FairLine is a risk-managed, tranched liquidity vault for DeepBook Predict on Sui. Be the house — pick your risk (protected senior or leveraged junior), earn the structural house edge, and verify every number on-chain.
FairLine began as an ML bot betting on direction. Live on testnet, it lost money: the spread was wider than the model's edge. On-chain, the lost money went to one place — the liquidity pool. The house. So we became the house, and turned the model into a defensive risk gate.
−830 dUSDC live. A 63%-accurate model still loses when the 2% spread beats the edge. The directional sleeve is now retired — the signal defends the house, it doesn't bet.
LP-primary: supply liquidity, earn the spread the losers pay. The ML/volatility signal now gates exposure against risk instead of gambling with it.
Prediction markets need liquidity, and the pool that provides it earns the spread every losing trader pays. But providing it well takes automation, risk management, and capital discipline most people don't have. FairLine packages that into one deposit.
Want prediction-market yield without running a bot? Deposit dUSDC into the senior tranche for protected, steady yield — or junior for leveraged upside. Withdraw anytime. The vault is the house, and now so are you.
The house edge is structural, not a bet — the 2% spread is paid by traders on both sides. It compounds into the FLP share price over time, on-chain and measurable.
An LP's real danger is a big directional move. FairLine's ML/volatility gate scales exposure down exactly then — turning a model that can't profit betting into one that's valuable defending.
No leverage, no emissions farming — just the structural edge a prediction market pays its liquidity pool.
Every BTC option on DeepBook Predict carries a ~2% spread, and traders lose to the house on average. Both accrue to the liquidity pool's reserves.
Deposit dUSDC and FairLine supplies it to that pool. You receive FLP-S (senior) or FLP-J (junior) shares, priced at the vault's net asset value.
As the edge accrues, the pool's redemption rate climbs — so your shares grow in value (junior leveraged, senior capped + protected). Withdraw any time.
The house wins on average, not every cycle — a large directional move can dip the share price, which is exactly what the ML/volatility gate guards against. Steady, modest yield — not a moonshot.
Supplies ~70% of capital to DeepBook Predict's pool to earn the structural house edge. Sticky by design — it sizes position by risk rather than churning in and out.
A trained model (5-fold CV) used defensively: a live Green / Amber / Red posture scales liquidity down when a large move is likely. The losing alpha model becomes a useful risk model.
Our own Move contract: deposit dUSDC into the senior or junior tranche, receive FLP-S / FLP-J priced at NAV, withdraw pro-rata. Real depositors, real TVL.
FairLine isn't a yield aggregator — it's the house of a prediction market, with the structure to match.
Senior is principal-protected by junior's first-loss buffer and earns a capped, steady yield (≤8% APR). Junior absorbs losses first and takes the leveraged upside. Pick your risk.
Every deposit and withdrawal prices at a freshly-marked NAV, recomputed independently from the on-chain PLP rate — so you never enter or exit at a stale, unfair share price.
The house edge doesn't scale forever, so the vault closes the door when more capital would dilute everyone's yield. We optimize depositor yield, not AUM — an on-chain cap, not a marketing promise. (A market maker always wants more capital; FairLine sometimes refuses it.)
FairLine routes a slice of its edge into rebates for predictors trading the markets it backs. More rebates → more volume → more edge → more rebates.
The same risk gate runs a live maker on DeepBook's core CLOB — resting real limit orders on the DEEP/SUI book, posture-gated and inventory-skewed. One risk brain, two venues: the prediction market and the orderbook.
These numbers are fetched in your browser directly from Sui's public RPC, right now — refreshing every 15 seconds. NAV is derived from the on-chain PLP redemption rate, not operator-set, and split across the senior and junior tranches.
Don't trust — verify. Every step is a real transaction you can open on the explorer.
DeepBook's MM makes sure a market has quotes. FairLine decides whether you get to own the house behind them — and at what risk. Different layer, not a competitor.
DeepBook's market maker solves liquidity — quotes from minute one. That's infrastructure. FairLine is the structured-product layer on top: who gets to be the house, and on what risk terms.
A market maker always wants more capital. FairLine caps capacity and turns deposits away when they'd dilute yield — because we optimize the depositor's return, not our AUM. That's a product, not a maker.
Not liquidity — a risk-tranched slice of house P&L. Senior is protected by junior's first-loss buffer; junior takes the leveraged upside. The waterfall is enforced on-chain.
Credibility is a feature. We never overclaim.
Live on Sui testnet. The Predict-house sleeve can't go to mainnet until Mysten ships DeepBook Predict there — but the CLOB-maker sleeve targets DeepBook's core orderbook, which is already on mainnet. Partial-ready, not faked.
The contract passed an internal security review. The whole trust surface is one number — the operator's mark on deployed capital. Reserve is trustless, settlement can't claim more than was returned, every mark is timestamped on-chain. We tighten it with a redemption-anchored settle that realizes NAV to chain-enforced reserve each checkpoint. No "audited" claims.
Every decision and transaction is on-chain. The numbers on this page come straight from the chain — check them yourself.
Deposit dUSDC into the senior or junior tranche, hold FLP-S / FLP-J, withdraw anytime — on Sui testnet, right in your browser or phone.