How to read a backtest report without lying to yourself

A backtest report is a very confident-looking document produced by a very literal machine. It answers the question you asked of it and has no opinion about whether that question was worth asking. Reading it well is a skill, and most of it is knowing which number is trying to flatter you.

Start with sample size, not profit

Total trades come first. Fifty trades and five hundred trades are different kinds of evidence: with fifty, a handful of lucky fills can carry the entire result, and removing one trade should visibly change the curve. If your best result disappears after deleting its best trade, you found the strategy's owner — randomness.

Equity curve: shape over slope

A smooth 45-degree curve is a warning sign, not a dream. Real equity lurches, stalls, and gives back. Look for the flat stretches and the long troughs in the middle, because that is the part you will actually experience. The slope is what people photograph; the shape is what people quit on.

Drawdown is the number you pay in

Max drawdown is usually the difference between a strategy you keep trading and one you abandon mid-run. Two questions do most of the work: how long was equity underwater, and how deep was it at the moment you would have doubted it. Duration is rarely discussed and routinely underestimates the psychological cost.

Then size it against your own tolerance, not the average. A drawdown that is survivable in a simulation is survivable because no money was at risk while you watched it.

MAE and MFE: the two numbers traders ignore

Maximum Adverse Excursion is how far a trade went against you before it resolved. Maximum Favorable Excursion is how far it went for you.

Put them on a scatter and a strategy starts explaining itself. Trades that survive a large MAE before reaching profit are telling you your stop is too tight or your entry is late. Trades whose MFE sits far beyond the exit are telling you the exit gives away ground. These two columns answer questions about stop placement that win rate cannot address at all.

The monthly matrix, and the shape of luck

A matrix of monthly returns shows you what an average hides. Three months carrying two years of profit is a different strategy than two years of small consistent months, even when the totals match. When a year looks good, find out which two months did it — and then ask whether those two months were a regime you expect to see again.

Long/short and exit-reason breakdowns

A strategy that only makes money on shorts in a falling period is a directional bet wearing a strategy's clothes. Same for exits: if every winner is a trailing-stop exit and every loser is a time exit, you know exactly which part of the rulebook is doing the work.

Then ask the only useful question

Not "was it profitable?" but "would I have believed this while it was losing?" Backtest reports describe the past accurately and the future not at all. The value of a report is not a forecast — it is a cheap, fast way to disqualify an idea before it costs real money, and a honest record of the assumptions you agreed to when you ran it.