Upload a TradingView or MetaTrader backtest. We recompute its numbers and cross-check them against what the report claims, run integrity lie-detectors, and stress-test the edge — then return a Backtest Health Score. It measures integrity and robustness, never profitability. We don't run backtests and we don't store your trades.
Your trade data is processed to compute the score and is not stored. There is no “connect TradingView” — export your list of trades and upload it.
Re-derives net profit, profit factor, win rate and drawdown from the trade list and flags any mismatch with the report's own summary.
Impossible chronology, zero-loss samples, near-perfectly-straight equity curves, and single-trade-dominated edges.
Removes the best trades one by one — if the edge flips negative after a couple, it was a lucky streak, not a system.
Bailey & López de Prado: adjusts the Sharpe for how many variants you tried and how long the sample really needs to be.
Reshuffles & resamples your realized trades thousands of times → the drawdown you could have had, terminal-equity fan, risk of ruin, and where your result sits on the luck curve.
Keep a private history of every check — verdict & metadata only, never your raw trade data, deletable any time.
It's a free tool that measures whether a backtest is internally consistent and robust — not whether it's profitable. We recompute the report's own numbers from its trade list, flag mismatches and structural red flags, and stress-test the edge by removing the best trades. Nothing here is a prediction or trading advice.
No — TradingView has no API for Strategy Tester data, so nobody can. You export your List of Trades (CSV) or MetaTrader statement (HTML) and upload it. We process it to compute the score and do not store your trade data.
No. The score measures integrity and robustness only — whether the numbers reconcile and whether the edge survives scrutiny. A robust-looking backtest can still lose live; an honest, modest one can score well. We never claim or estimate profitability.
Below about 20 trades we refuse to assign a score, because every metric is dominated by luck. Even at 30–100 trades the error bars are wide — we show the trade count prominently and gate each metric on it.