BacktestProof

Analyze NT8, MT5, TradingView backtests — robustness score, overfitting detection, AI key findings.

BacktestProof — Algo Strategy Robustness Analyzer Stop deploying strategies that look great in backtests but fail live. BacktestProof analyzes your backtest exports and gives you an honest robustness score — before you risk real money. What it does Drop your backtest export file into BacktestProof. In seconds, you get a 0–100 robustness score across 19 criteria organized into 4 categories: Performance, Risk, Consistency, and Anti-Overfitting. No account needed. No data leaves your browser. Who it's for Algo traders using NinjaTrader 8, MetaTrader 5, or TradingView. Developers backtesting strategies who want a second opinion. Anyone who wants to know if their backtest results are real or just curve-fitted noise. Supported platforms ✓ NinjaTrader 8 (CSV export) ✓ MetaTrader 5 (CSV, HTML report, XLSX Strategy Tester) ✓ TradingView (CSV export) Auto-detected. No manual column mapping. Works in any language — French, English, Japanese, Russian, and more. The 19-criteria scoring system Performance (25 pts): Profit Factor — is the edge real or marginal? Sharpe Ratio — annualized by actual trade frequency, not a fixed 252 Win Rate — in context with average win/loss Avg Win / Avg Loss — the full picture of expectancy Risk (27 pts): Max Drawdown — as % of peak equity, never inflated Recovery Factor — how fast does it bounce back? Consecutive Losses — worst losing streak Drawdown Duration — how long underwater? Directional Bias — does it only work long or only short? Consistency (28 pts): Profitable Months — what % of months were green? K-Ratio — is the equity curve smooth or erratic? Ulcer Index — how painful were the drawdowns? Outlier Dependency — does the whole performance rest on 3 lucky trades? Edge Decay — is the strategy improving or degrading over time? Time Concentration — is performance clustered in one period? Anti-Overfitting (20 pts): Trade Count — enough trades for statistical significance? Backtest Duration — tested across multiple market regimes? Yearly Variance — consistent year over year or one lucky year? Monte Carlo analysis Grading ≥ 70 → Strong — deployable with proper position sizing 50–69 → Average — profitable but needs refinement < 50 → Weak — significant risks, do not deploy live AI Key Findings (optional, €5 one-time) After scoring, unlock AI-powered analysis for any strategy. Get 3 specific, actionable insights tailored to your exact metrics — what's working, what's dangerous, what to fix. Powered by Claude (Anthropic). One payment unlocks all future analyses permanently. Privacy first Your backtest files never leave your browser. All parsing and scoring happens locally. Only aggregated metrics (score, statistics) are sent to the AI service when you explicitly request it. No tracking, no analytics, no account required. How to use Export your backtest from NinjaTrader, MT5, or TradingView Click the BacktestProof icon in your Chrome toolbar Drop the file — platform auto-detected instantly Read your robustness score and category breakdown Optionally unlock AI Key Findings for deeper insights File formats accepted .csv (NT8, MT5 raw export, TradingView) .html (MT5 Save as Report — any language) .xlsx (MT5 Strategy Tester report) Algo trading is hard. Most strategies that look good in backtests fail live because of overfitting, curve-fitting, or hidden risks that standard backtest reports don't surface. BacktestProof was built to give you a clear, honest signal before you put real capital on the line. BacktestProof is an analytical tool only. Nothing in this extension constitutes financial advice. Past backtest performance does not guarantee future results.