Win rate vs risk reward ratio: the real trade-off

You have read that reward-to-risk matters more than win rate. You have also read that a high win rate is what keeps you executing. Both claims are half-right, and neither tells you what to do with the two numbers already sitting in your own journal.

One equation connects them. Once you have it, the question stops being which number matters more and becomes which combination of the two you can actually execute.

The one equation behind both numbers#

Your reward-to-risk ratio is what a winner pays relative to what a loser costs. Risk 1R, target 2R, and you are running a 2:1 system. Your win rate is the share of trades that reach the target instead of the stop.

Breakeven is the point where the two cancel out:

Breakeven win rate = 1 / (R + 1)

R is the reward-to-risk ratio. At 2:1 that is 1 / 3 = 33.3%. Win a third of your trades at 2:1 and you end flat, before costs.

The right-hand column is the same equation solved for a target expectancy instead of zero: 1.2 / (R + 1). It is there because breakeven is not the goal — you need a margin above it wide enough to cover costs and the error in your own estimate.

Two systems, identical expectancy#

Take two rows from that table and build a system from each.

System A wins 60% of the time at 1:1.

(0.60 x 1R) - (0.40 x 1R) = 0.60R - 0.40R = +0.20R per trade

System B wins 30% of the time at 3:1.

(0.30 x 3R) - (0.70 x 1R) = 0.90R - 0.70R = +0.20R per trade

Identical. Over 200 trades both expect +40R. If 1R is 1% of a $25,000 account — $250 — both expect roughly $10,000 before compounding.

On a spreadsheet they are the same system. In an account they are nothing alike.

Why identical expectancy does not feel identical#

What drives the experience is the loss rate, not the expectancy. System A loses 40% of the time. System B loses 70% of the time.

The probability that the next k trades are all losses is the loss rate raised to the power of k:

Read the eight-loss row. In System B, measured from any point you care to pick, an eight-trade losing run is roughly a one-in-seventeen event — and across a year of trading you will pass through hundreds of such points. In System A the same run is about one in fifteen hundred.

The gap shows up again in how long it takes to know whether either system works at all, because per-trade variance is what sets the sample size you need:

System A: E[X²] = 0.60(1²) + 0.40(1²) = 1.00    variance = 1.00 - 0.20² = 0.96
System B: E[X²] = 0.30(3²) + 0.70(1²) = 3.40    variance = 3.40 - 0.20² = 3.36

3.36 / 0.96 = 3.5. System B needs about three and a half times as many trades to reach the same confidence that its edge is real. Same expectancy, three and a half times the evidence required. The arithmetic behind that sample size is worked through in how many trades before you know a strategy works.

If you would rather see the two distributions than take the algebra on faith, put each pair of numbers into the Monte Carlo simulator and compare the spread of outcomes rather than the averages. The averages will agree. The tails will not.

My main setup is SMC. Across 319 trades it wins 25% of the time, the average winner is 3.6R, and the loser is 1R on nearly every stop. That is System B, and slightly past it: nearer 25% at 3.6:1 than 30% at 3:1, with the same expectancy, about +0.18R a trade. The written target averages 4.8R, so the low win rate is what that exit rule produces; 3.6R is where the winners actually closed.

The number that breaks the table#

Everything above runs on your planned reward-to-risk. Your account runs on your realized one.

Say you plan every trade at 1R stop and 2R target. Breakeven is 33.3%, you hit 40%, so you expect +0.20R per trade. Now price the trades as they actually closed.

You took partials into resistance, so the average winner finished at 1.3R rather than 2R. A couple of stops gapped and you widened one, so the average loser cost 1.1R rather than 1.0R.

Realized reward-to-risk = 1.3 / 1.1 = 1.18
Breakeven win rate      = 1.1 / (1.3 + 1.1) = 45.8%

That second line is the same equation written without rounding: 1 / (R + 1) is identical to average loser / (average winner + average loser), which is the form worth remembering because it takes your two journal averages directly.

Your 40% win rate is now below breakeven. Run the expectancy on the real numbers:

(0.40 x 1.3R) - (0.60 x 1.1R) = 0.52R - 0.66R = -0.14R per trade

Same setups, same win rate, same stated plan. The system moved from +0.20R to −0.14R per trade because the average winner came in 0.7R short of its target. Across 100 trades at $250 per R, that is the difference between +$5,000 and −$3,500.

No breakeven table can show you this, because the table takes your intentions as its input. The realized number only exists if you record the planned stop, the planned target and the actual exit on every trade — and then divide one average by the other.

The 3.6R above is my full history. Across my last 100 SMC trades the average winner is 4.2R and the average loser is 1.0R, against a written target of 4.4R on the winners. Realized reward-to-risk is 4.2:1, just under what I thought I was trading. Twenty-four of the twenty-five winners closed at the target. One closed at 0.6R instead of 4.5R, and that trade is the entire gap. Over my last 50 every winner reached its target, so the realized ratio, 4.6:1, came out level with what I had written those trades for.

Choosing the profile you can execute#

Neither profile is better. They fail differently, and the useful question is which failure you can sit through.

Lean toward the high-win-rate profile if what has broken you before is a long run of losses: you abandon systems after four reds, or size up to win it back. A 1:1 system at 60% hands you that run far less often.

Lean toward the high-reward profile if what has broken you before is cutting winners early. A 3:1 system is mostly small losses punctuated by the occasional large win, and it tolerates a low win rate — but it demands that you are still taking the setup after ten losses in a row.

If you are inside a prop evaluation the constraint is usually a drawdown limit rather than a return target, and that pushes against the high-variance profile. The ten-loss run System B produces 2.8% of the time costs 10R; at 1% risk per trade that is 10% of the account. Check what your own limit permits before you commit to a profile — the considerations are laid out for prop firm evaluations.

None of this arithmetic survives if 1R moves around. Risk 0.5% on one trade and 2% on the next and your R-multiples stop being comparable, which leaves your win rate attached to nothing. Fix risk per trade first; the position sizing calculator turns a balance, a risk percentage and a stop distance into the size.

The setup I stopped taking is FCS. Across 143 trades it won about 29% of the time at 2:1, and the winners did close at the target. I sat through ten losses in a row in April 2026 and kept taking it. From 30 April to 13 May the next thirteen trades were twelve losses and one win, about −10R, and the last six were all losses. I have not taken it since. SMC I kept trading.

Where this arithmetic stops#

Four things the equation does not give you.

Your win rate is an estimate, not a constant. Forty wins in 100 trades is consistent with a true rate across a fairly wide band either side of 40%. Every comparison above inherits that uncertainty, which is why a system sitting a point or two above breakeven is not meaningfully distinguishable from one sitting a point or two below.

Costs are missing from the table. Spread, commission and financing come out of every winner and get added to every loser. A 1:1 system at exactly 50% is a losing system once they are counted, and the shorter your average hold, the larger the share of your edge they take.

Expectancy assumes trades are independent. If your losses cluster into one regime, one session or one setup that stopped working, the streak probabilities above understate what you will live through. How long a losing streak is normal takes that apart separately.

You do not choose a win rate directly. You choose a setup, a market and an exit rule, and those produce a win rate. Treating this as a menu you pick from is the main way the exercise gets misused — it is a measurement tool for the pair you already have.

What to do with this#

Pull your last 100 closed trades and compute three numbers: your win rate, your average winner in R, and your average loser in R.

Divide the average winner by the average loser. That is your realized reward-to-risk. Put it through 1 / (R + 1) and you have the win rate the system requires. Compare it against the win rate you have.

A positive gap means an edge worth sizing properly. A negative gap means you now know which of the two numbers to work on — and you know it from your own record, rather than from an argument about which one matters more.