Running a Backtest
What happens during the six-second validation and how to read the output.
Starting a run
From Strategy Lab, choose your date range and starting capital, then press Run backtest. The engine loads candles, replays every rule bar by bar, records each fill with fees and slippage, and streams progress as it goes.
Free tier does not include backtests. Pro allows fourteen per week and Elite forty, with instant activation.
The seven gates
Every run passes through the same sequence: sample size (at least thirty closed trades), walk-forward efficiency (0.50 or better), Monte Carlo survival (P95 drawdown within tolerance), parameter stability (a plateau rather than a spike), overfitting audit (PBO below 0.15), realistic execution (fees and slippage applied, pessimistic fills available), and positive out-of-sample expectancy (still profitable on the data it wasn't tuned on).
A run that clears all seven earns the Golden Badge. A run that fails any one is reported as withheld, with the failing metric and its threshold shown explicitly.
1RUN 6afd09f3 SOL · 1H · 12MO 5.8s23✓ SAMPLE SIZE 61 trades (≥30)4✓ WALK-FORWARD 0.70 (≥0.50)5✓ MONTE CARLO P95 -17.2% (≤25%)6✓ PARAMETER STABILITY plateau7✓ OVERFIT AUDIT (PBO) 0.11 (<0.15)8✓ EXECUTION fees + slippage ON9✓ OOS EXPECTANCY positive1011◆ GOLDEN — all seven gates passed
Reading the report
The Overview tab shows equity against buy-and-hold with the out-of-sample region shaded. Drawdown is plotted underwater from peak. Per-trade P&L shows every loser at the same visual weight as every winner.
The Trades tab lists each fill with entry, exit, duration and cost. The AI Explanation tab has Quant AI walk through what the engine found in plain language.
If a strategy closed 23 trades, the report says 23. We never filter the trade list to improve the picture.