Why an 80% Win Rate Still Loses: Factor in the Size of Profits and Losses Along with Costs

Author: Thomas | Trading Knowledge | October 6, 2026

Eight out of ten trades win, yet the account still loses money—this is not a contradiction. Win rate only answers how many trades are profitable; it does not answer how large each profit or loss is, nor does it explain fees or unrealized risk. When evaluating gold trading or EA records, these definitions should be aligned first.

An 80% Win Rate Can Still Be a Negative Return

The following numbers are only a calculation demonstration, not the real performance of any account, nor today's gold quote. Assume ten closed trades: eight each profit USD 10, and two each lose USD 50, excluding fees for now. Total profit is USD 80, total loss is USD 100, and the net loss is USD 20; the win rate calculated by number of trades is nevertheless 80%.

If we further assume an additional USD 1 commission per trade, and the above profit and loss does not yet include this commission, the ten trades incur an additional USD 10 in total, resulting in a final net loss of USD 30. A high win rate and a loss can coexist because a few larger losses offset many smaller gains.

Distinguish Actual Averages from the Planned Profit/Loss Ratio

In this example, average profit is USD 10, and the absolute value of average loss is USD 50. Under the definition of 'average profit divided by the absolute value of average loss,' the ratio is 0.2. When using any profit/loss ratio, first specify the numerator and denominator to avoid confusing reward-to-risk with risk-to-reward representations.

The ratio between planned take-profit and stop-loss is also not equal to the average profit-to-loss ratio in actual records. Early exits, scaling out, slippage, changes in size, and execution deviations can all make actual results different. Do not use target distances on a chart to replace the actual fill statistics in the account.

What a Formula for the Average Result Can Show

Under the same statistical definition, the sample's average result per trade can be written as: win rate multiplied by average profit, minus loss rate multiplied by the absolute value of average loss, then minus average fees per trade not yet included. In the demonstration, excluding fees, it is 0.8 times 10 minus 0.2 times 50, which equals a loss of USD 2 per trade; after adding the USD 1 commission per trade, it is a loss of USD 3 per trade.

The averages here come from these ten hypothetical records; they are sample results, not outcomes that will necessarily occur on every future trade. If break-even trades exist, it must be made clear how they are counted in the number of trades and proportions; for samples with different numbers of trades, a win rate calculated by number of trades cannot be treated as a money-weighted return.

Include All Fees, but Do Not Deduct Twice

In MT4 account history, you can check fields such as trade results, commissions, and overnight fees, export for a consistent period, and inspect the records. First clarify which kind of profit and loss the report shows, then add items not yet included. Some fees are already reflected in the fill results; do not deduct them a second time just to make things look more complete.

Spread, commission, and slippage are not the same concept. If profit and loss are calculated based on actual entry and exit fill prices, the spread and some execution impact are usually already reflected in the result; during review, their impact can be analyzed separately, but double-counting should be avoided. Overnight items may be an expense or income and should retain their true sign rather than always being written as a loss.

What Might Be Missed by Counting Only Closed Trades

A high win rate on closed trades does not mean the account's overall risk is low. Unrealized losses on open positions, exposure from continuously adding to positions, deposits and withdrawals, and the observation period can all affect judgment. Multiple screenshots of green orders cannot replace complete records, and one large loss can also change prior conclusions from a small sample.

For an EA in particular, verify the unit of statistics: is it one order, one complete trade idea, or a group of positions? Splitting the same trade into many small profitable orders may raise the win rate calculated by orders without improving the overall result. Before comparing different systems, first unify the unit of statistics and the time period.

Keep Four Records Together During Review

Put win rate, average profit, absolute value of average loss, and fee definition together, then compare them with drawdown, positions, and sample size. Keep original fill records, state whether they are live, demo, or backtest, and avoid changing the statistical interval after results appear or cherry-picking favorable trades.

This information helps explain records, but it does not provide any guarantee of future profit. If actual losses are found to exceed the original limit, the cause should be evaluated according to the established review process; problems must not be concealed by widening the stop-loss, increasing position size, or changing the statistical definition.

Leveraged trading can cause significant losses. This article is for explaining calculation principles and review knowledge; the hypothetical numbers do not constitute trading, position, or return advice, and historical samples cannot represent future performance.