Backtesting that tries to break your strategy

Trade by trade, on the candles you can see

The rules replay over the same candle history the terminal charts — not a smoothed series, not a vendor index — and every entry and exit is marked on the chart at the bar where it happened. You can scroll back and look at the trade you are suspicious of instead of trusting a summary.

Three numbers, never one

A single return figure is the easiest thing in finance to make look good. Every backtest reports the trade count, the win rate and the maximum drawdown together, because the three of them are the smallest set that can catch a bad strategy.

• Trade countHow much evidence there actually is behind the result.
• Win rateHow often it was right, which alone tells you very little.
• Max drawdownThe worst run of losses — the number that decides whether you could have held on.

A small sample is called a small sample

When a backtest returns too few trades to mean anything, the result carries a warning saying so in those words: weak evidence, not a performance claim. This is deliberate and it is the opposite of what a tool trying to sell you a strategy would do. Twelve winning trades is a coin that came up heads twelve times, and you should be told that rather than left to work it out after it costs you money.

What happens after the backtest

Nothing goes live straight from history. A strategy that survives the replay runs forward on live prices with imaginary money first, meeting candles nobody has seen yet. Only after that does it start sending live alerts.

What a backtest cannot tell you

It cannot tell you the future, and it cannot fully model slippage, a gap through your stop, or the day you decide to override the rule. Backtesting is evidence against a bad idea rather than proof of a good one. Treated that way it is the most useful hour you will spend on a strategy; treated as a promise it is the most expensive.