Start with the equity curve
Before any ratio, look at the shape of the equity line. Ratios compress a trading record into one number and hide the path it took.
Two records with identical profit factor can be entirely different experiences. One rises steadily. Another reaches the same final value after a drawdown that would have forced most people out. The ratio does not distinguish them. The curve does.
Drawdown is the number to respect
Two properties matter and they are constantly confused.
- Maximum drawdown is the single worst peak-to-trough decline on record. A historical fact about one sample.
- Drawdown duration is how long it took to recover to the previous peak. This determines whether a system is survivable in practice.
A method with a modest maximum drawdown and a four-year recovery period is a method most traders abandon long before it works.
Ratios and what they hide
| Measure | What it tells you | What it leaves out |
|---|---|---|
| Profit factor | Gross wins against gross losses | Whether it came from many small wins or one outlier |
| Expectancy | Average result per trade | How variable those results are |
| Sharpe ratio | Return per unit of variability | The distribution behind the average |
| Win rate | Share of profitable trades | Payoff ratio, which matters far more |
A 30 percent win rate with a four-to-one payoff is a strong record. A 70 percent win rate with a one-to-two payoff loses money. Win rate alone means almost nothing.
The sample size problem
Twelve trades in a month can produce an excellent-looking expectancy by chance. The variance of a small sample is enormous.
Fewer than thirty observations is noise. Thirty is where a difference becomes worth examining, not where it becomes reliable. Many strategies that look profitable over a quarter do not survive a full year.
Never optimise on the same data you review
Tuning parameters until the historical record looks clean guarantees a curve-fit and nothing else. Keep part of the history untouched and use it as the only honest test.
Breakdowns worth running
- By session. Regular hours versus overnight on index futures often behaves like two different strategies.
- By instrument. One contract carrying a result is usually carrying the whole result.
- By trade type. Breakouts, pullbacks and reversals are different methods and should not share statistics.
Statistics are descriptive. They tell you what your record contains, which is genuinely more than most people know — and nothing about what it will contain next.
Next: what a terminal should provide