Backtest AI Labs
Crypto strategy backtests on real historical market data
Every strategy below is a deterministic simulation — the same inputs always produce the same result, and no model inference runs during a backtest. Each result ships with a Methodology panel stating its data sources, which costs are measured, which are fixed assumptions, and which risks are not priced at all. Read it before acting on any number here.
Backtest Rally Card Ideas → Neural Trading (BCI) →Cross-Exchange Arbitrage Live Data
Compares hourly closing prices across every discovered pool for a token and simulates buying the cheapest venue and selling the dearest. Gas, bridge and latency costs are fixed estimates, not historical measurements.
A gap between two hourly closes is not proof a trade was executable at the same instant. MEV, failed transactions and router paths are not modelled.
Funding Rate Carry 3 Exchanges
Simulate delta-neutral funding rate arbitrage across Hyperliquid, Binance and Bybit, using each venue's actual historical funding rates at its own settlement interval and its real taker fee.
Delta-neutral is not risk-free. Being long one venue and short another reduces directional exposure; it does not remove basis, liquidation, execution, funding-inversion or counterparty risk — and this simulation prices none of them. Leverage above 1x scales the modelled income linearly and the unmodelled liquidation risk faster than that.
DEX AMM Simulation 3 Strategies
Backtest mean reversion, momentum or LP-vs-Hold against a modelled AMM. Price comes from hourly closes; slippage from constant-product math.
This is not an order-book replay and it does not read historical pool reserves. Depth is derived from the pool's present-day liquidity held constant across the window. Concentrated-liquidity (V3) tick math is not implemented.