About the Competition

Task

You will build and deploy an autonomous trading algorithm that trades a fixed universe of perpetual futures on the Hyperliquid testnet, via its public API. Your algorithm will manage the full trading lifecycle: signal generation, order placement on a live order book, position and margin management, funding-rate dynamics, and risk control across assets with different trading regimes. The tradable universe will span multiple asset classes. Further details on the assets and oracles will be shared soon.

The Trading Platform

Hyperliquid is a layer-1 blockchain whose core state is a fully on-chain central limit order book supporting perpetual futures with cross and isolated margin, funding, and liquidation. Its HIP-3 standard allows an approved deployer to launch and operate independent perpetual markets — choosing the underlying, the oracle, contract specifications, margin parameters, and fee schedule — on the shared matching engine. HIP-3 is available on the Hyperliquid testnet, which mirrors mainnet functionality with valueless test tokens distributed via a faucet.

You'll trade on a dedicated HIP-3 perpetuals exchange on the testnet. The tradable universe will comprise multiple asset classes, requiring your algorithm to handle several phenomena such as regime heterogeneity, calendar effects, oracle-update discontinuities, and cross-asset risk.

Market-making bots will quote continuously in all markets to guarantee baseline liquidity, and per-market position limits will prevent any single team from cornering a book. Your team will receive one whitelisted API wallet, funded with a testnet balance identical to every other team — top-ups aren't permitted. Technical instructions and a backtesting dataset will be provided soon, so you can test your strategy ahead of the competition.

Eligible Teams

Eligible Teams: Teams must be made up of 1 to 4 members. Should the team be selected as a winner, at least one member must be available to attend ICAIF 2026 in Milan in person to present its work.

Evaluation Protocol

Teams are evaluated according to three criteria:

• Total net return on initial capital, marked to oracle prices at the end of the Live Trading phase, min–max normalized across teams.
• Negative of the standard deviation of daily equity returns (lower volatility scores higher), min–max normalized.
• Jury score on a 2–4 page technical report: methodology and depth of AI/ML techniques, originality, rigor and reproducibility, and clarity. Each report is scored independently by at least three jurors.

Fairness and Anti-Gaming Rules

One trading account per team; accounts are whitelisted and monitored on-chain. To be ranked, a team must trade on at least 10 distinct days and reach a minimum cumulative notional volume. Wash trading, self-crossing, collusion between teams, and attacks on the exchange or oracle infrastructure are prohibited. All trading activity is recorded on-chain and can be reviewed for violations, which lead to disqualification. Fully manual trading is not allowed: entries must be algorithmic, as verified through the technical report and a code walkthrough with the jury.

Live Leaderboard

A public web leaderboard, updated in near real time from on-chain data, will track team performance throughout the Live Trading phase.

Awards

The top teams will be invited to present at the ICAIF 2026 Competition Track in Milan. Prize details will be announced soon.