A trading strategy does not need to win more than half its trades to make money. It needs its winning trades to cover its losing trades and trading costs. The break-even point depends on two numbers: the win rate and the payoff ratio, meaning average profit per winning trade divided by average loss per losing trade.
The break-even win rate is 1 ÷ (1 + payoff ratio), provided wins and losses use the same cost basis and there are no flat outcomes. A payoff ratio of 2 requires a 33.33% win rate to break even. A ratio of 0.5 requires 66.67%. This relationship is a practical application of calculating a trading strategy’s expected value.
What Win Rate and Payoff Ratio Measure
Win rate measures frequency: how often a completed trade produces a profit. If 45 of 100 trades are profitable and the other 55 lose money, the observed win rate is 45%.
Payoff ratio measures size. If winning trades average $150 and losing trades average $100, the payoff ratio is 1.5. Use the positive magnitude of the average loss, not a negative denominator.
Payoff ratio = Average winning trade ÷ Average losing trade
These measures belong together. Counting profitable trades without considering their size leaves out half the calculation. The CME Group lesson on mathematical expectation connects winning frequency with average winning and losing amounts rather than treating accuracy alone as a measure of success.
Keep the measurement basis consistent. Compare dollar profits with dollar losses, or returns expressed in the same risk units. Do not divide an average percentage gain by an average dollar loss, or combine profits after fees with losses before fees.
Deriving the Break-Even Win Rate
Let p represent the probability of winning, W the average winning amount, and L the average losing amount. With no flat trades, the probability of losing is 1 − p.
The expected result per trade is:
E = pW − (1 − p)L
At break-even, expected profits and expected losses are equal:
pW = (1 − p)L
Rearranging gives p = L ÷ (W + L). Dividing the numerator and denominator by L produces the payoff ratio version:
Break-even win rate = 1 ÷ (1 + B), where B = W ÷ L.
| Payoff ratio | Average win if average loss is $100 | Break-even win rate |
|---|---|---|
| 0.5 | $50 | 66.67% |
| 0.75 | $75 | 57.14% |
| 1 | $100 | 50.00% |
| 1.5 | $150 | 40.00% |
| 2 | $200 | 33.33% |
| 3 | $300 | 25.00% |
| 4 | $400 | 20.00% |
The table assumes no costs, or that costs are already included in both averages. The percentages are rounded: a ratio of 2 has an exact threshold of one third, not precisely 33.33%.
You can also reverse the calculation. At a given win rate, the required break-even payoff ratio is B = (1 − p) ÷ p. A 40% win rate therefore requires a ratio of 1.5. Profitability requires exceeding the threshold, not merely reaching it.
Why a Higher Win Rate Can Still Lose Money
Consider a hypothetical strategy with 80 winning trades and 20 losing trades. Each winner earns $25, while each loser costs $150. Before trading costs, the result is:
(80 × $25) − (20 × $150) = −$1,000
The 80% win rate looks attractive in isolation. Its payoff ratio is only 0.1667, however, which requires approximately 85.71% winners to break even. Being right is not the same as making money.
Now consider a strategy that wins 40 trades out of 100, earning $200 per winner and losing $100 per loser:
(40 × $200) − (60 × $100) = $2,000
This strategy loses more often than it wins, yet produces a positive result before costs. Its 40% win rate exceeds the 33.33% threshold associated with a payoff ratio of 2.
Neither example establishes that one trading style is preferable. They show why a win rate has little meaning without the corresponding payoff. Raising the win rate by accepting smaller profits or tolerating larger losses can make the result worse.
Use Realized Payoffs, Not Profit Targets
A planned reward relative to risk is not the same as an observed payoff ratio. A trade entered with a $100 planned loss and a $200 profit target has a planned reward of twice the risk. That does not prove average winners will be $200 or average losers $100.
Partial exits, early closures and changes to position size can alter the completed result. Execution also matters. For stock stop orders, the stop price triggers a market order but does not guarantee the execution price; a stop-limit order controls the acceptable price but may remain unfilled. The SEC investor bulletin on stop orders details this distinction.
Calculate the realized ratio from complete trades, including every entry and exit. Treating three partial profit-taking exits as three wins while counting a fully stopped position as one loss inflates the hit rate without changing the money earned.
Nor can you assume a more distant target improves profitability. It may increase the profit on trades that reach it, but the relevant question is what happens to both the win rate and the average completed payoff.
How Trading Costs Move the Break-Even Point
If average wins and losses exclude trading costs, introduce an average cost per completed trade, c:
E = pW − (1 − p)L − c
Solving for zero expectancy gives:
Break-even win rate = (L + c) ÷ (W + L)
This assumes the average cost stays unchanged in the scenario being evaluated. Include entry and exit commissions and other applicable trading expenses. Execution costs require care: spreads and slippage already reflected in actual fill prices must not be deducted again.
Frequent small charges can consume a trading margin. FINRA’s day-trading risk disclosure warns that accumulated commissions can increase losses or substantially reduce earnings even when the charge per trade is small.
Suppose average gross winners are $150, average gross losers are $100, and total costs average $5 per completed trade. The gross break-even win rate is 40%, but the threshold after costs becomes:
($100 + $5) ÷ ($150 + $100) = 42%
At a 45% win rate, expected net profit is (0.45 × $150) − (0.55 × $100) − $5 = $7.50 per trade. At 40%, the strategy loses an expected $5 per trade.
Alternatively, if every trade incurs exactly $5 and no winner changes classification, use net average wins of $145 and net average losses of $105. Their payoff ratio is approximately 1.381, giving the same 42% threshold. Do not subtract the $5 again.
Calculate the Numbers From a Consistent Trade Record
Start with completed positions over a defined period and apply one counting rule throughout. A useful calculation sequence is:
- Combine all entries, partial exits and remaining exits belonging to each trade.
- Calculate the net profit or loss for each completed trade.
- Count winners and losers, then calculate their separate arithmetic averages.
- Divide average profit by average loss and calculate the corresponding break-even win rate.
Flat trades need their own category. If some trades finish at exactly zero after costs, winning and losing frequencies do not sum to one across all trades. Use E = pwW − plL, where the two probabilities represent winners and losers separately. Alternatively, calculate the win rate among nonflat trades for the break-even comparison, while reporting how often flat trades occur.
Position size also affects interpretation. Dollar averages describe the results of the sizing decisions actually used. Risk-normalized outcomes describe profits and losses relative to each trade’s initial planned risk. Both can be useful, but they answer different questions.
Check that the calculation reconciles with total net trading profit. Broader investment and trading performance measurement then puts that result beside capital employed, time and losses along the way.
Allow for Uncertainty in Both Inputs
An observed win rate is an estimate, not a known future probability. Suppose a hypothetical test records 45 winners in 100 independent trades with a stable winning probability. Using the NIST method for Wilson confidence intervals, the approximate 95% interval for that probability runs from 35.6% to 54.8%.
That interval includes the 42% threshold in the cost example. The observed 45% rate alone therefore does not establish a comfortably positive margin. The interval also addresses only winning frequency, not uncertainty in average profit and loss. Dependence between trades or a changing strategy can make the simple binomial model unsuitable.
Stress the payoff assumptions too. Keeping the win rate at 45% and average winner at $150, but increasing average gross loss from $100 to $150, changes expected net profit from $7.50 to a loss of $20 per trade after the same $5 cost.
A backtest needs another check: how many versions were tried before selecting the reported result? Repeatedly searching the same data can select historical noise rather than a repeatable advantage. The research paper The Probability of Backtest Overfitting examines this selection problem in investment strategy testing.
For a practical review, keep a record of rule changes and evaluate the chosen rules on data not used to select them. Treat walk-forward testing and out-of-sample validation as further checks, not certificates of future profitability.
Break-Even Is Not a Position-Sizing Rule
Even a positive expected result says nothing by itself about an acceptable trade size. Two hypothetical strategies can share the same expectancy while producing different outcomes:
A strategy winning 40% of trades at a payoff ratio of 2 earns an expected 0.2 average-loss units per trade. A strategy winning 60% at a ratio of 1 also earns 0.2 units. Their winning frequencies and payoff patterns differ despite that identical average.
Neither calculation describes the largest possible loss, the clustering of losing trades or the exposure from positions open at the same time. A strategy can exceed its break-even threshold and still suffer losses large enough to exhaust the capital allocated to it. That is a separate risk-of-ruin assessment.
Use the break-even point as a diagnostic boundary. Measure realized outcomes after costs, compare the observed win rate with the required rate, and test how that margin changes under less favorable assumptions. The useful question is not whether the strategy wins often. It is whether its wins plausibly cover its losses and expenses, with enough room for estimation error.