Seven out of ten trades end in profit, yet a loss of EUR 250 is sitting in the account at the end of the day. This may sound contradictory at first, but it is mathematically quite possible. The win rate only says how often a strategy wins. It does not reveal how large the individual profits and losses actually are. A trader can be right frequently and still lose money if the few losing trades are significantly larger than the many small gains.
A reliable evaluation therefore needs more than a positive win rate. It needs the average win, the average loss and the cost of every trade – as well as comprehensive statistics from the trading journal at all times.
Trading win rate is only part of the calculation
The win rate is calculated as follows:
Win rate = winning trades ÷ number of all trades
With 40 winning trades and 60 losing trades, the trader has achieved a win rate of 40%, which by itself says nothing at all about the actual result. Expectancy describes the average amount a trading idea generates per trade when it is repeated often enough. The central calculation is:
Expectancy = win rate × average win – loss rate × average loss – costs
Expectancy is not a forecast for the next trade either, because an individual loss can occur even with a statistically positive strategy. The unit “R” is suitable for the calculation. One R corresponds to the amount that is put at maximum risk in a trade. With a risk of EUR 50, a losing trade of 1R therefore equals EUR 50. A gain of 2R equals EUR 100. This makes different account sizes and position sizes comparable.
A high win rate can be negative
The result can therefore still be negative despite a high win rate and thus a high expectancy. One example:
- 70% winning trades
- average win: 0.5R
- 30% losing trades
- average loss: 2R
0.70 × 0.5R – 0.30 × 2R = 0.35R – 0.60R = –0.25R
The strategy wins seven out of ten trades. Nevertheless, before costs it loses an average of 0.25R per trade. The reason lies in the distribution of the gains. Each of the seven winners brings only 0.5R. Each of the three losing trades costs 2R. Two large losses therefore completely wipe out the gains from several successful trades. A high win rate can be psychologically comforting. It often creates the feeling that the trading idea is working. The account balance, however, does not look at the number of positive trades, but at their exact monetary value.
The opposite is possible as well:
- 40% winning trades
- average win: 2R
- 60% losing trades
- average loss: 1R
0.40 × 2R – 0.60 × 1R = 0.80R – 0.60R = +0.20R
This strategy wins fewer than half of its trades, but still has a positive expectancy before costs. This is not proof that a 2:1 risk-reward ratio is fundamentally better. The figures must fit the actual strategy, the market and the costs. The calculation merely shows why win rate alone is not a quality criterion.
The risk-reward ratio does not describe later success
The risk-reward ratio compares the possible gain with the planned loss. A trade with a 2R profit target and 1R risk has a planned ratio of 2:1. This does not mean that the profit has to occur. It merely describes the original trading plan. In practice, the results often deviate from the plan:
- the stop-loss is reached before the target
- a profit is taken early
- the position is closed because of a news event
- slippage worsens the exit
- part of the position is handled differently
- fees reduce the actual result
A planned profit of 2R can therefore become a realised profit of 0.8R, while a planned loss of 1R can become more than 1R because of a rapid price movement. The ratio should therefore not be calculated only before entry. After a sufficient number of trades, the ratio actually achieved also belongs in the evaluation. A trading journal should record at least:
- planned risk in R
- actual result in R
- fees and other costs
- reason for entry
- reason for exit
- deviations from the trading plan
- market phase and time of day
Costs turn a small edge into a loss
A strategy can look slightly positive before costs and become negative after costs. Assume that expectancy before costs is 0.05R per trade. With a risk of EUR 50, that corresponds to EUR 2.50. If fees, spread and slippage together cost an average of EUR 3 per trade, a loss of EUR 0.50 remains. The edge was therefore not large enough to cover the trading costs.
With only a few trades, this difference is hardly noticeable. With 100 trades, an average disadvantage of EUR 0.50 adds up to EUR 50. The amount increases accordingly with larger positions and higher costs. This explains why backtests often show better results than real accounts. Historical price data does not always contain the actual spreads, execution prices and delays. Strategies with small price targets and a high trading frequency are particularly affected. The smaller the planned profit per trade, the larger the share of costs in the result.
A sample of 20 trades proves little
After a short winning streak, it is easy to get the impression that a strategy works. Twenty trades, however, may be too few to prove this statistically. In a small sample, chance and individual outliers have a particularly strong effect on the statistics. An unusually large gain can increase the average, while a single mistake can shift it in the other direction. The market phase in which the sample is created also plays a role. A strategy that works in a calm uptrend may be less reliable in a sideways phase or during rapid news-driven movements. It is therefore not enough to look only at the overall results.
Useful evaluations can separate trades, for example, according to:
- setup
- time of day
- market phase
- long and short direction
- holding period
- planned and actual exit
- rule-compliant and faulty trades
An average gain of 1R can result from a stable series. It may also have been distorted by two exceptionally good trades. The median, meaning the middle value in a sorted series, can therefore provide additional information. The longest losing streak also belongs in such a statistic. A strategy with positive expectancy can produce several consecutive losses. Anyone who does not know this possibility may increase the risk after three red trades and thereby damage the long-term statistics precisely because of that.
Selling winners too early, holding losses too long
Many evaluations fail above all because of deviations from one’s own plan. A winning trade is closed at 0.4R because the price pulls back briefly. A losing trade remains open because a recovery is expected. This shrinks the average wins while the average losses grow. The win rate may even rise as a result. Taking profits early leads to more small winners, but the account still suffers in the end if the few losing trades are significantly larger.
A trader must therefore record not only whether a trade was won or lost. Much more important is whether the position was closed in the way the original trading idea intended. Only a rule-compliant loss is statistically healthy; an unplanned large loss is more dangerous, even if the next trade happens to offset it.
Expectancy does not replace risk limits
Positive expectancy does not protect against longer losing streaks. Even a functioning strategy can lose several times in a row. Position size must therefore always be chosen so that such a streak does not throw the account off balance. Anyone risking 1R per trade must know how ten or 15 losing trades affect their capital and their own state of mind. Anyone who increases position size after every loss changes their own statistics and significantly increases their financial risk. Expectancy evaluates the trading idea. Position size determines how strongly that idea affects the account. Both calculations belong together. A poor expectancy remains poor even if the risk is reduced. A good expectancy can also become unusable if positions are chosen too large.
Separating these quantities makes it easier to recognise whether a strategy actually has an edge or merely produces frequent small gains.
