How many trades does a backtest need?
The short answer
Aim for at least 100 trades, and more is better. With 30 trades your win rate could be off by about 18 points either way. With 100 trades it is about 10 points, and with 500 about 4. Test across different market conditions, then confirm with walk-forward and Monte Carlo tests.
Last updated
How sample size changes your margin of error
Margin of error on the win rate at 95% confidence, for a strategy that wins about half its trades.
| Trades in backtest | Win rate could be off by |
|---|---|
| 30 | ± 17.9 points |
| 50 | ± 13.9 points |
| 100 | ± 9.8 points |
| 200 | ± 6.9 points |
| 500 | ± 4.4 points |
| 1,000 | ± 3.1 points |
More than just the number of trades
- Cover different markets: rising, falling and sideways periods.
- Use walk-forward testing: build on one period, test on a later one.
- Run a Monte Carlo test to see a range of possible outcomes.
- Avoid look-ahead bias: only use data known at the time.
- Include realistic costs before you trust the result.
Free tools
Good to know
Past results do not guarantee future results. A large sample makes a backtest more reliable, not certain. Nothing here is financial advice.
Frequently asked questions
Is 30 trades enough for a backtest?
Usually not. With 30 trades, a win rate could be off by about 18 points either way, so luck can look like skill.
How many trades make a backtest reliable?
Aim for at least 100 trades, and several hundred if you can, spread over different market conditions.
What else should I check besides the number of trades?
Use walk-forward and Monte Carlo tests, avoid look-ahead bias, and include realistic costs. AlgoBars includes these in its free backtesting.
More answers
Start free. Today.
Every feature, free. No card. No catch.