Financial Instruments Business Operator Kanto Finance Bureau Director (Kinsho) No. 1960 / Membership Association Investment Trusts Association, Japan Membership No. 012-02323 Go to GogoJungle Home
icon List
SEMURAラボ
09/26/2026 19:36
Release： 09/26/2026 19:36
Update： 09/26/2026 19:36
SEMURAラボ
Investment style : Day Trade
Investment career : More than 10 years
Investment amount : 10 million - 50 million JPY
Hello, Semura Lab. We are diligently pursuing logic development in our daily work. ■ Lab Policy "Build long-lasting, robust systems that reflect the realities of the market." This is the most important guiding principle for our lab. Even if backtests show clean, upward-trending performance, in real markets you can be exposed to slippage. Such stories are not rare. Therefore, we design with the assumption of real-world frictions like spreads and slippage from the outset. ■ What we are particular about - We value "why it works" more than appearance of performance - We avoid over-optimization on past data and emphasize reproducible grounds - We strive for rigorous verification that does not incorporate future information - We do not skip durability checks to see if it holds in unknown markets - Not only the reasons it moves, but also honestly communicate when market conditions are not favorable ■ Why we run this lab I myself have been disappointed many times by products with exaggerated claims. "Then we should deliver the facts as they are." That is the origin of continuing logic development and verification. We may not say flashy things, but we will not compromise on facing the facts. ■ What we provide We are developing tools and logic centered on MT4 that aid operational decisions and understanding of the environment. We aim to be tools that confront the market with solid grounds, not just a vague sense of something. ■ To everyone who purchases It’s not the end at purchase; we want to accompany you in ongoing operation as well. If you have questions or notice anything, please feel free to contact us. We would be glad to be of some help.
Article cover image

[Development Log] Why Look at Expectation Value Rather Than Winning Percentage — How Should We Read Backtest Metrics

When looking at backtest results for FX automated trading (EA), many people first focus on the “win rate.” If the win rate is high, it’s easy to feel that the logic is good, but in fact this intuition can be a major trap that leads you to select the wrong logic. This time, we will explain how to correctly relate to the win rate indicator and how to interpret backtest metrics.


■ Why there are logics with high win rate but no asset growth

When you look at a backtest report, most people’s eyes first go to the win rate. Hearing “80% win rate” naturally gives a sense of reassurance. However, a high win rate does not necessarily mean the logic is good.

For example, suppose there is a logic with an 80% win rate where profits are small when you win and losses are large when you lose. In this case, if you win small 8 times out of 10 and lose big 2 times, the total assets can decrease. The win rate number only reflects the proportion of winning trades and does not reflect the size of gains or losses per trade at all.

Conversely, even with a win rate around 30%, a “small loss, big win” logic that sizes profits well when you win and strictly limits losses when you lose can lead to a rising overall asset, which is not uncommon. If you only look at win rate, such logics may appear “weak,” and you might miss actually excellent logic.


■ The concept of expectancy

To avoid these traps, the important concept is “expectancy.” In simple terms, expectancy indicates how much profit or loss you can expect on average per trade.


Expectancy can be expressed by the following formula.


Expectancy = (Win rate × Average win) − (Loss rate × Average loss)


As this formula shows, expectancy considers not only win rate but also the size of gains when winning and the size of losses when losing, as a package. A low win rate with large gains on wins can yield positive expectancy, while a high win rate with large losses can yield negative expectancy. To assess the true advantage of a logic, you must judge it by expectancy rather than win rate alone.


■ How to interpret backtest metrics

So, when examining backtest results, where should you focus specifically? First, the big premise is that win rate and average win/average loss should be evaluated together. It’s important to consider the balance of profit and loss, i.e., how much you gain when you win and how much you lose when you lose, rather than evaluating win rate alone.

Also, the Profit Factor (Total profit ÷ Total loss) is another metric to check alongside expectancy. If the Profit Factor is above 1, you at least have a total profit, but relying on this number alone is risky. Sometimes a favorable number arises by chance with a small number of trades, so you should also verify the number of trades and the length of the testing period. As a rule of thumb, expectancy and Profit Factor calculated from only a few dozen trades can be heavily affected by statistical noise and should be treated cautiously.

Furthermore, you should not overlook maximum drawdown. Even with positive expectancy, if the drawdown during the run is too large, you may not be able to endure it psychologically in live trading and abandon the strategy. By looking at expectancy, Profit Factor, and drawdown together, you can see the strategy’s capability more three-dimensionally. Additionally, checking the ratio of average win to average loss (risk-reward ratio) can further reveal the character of the logic (gradual vs. all-or-nothing).


■ How to relate to the win rate indicator

The win rate itself is not a meaningless metric. From the perspective of psychological stability, a logic with a high win rate can be easier to operate in live trading. When you have a string of losses, you might want to stop the logic emotionally, even if expectancy is positive, which is not uncommon.

However, that is merely a matter of “ease of continued operation” and not a criterion for judging the logic’s superiority. In actual logic selection, a practical approach is not “high win rate so safe” but “positive expectancy and drawdown within a personally tolerable range,” looked at along two axes.


■ Summary: look at what lies behind the numbers

When looking at backtest reports, if you are pulled by just one number like the win rate, you may miss the essential advantage. Look at win rate, average win/average loss, expectancy, Profit Factor, and drawdown as a set to accurately evaluate a logic’s capability.

Do not be swayed by flashy win-rate numbers; carefully interpret the profit-loss balance behind them. That is the perspective we should maintain.



Also, discussions on how to read backtest metrics and points of interest from actual testing are being shared at the Semura_Lab Yorozu Consultation Desk. If you’re interested, feel free to join.
https://www.gogojungle.co.jp/mypage/community/485