Backtesting is the process of applying a defined trading or investment rule to historical data to estimate how it would have behaved under stated assumptions about signals, fills, costs, and position management.
How it works
A backtest converts historical observations into hypothetical signals and outcomes using rules that should be fixed before evaluation. It should record every eligible sample rather than only memorable chart examples.
Useful outputs include the distribution of trade results, win rate, average win and loss, drawdown, expectancy, turnover, and sensitivity to costs and assumptions.
Why it matters
Backtesting provides more evidence than a handful of anecdotes and exposes implementation ambiguities that are easy to hide in prose.
It is not live trading. Bad data, look-ahead bias, overfitting, survivorship bias, unrealistic fills, and ignored costs can make a backtest materially misleading.
A simple market example
A rule is applied to 24 consecutive historical signals. Running every signal reveals losing streaks and drawdowns that would disappear if only the best charts were selected.
Common mistakes
Skipping inconvenient historical signals after seeing their outcomes.
Optimizing the same sample repeatedly and treating the best curve as independent evidence.
Frequently asked questions
What is a good backtest sample size?
There is no universal number; it depends on trade frequency, regime diversity, and the uncertainty of the statistic being estimated.
Does a profitable backtest prove an edge?
No. It is evidence to investigate, not proof of future profitability.
Should costs be included?
Yes. Spread, commissions, slippage, financing, and realistic fill assumptions can materially change results.
Educational content only. Definitions describe common market usage and may vary by jurisdiction, instrument, or institution.