crypto backtesting
Real candles, fees, funding and exchange rules with minute-level fills. Every result comes with an honesty grade and an overfit verdict, so you know which numbers to trust.
Official Binance broker · Non-custodial · Live only when you confirm
sample data
A strategy that peeks one bar ahead, fills at the perfect price, ignores funding, or wins after a hundred parameter tweaks on the same data looks brilliant and fails live. OptAlgo checks for each of these and tells you, in plain words, which ones apply.
A: every fill settled on real data with fees, funding and exchange rules. B: a few standard assumptions. C: usable, read the findings (for example a short warm-up or missing coverage). D: not realistic, fix and rerun. The result lists what was simulated, what wasn't, and how it differs from live trading.
OptAlgo counts the variants of your strategy you ran on overlapping data in the last 30 days and computes the probability that your Sharpe beats what the best of that many worthless variants would reach by luck (Bailey & López de Prado). Verdicts: ok, caution, likely overfit, or no edge when the run lost money.
Walk-forward, nested walk-forward, purged cross-validation, CPCV, Monte Carlo and permutation tests run in one job, with a verdict: robust, inconclusive or likely overfit. Baskets test the same rules on several coins at once.
A zoomable price chart with every entry and exit, equity and drawdown curves, return by year, fees and funding totals, and downloads for the trades CSV, the result JSON and the exact code that ran.
faq
Exchange candles down to 1 minute, with real fees, exact funding payments and exchange rules such as minimum order sizes. The list of symbols, timeframes and history start dates is live in the app and through the markets tool.
Validation is always free. The Free plan includes 50 backtest units a month (1 unit is up to 10 seconds of run time). Plus has 30 units a day and Pro 200 a day, with longer history and bigger baskets.
Yes. Strategies are vectorised Python with numpy and the OptAlgo SDK. Write it yourself or let Claude, ChatGPT or Codex write it. The code runs in a sandbox and is checked for lookahead and determinism first.
A Sharpe ratio adjusted for how many strategies you tried. The more variants you test on the same data, the higher the best one scores by luck alone. The deflated Sharpe probability tells you whether your result clears that bar.