Definition of Average Win Loss
Average Win Loss is a performance metric that summarizes how large wins are relative to losses by using averages. In practice, you take the set of winning trades and compute their average outcome, then take the set of losing trades and compute their average outcome, and compare the two.
A typical question it addresses is: “When this system has a winning trade, how much does it usually make, and when it has a losing trade, how much does it usually lose?” This is especially useful in forex because trade outcomes can differ widely depending on volatility, position size, and execution quality.
How Average Win Loss works in forex
Start with clearly defined trade outcomes. For each closed trade, compute a realized profit or loss in your chosen measurement (for example, account currency units, or pips converted to money using your position size). Then separate trades into two groups:
- Winning trades: trades with a realized profit above zero.
- Losing trades: trades with a realized loss below zero.
Next, compute:
- Average Win = (sum of profits from winning trades) ÷ (number of winning trades)
- Average Loss = (sum of losses from losing trades) ÷ (number of losing trades)
Because losses are usually negative, many people compare magnitudes by taking the absolute value of the average loss. For example, if Average Win is +10 and Average Loss is −7, then the magnitude comparison would treat the loss magnitude as 7.
Simple worked example (with explicit assumptions)
Assume you have 5 closed trades with realized outcomes in account currency: +20, +10, −5, −15, +5.
- Winning trades: +20, +10, +5 → Average Win = (20 + 10 + 5) ÷ 3 = 35 ÷ 3 ≈ +11.67
- Losing trades: −5, −15 → Average Loss = (−5 + −15) ÷ 2 = −20 ÷ 2 = −10
A magnitude comparison would show average wins of about 11.67 versus average loss magnitude of 10.
What it measures—and what it does not measure
Average Win Loss helps you review the typical size of profitable versus unprofitable trades. It does not, by itself, describe:
- whether wins happen more or less often
- whether the strategy’s total results are positive over time
- whether the distribution is stable (for instance, whether results rely on a few unusually large wins or avoid unusually large losses)
That’s why Average Win Loss is usually interpreted together with other information, such as win rate and overall expectancy (defined from realized outcomes).
Limitations and failure modes
Several limitations can make Average Win Loss misleading if you use it as a standalone headline number.
- Definition and filtering can change the result. If you include or exclude certain outcomes (partial closes, commissions, swaps, or trades closed due to non-market reasons), the average can shift even if market behavior is unchanged.
- Costs and execution matter. If realized profit/loss does not include all relevant costs for your setup, the “average win” and “average loss” can be overstated.
- Outliers can dominate averages. A small number of rare large wins or rare large losses can strongly affect the average, even if most trades look similar.
- Changing conditions break historical stability. Market volatility, liquidity, and spreads can change across time, so historical relationships do not establish future results.
One practical failure mode is when Average Win Loss looks favorable because average losses are smaller in magnitude, while overall performance still deteriorates due to a lower win rate or a cost increase that reduces net outcomes.
How to verify Average Win Loss independently
To verify Average Win Loss for your own dataset, you can reproduce the calculation from your trade ledger:
- Collect the realized profit/loss for each closed trade.
- Clearly label each trade as a win or a loss using your definition (e.g., profit > 0 versus profit < 0).
- Compute Average Win and Average Loss using the formulas above.
- Check sensitivity by repeating the calculation after changing one assumption at a time (for example, including vs excluding certain costs, or using partial-trade handling consistent with your ledger).
If your results vary greatly under reasonable definition choices, that variability is itself an important signal about how dependent the metric is on process details.