Risk-Reward Ratio vs Win Rate: What the Numbers Actually Show
Ben Ghabili · Published
A strategy can win most trades and still lose money. Another can lose more often than it wins and produce a positive result. The frequency of winning trades needs the size of wins, size of losses and costs beside it.
There is another distinction to preserve: a target drawn on a chart is a planned payoff, not a realised outcome or a probability of success.
What is the difference between risk-reward ratio and win rate?
A planned risk-reward ratio compares an assumed loss with an intended gain for a trade. Win rate measures the fraction of trades classified as winners in a specified record. Realised average wins and losses describe what those trades actually produced. These measures are related, but none can substitute for the others.
State the ratio's orientation. Risk:reward of 1:3 means three units of intended reward for one unit of assumed risk. The equivalent reward divided by risk is 3. Writing “the ratio is 3” without its definition invites confusion.
Also state whether outcomes are before or after costs. A tiny gross winner can become a net loser, so classifications and averages must use a clear basis.
Can a high win rate still lose money?
Yes. A high win rate can be outweighed by small wins, larger losses or costs. Evaluate the weighted result rather than treating the winning percentage as a complete performance measure.
Consider two fictional records of 100 completed trades each. Every trade uses the same fixed initial planned-risk unit: 1R = $100. All outcomes are invented, there are no zero-result trades, and the win/loss classification below is before costs. No compounding or portfolio-capital assumptions are implied.
| Measure | Record A | Record B |
|---|---|---|
| Winning trades | 70 | 40 |
| Losing trades | 30 | 60 |
| Win rate before costs | 70% | 40% |
| Average gross win | +0.4R | +2R |
| Average gross loss magnitude | 1R | 1R |
| Gross total | −2R | +20R |
For A, 70 × 0.4R − 30 × 1R = −2R. For B, 40 × 2R − 60 × 1R = +20R.
Now assume an invented all-in cost of 0.05R per completed trade, covering entry and exit. Each record incurs 100 × 0.05R = 5R.
| Net measure | Record A | Record B |
|---|---|---|
| Net total | −7R | +15R |
| Net average per trade | −0.07R | +0.15R |
| Dollar total with fixed $100 per R | −$700 | +$1,500 |
The lower win rate produced the better result in this constructed comparison. That is not a rule favouring low win rates. It demonstrates why frequency alone is insufficient.
These summaries establish arithmetic for invented inputs, not a real trading edge, drawdown or statistically reliable future return.
How do you combine win rate and average payoff?
For a simplified model containing only wins and losses, the average net payoff is:
Average net payoff = p × W − (1 − p) × L − c.
Here p is the win fraction or assumed win probability, W the average gross win, L the positive magnitude of the average gross loss, and c the assumed cost per trade. Keep the units consistent.
For an observed record, this gives a sample average using its realised outcomes. For a future model, the inputs are assumptions or estimates. Calling the calculation “expectancy” does not turn an observed win fraction into a known future probability.
If your record includes scratch trades, variable costs or different outcome classes, include them in the actual arithmetic. Do not force them into a two-outcome model without adjusting the inputs and definitions.
What win rate is needed to break even?
Under the simplified win/loss model, the break-even win fraction is (L + c) ÷ (W + L). It depends on the realised or assumed payoffs and costs, not just the target written before entry.
Ignoring costs and assuming every winner and loser exactly matches the stated payoff:
| Risk:reward | Assumed loss | Assumed win | Break-even win rate |
|---|---|---|---|
| 1:1 | 1R | 1R | 50% |
| 1:2 | 1R | 2R | Approximately 33.33% |
| 1:3 | 1R | 3R | 25% |
Costs change the boundary. With average wins of 2R, average losses of 1R and a flat cost of 0.05R, break-even is 1.05 ÷ 3 = 35%, rather than approximately 33.33% before costs.
Using the fictional records' payoffs and costs, A would need 1.05 ÷ 1.4 = 75% winners to break even; B would need 35%. Their supplied win rates of 70% and 40% sit on opposite sides of those model boundaries.
This does not establish a safe target win rate. If the inputs change, the threshold changes, and future inputs remain uncertain.
Does an attractive planned ratio prove a good trade?
No. A planned ratio describes payoff geometry under assumed exits. It does not establish the probability of reaching the target, the actual average win or loss, or executable prices.
For a separate fictional long-stock plan, suppose entry is $50, stop is $48 and target is $56, before fees. The assumed downside distance is $2, and the intended upside is $6: risk:reward of 1:3.
Placing the target at $56 does not show how often it will be reached. A stop can execute at a different price, and partial exits or changed rules can make realised payoffs differ from the original plan. The target and stop do not supply an observed win probability.
Changing the target to improve the displayed ratio changes the proposal. It does not independently improve its evidence.
What should you record before judging a strategy?
Record the planned terms separately from the outcomes. A useful review includes the number of eligible trades, outcome definitions, realised average wins and losses, costs and whether the risk unit remained consistent.
Inspect which trades are missing as well as which ones won. A selected set of successful examples cannot establish a strategy-wide win rate. Check the result across relevant periods and preserve failures rather than retelling only the best cases.
The overall average also omits the sequence of outcomes. Two records can share an average while having different loss runs and capital demands. Position sizing and drawdown still require separate review.
For your next performance claim, place win rate, realised payoff sizes and net average on the same line. Keep the planned ratio on a separate line. If only the planned ratio is available, the outcome evidence is still missing.
Nothing here is investment advice.