How Many Trades Do You Need Before You Can Trust Your Stats?
Last updated: November 2026 · 4 min read · Educational content, not financial advice
Key takeaways
- A strategy with no edge ends in profit about half the time, whatever the sample size.
- After 50 trades a measured 40% win rate could be a true 26%–54%; narrowing that takes hundreds of trades.
- The smaller your edge, the more trades you need to confirm it, so treat results from 10–30 trades as noise.
Most traders judge a strategy after a few dozen trades. The trouble is that luck has a huge influence over small samples. A strategy with no edge can look brilliant, and a strategy with a real edge can look broken. This guide puts numbers on that so you know what your results can and cannot tell you.
A strategy with no edge is profitable about half the time
Here is the chance that a strategy ends a run of trades in profit, using a 1:2 risk-to-reward ratio. These figures are exact binomial calculations with independent trades:
| Strategy | 20 trades | 50 trades | 100 trades | 200 trades | 500 trades |
|---|---|---|---|---|---|
| No edge (33.3% win rate, expectancy 0R) | 52% | 51% | 48% | 51% | 50% |
| Losing edge (30% win rate, −0.10R) | 39% | 32% | 22% | 16% | 5% |
| Small real edge (40% win rate, +0.20R) | 75% | 84% | 91% | 98% | about 100% |
Read the first row carefully. A strategy with exactly zero edge ends in profit about half the time at every sample size (the small differences come from trades being whole numbers). A strategy that loses money on average still shows a profit in roughly four runs out of ten at 20 trades. And a strategy with a genuine +0.20R edge still ends 20 trades in the red about a quarter of the time. Twenty trades cannot separate skill from luck.
How uncertain is your win rate?
A measured win rate is an estimate. A rough 95% range around it is:
| Trades (n) | If you measure 40% | If you measure 55% |
|---|---|---|
| 20 | ±21.5 points | ±21.8 points |
| 50 | ±13.6 points | ±13.8 points |
| 100 | ±9.6 points | ±9.8 points |
| 200 | ±6.8 points | ±6.9 points |
| 500 | ±4.3 points | ±4.4 points |
After 50 trades, a measured 40% win rate is consistent with a true win rate anywhere from about 26% to 54%. That range includes both a losing strategy and a very good one. It takes several hundred trades to narrow it to a few points.
How many trades to confirm an edge?
Suppose your break-even win rate is 33.3% (at 1:2). To be reasonably confident (about 95%) that your true win rate is above that line, the margin of error has to be smaller than the gap between your measured win rate and the break-even rate. For a win rate near 40%, that gives:
| Measured win rate above break-even | Trades needed |
|---|---|
| 20 points (for example 53% vs 33%) | about 24 |
| 15 points | about 41 |
| 10 points | about 93 |
| 5 points (for example 38% vs 33%) | about 369 |
The smaller your edge, the more trades you need to be sure it exists. Thin edges are hard to prove, and they are also the ones that costs can easily erase.
How uncertain is your expectancy in R?
Expectancy has its own uncertainty, because wins and losses have different sizes. For a strategy with a true 55% win rate at 1:2 (expectancy +0.65R), the spread of single-trade results is about 1.49R. The 95% range for the measured average is about 1.96 × 1.49 ÷ √n:
| Trades | 20 | 30 | 50 | 100 | 200 | 500 |
|---|---|---|---|---|---|---|
| Margin of error | ±0.65R | ±0.53R | ±0.41R | ±0.29R | ±0.21R | ±0.13R |
After 30 trades, a strategy whose true expectancy is +0.65R can easily measure anywhere between roughly +0.1R and +1.2R. After 20 trades, the range even touches zero. Only after a couple of hundred trades does the measurement settle within a fifth of an R.
Practical rules
- Do not change a strategy based on 10–20 trades. The results are mostly noise.
- Aim for 100+ trades of the same setup before calling a number reliable, and several hundred before relying on a small edge.
- Record every trade, including the ones you would rather forget. Dropping losers inflates the numbers.
- Test on data you did not use to design the rules. A setup tuned to fit past charts will look better there than it will live.
- Size positions as if your edge could be smaller than measured. The margins of error above are an argument for modest risk per trade.
The assumptions behind these numbers
The calculations assume independent trades, a constant win rate and a constant risk-to-reward ratio. Real results include changing market conditions, which add more uncertainty, not less.
Try the numbers yourself
Set the number of trades to 20, then 100, and watch the outcomes spread.
Related guides
This guide is for education only. It is not financial advice or a trade recommendation, and trading involves substantial risk of loss. See the Terms of Service.