ARMFMetrics of a forex trader's trading system
The metrics of a trading system and the statistics of trades answer one question: does the strategy work at a profit and at what price that profit came. Individually none of them describes the system — a losing strategy can have a high win rate, and a large profit factor over ten trades means nothing. We take the eight figures apart together.
A trader's trade statistics: eight figures and what each one measures
| Metric | Formula | What it shows | What it does not show |
|---|---|---|---|
| Win rate | winning ÷ all trades | The frequency of wins | Nothing about the size of a win |
| The R/R ratio | average win ÷ average loss | The asymmetry of outcomes | How often the target is reached |
| Expectancy | p × R − (1 − p) | The average result of a trade in R | The spread around the average |
| Profit factor | earned ÷ lost | The return per unit of losses | How profits are distributed over time |
| Maximum drawdown | maximum (peak − equity) ÷ peak | The worst moment lived through | The probability of it repeating |
| Recovery factor | profit for the period ÷ max drawdown | The ratio of result to risk | The duration of drawdowns |
| Sharpe ratio | (return − risk-free) ÷ standard deviation | The result per unit of fluctuation | Asymmetry: it punishes growth too |
| Number of trades | simply the count | The reliability of every other metric | Nothing about the quality of the system |
The last row is not a formality. All the metrics above are estimates from a sample, and over twenty trades their spread is such that a losing system easily looks profitable. The reference for first conclusions is a hundred trades, for confident ones several hundred.
How the metrics are linked to each other
The win rate, the R/R ratio and the profit factor are not independent values. Knowing two, the third can be computed, and that is a useful check: if the numbers calculated from the journal do not add up, there is an error in the record-keeping somewhere.
profit factor = (win rate × average R) ÷ (1 − win rate)
| Win rate | Average R | Expectancy, R | Profit factor | Verdict |
|---|---|---|---|---|
| 30 % | 3.0 | +0.20 | 1.29 | Profitable, but with rare wins |
| 40 % | 2.0 | +0.20 | 1.33 | Profitable, the working mode |
| 45 % | 2.0 | +0.35 | 1.64 | A good buffer above break-even |
| 50 % | 1.0 | 0.00 | 1.00 | Zero before costs, losing after them |
| 60 % | 1.0 | +0.20 | 1.50 | Profitable thanks to frequency |
| 70 % | 0.4 | −0.02 | 0.93 | Losing, despite the high win rate |
The last row is the typical trap. A system with seventy percent winners looks convincing, but if the average win is less than half the average loss, the expectancy is negative. That is exactly how strategies with a distant stop and a close target work: a long series of small profits and rare large losses.
Reference points: which values count as working ones
The boundaries are conditional and depend on the trading style, but the order of magnitude is useful as a first check.
How to read other people's numbers. A profit factor without the number of trades and the period means nothing. Before comparing systems, bring them to a common denominator: how many trades, over what period, including which costs and on which instrument.
How to read a forex trader's metrics together rather than separately
A single figure almost always allows two opposite explanations. Pairs of metrics remove the ambiguity: below are the four combinations that occur most often and the conclusion for each.
Frequently asked questions
Which metric matters most?
Expectancy together with the number of trades: the first shows whether there is an edge, the second whether that value can be trusted. Everything else refines the picture but does not replace these two.
What is the Sharpe ratio for in trading?
To compare systems with different sizes of fluctuation: it shows at what price in return dispersion the result was obtained. For trading systems with rare large wins the figure is understated, so it is looked at together with the maximum drawdown rather than instead of it.
What is the Sharpe ratio in simple words?
It is the ratio of the result above the risk-free rate to its fluctuations. The higher it is, the smoother the result curve was. Its main drawback: it punishes drawdowns and sharp growth equally, so for systems with rare large wins it is understated.
How many trades are needed before the metrics can be trusted?
For first cautious conclusions about a hundred, for stable estimates several hundred. Over twenty trades the spread is so wide that a losing system is equally likely to show good and bad numbers.
Can the metrics of different systems be compared?
Only under identical conditions: the same period, the same treatment of costs, a comparable number of trades. A system tested without the spread is not comparable with one traded on a live account.
Which metrics should be counted separately by currency pair?
The win rate, the average R and the profit factor. The overall figure often rests on one pair while the rest run at a loss: a breakdown by instrument shows this at once and usually leads to a shorter list.
Does the swap have to be included in the metrics?
Yes, if positions are carried overnight. The swap reduces the result of a trade regardless of the price move, so excluding it inflates both the average R and the profit factor.
How do you count metrics when trade sizes differ?
In units of risk rather than in money. A result in R does not depend on the size or on the account, so trades with different lots become comparable.
What is the recovery factor?
The ratio of the profit for a period to the maximum drawdown. It shows at what price in risk the result was obtained: a value of 2 means twice as much was earned as the worst dip along the way.
How do a forex trader's metrics differ from exchange ones?
In the composition of the costs and the working regime: the overnight charge for holding a position and the widening of the spread in inactive hours land directly in the result of a trade. The formulas themselves — win rate, expectancy, profit factor — are the same.
Which metrics are useless on a small sample?
All of them, but especially the profit factor and the maximum drawdown: the first breaks from one large trade, the second simply has not happened yet. Over twenty trades only the number of trades and adherence to the calculated risk are meaningful.
Read next
