Average win is the mean result of your winning trades only, average loss the mean of your losers only. These two figures describe the shape of your trading far better than P&L: they say whether you live off a few big hits or a steady accumulation.
What the ratio compares
Contents
These two averages are computed on distinct populations, and that is the whole point. Mixing winners and losers into a single average gives expectancy, which answers a different question. Kept apart, the two figures say how your trades end, on each side.
Average loss is the more informative of the two, because it is supposed to be chosen. Your stop determines what you lose: if average loss hugs your planned risk, your stops hold; if it clearly exceeds it, they are widened mid-trade or skipped, and that is the most expensive flaw in the job.
Average win is largely imposed: the market decides the size of the move. What you control is the share of it you take, and that is what exit efficiency measures.
Average win = Σ gains ÷ number of winners · Average loss = |Σ losses| ÷ number of losers
Trades closed exactly at break-even enter neither average.
The calculation runs on the net: an average computed before fees overstates gains and understates losses, twice in the same direction.
These averages are very sensitive to extreme values. Look at the median too: if it sits far below the mean, one trade is carrying everything.
The 1.67 RRR looks fine, and yet the real flaw is elsewhere: the stops do not hold. Bringing the average loss back to $200 would lift the RRR to 2.25 without changing a single entry.
Compare your average loss to your planned risk. In this page's example, $270 against $200 planned: the 35% gap does not come from the market, it comes from stops widened along the way or never placed.
It is the most profitable check in the group, because average loss is the only one of the two averages you truly choose.
Then compare your average win to your average MFE, the best moment your trades reached. An average win far below it signals rushed exits or misplaced targets.
The gap is measured in currency in a journal, and it points directly at what needs work: not entries, exits.
The first flaw is reading them in currency across different sizes. Going from one to two contracts doubles both averages without anything having changed in how you trade.
The second is the average pulled by an extreme. A single exceptional winner raises the average win of the whole sample, and the median, much lower, is then the real typical size.
The third is computing before fees. It overstates gains and understates losses, two errors in the same direction, and it flatters the win/loss ratio accordingly.
The classic mistake. Reading both figures in currency on an account whose position size varies. An average win that rises because you doubled your size is not a method improvement, and it mechanically comes with an average loss that rises just as much.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| Average win / average loss | Σ gains ÷ winners · |Σ losses| ÷ losers | The typical size on each side | Checking stops and targets |
| RRR | average win ÷ average loss | How winners weigh against losers | Judging the pair with win rate |
| R-multiple | result ÷ risk taken | The result in units of risk | Comparing despite different sizes |
| Exit efficiency | gain banked ÷ MFE | The share of the move taken | Judging your exits |
Beyond a 20% gap, a handful of trades with no recorded stop almost always explain everything. Finding them in the journal takes one session, and they are worth more than any new entry.
The gap is money left on the table. If it is wide and steady, set your targets on what your trades actually reach rather than on a round number.
If the two diverge sharply, a single trade carries the average. The median then describes your typical trading, and that is what needs improving.
Not necessarily. What matters is the pairing with win rate: an average win below the average loss stays profitable if you win often enough. The threshold is computable, it is 1 ÷ (1 + RRR) applied to your accuracy.
Find the trades involved in your journal rather than changing method. In most cases, a handful of trades with no recorded stop explain the whole gap. It is an execution problem, not a strategy one, and it is fixed in one session.
Do not exclude them, but compute the median alongside. If mean and median diverge sharply, you know the average is describing an accident rather than a behaviour, and that information is the useful one.
By setting targets on what the market actually offers, measured by the MFE of your past trades, rather than on a round number. Many traders systematically exit at 1R while their winners reach 2.4R on average before turning.
On the net, always. Before fees, average win is overstated and average loss understated, and the win/loss ratio comes out flattered twice. On a high-frequency style, the difference is enough to make a losing system look balanced.
In Tradoshi, Average win and Average loss are computed on the net, fees included, and shown in the Reports screen. The Dashboard offers the Average win / loss chart, which shows both side by side with their ratio.
Both checks on this page happen in the same screen: Average % of capital risked is compared with your average loss, and Average MFE and Money left on the table are compared with your average win. The version in units of risk is Average realized R.
To go deeper on this: read the full guide. Every term is gathered in the glossary.