), and it deploys instantly and trades immediately. The starting liquidity is virtual, not deposited capital: launching costs nothing beyond gas and requires no liquidity from you. No platform fee to create, and the creator keeps 20% of every fee their market generates for the life of the token, an annuity rather than a launch fee.
Stable+ appreciates through slippage retention with an elastic supply and a price that cannot decrease, the up-only base asset. Floor+ adds a free market on top of a floor that only ratchets upward, tuned by a stability dial the creator sets from 1 to 90 at launch; the dial is the stability percentage, and 100 is Stable+. Because selling raises the floor, the death spiral is structurally impossible. Both are contract properties, not guarantees of outcome: the floor is collateral, not a commitment.
Collateral carries an enforced price floor, so the system collateralizes itself: borrow up to 100% of floor value with no lenders, and no liquidation from price moves. There is no price liquidation anywhere on the platform, on anything. Leverage of up to 36× is calculated against that floor, not the market price, so the real constraint is the spot-to-floor ratio; 36× is a theoretical maximum, not a constant you are guaranteed. A flash crash can't liquidate you; the loan's expiry date can.
It puts one dollar to work three times. Buy STASIS and stake it to earn a share of every platform fee. Borrow against the staked position at 100% LTV with no price liquidation. Bet the borrowed funds and you get the pool position plus the market's Predict+ token. One deposit: vault yield, token upside, and the bet itself.
Stakers receive 48% of all platform fees (trading, lending, and prediction), paid as real USDB, not emissions. Today fees split: company 60%, creator 20%, STASIS vault 16%, reward phase holders 4%. At TGE, 48% of fees - four fifths of the company's current 60% - route to the BASIS staking contract, leaving BASIS stakers 48%, creator 20%, STASIS vault 16%, treasury 12%, reward phase holders 4%. Two hard contract limits back it: the BASIS staking share can never fall below 30%, and every fee setter is capped at 200bps. It's a share of actual revenue, so it grows with adoption. At the 1.5% headline rate, BASIS stakers take 0.720% of volume, or $7,200 per $1M traded.
It is a permissionless protocol: no KYC to use the contracts, with access geo-managed at the interface and restricted jurisdictions blocked. The public beta runs on a test stablecoin (USDB), so no real funds are at risk. The platform runs a suite of more than 20 smart contracts, deployed in phases, on BNB mainnet: MEV-resistant, noncustodial, and verifiable on BscScan. They are self-audited today, with the public AI-generated review reports on our GitHub. A full third-party audit by Hashlock is the first use of funds and gates real-money settlement.
Veilon AG, a Swiss company, develops and owns the software, with counsel from Ernst & Young. Veilon does not operate the venue. The venue runs offshore across roughly 190 countries, geo-blocking restricted jurisdictions at the interface. Parimutuel is not a novel structure: it is a mechanism regulators have supervised for a century, and Polymarket is the recent precedent for the offshore-to-onshore arc.
Prediction markets are the door. The platform is why nobody leaves.
Enter the app