Where does the 1:3 rule come from?
The rule comes from a correct observation applied as if it were a law: with a 1:3 ratio you only need to be right 25% of the time to break even, which sounds like an enormous margin of safety.
The arithmetic is right. The mistake is treating win rate as independent of the target — as though you could move the take-profit further away and keep winning as often. Price does not cooperate with that.
Every unit you move the target further from entry reduces the probability of reaching it before the stop. The break-even win rate falls and the actual win rate falls with it, and which falls faster is an empirical question about your instrument, timeframe and setup.
Sometimes the answer favours the wider target. Sometimes it does not. That is the whole finding, and it is instrument-specific rather than universal.
How do you test a reward-to-risk ratio properly?
Hold the entry and the stop completely fixed, vary only the target, and compare expectancy net of costs across the range — because changing two things at once tells you nothing about either.
Concretely: take your setup's historical occurrences, fix the stop at whatever your rule says, and then measure the outcome at targets of 1R, 1.5R, 2R, 3R and 4R on the same trades. Each target gives a win rate and an expectancy, and the curve between them is the answer.
The shape of that curve is the interesting part. It is usually not monotonic — expectancy typically rises to some point and then falls as the target moves past where the instrument actually travels — and the peak is rarely at exactly 1:3.
Watch for the sample-size illusion while doing this. Wider targets produce fewer winners, so the estimate for a 5R target rests on far fewer positive observations than the one for 1R, and its apparent superiority may be noise.
Does the ratio interact with the market you trade?
Strongly, and this is why a single universal ratio was never going to hold: the achievable reward-to-risk depends on how far the instrument tends to travel relative to its noise.
A trending instrument gives room for wide targets, and a strategy that exits at 1R in a market that regularly runs 5R is leaving most of the edge behind. A range-bound instrument punishes the same target, because price turns around before getting there.
Timeframe interacts the same way. Costs are roughly fixed per trade, so on a short timeframe a small target is competing with a cost that does not shrink — which pushes the optimal ratio wider for reasons that have nothing to do with the trading rule.
The generalisable advice is therefore not a number. It is that the ratio is a parameter to be measured per instrument and timeframe, and that measuring it takes one afternoon.
Is a fixed target the right approach at all?
A fixed target is the easiest exit to test and rarely the best one, because it discards information that arrives after entry — but the alternatives are much easier to fit.
Trailing stops, time-based exits and structure-based exits all adapt to what price does. In a favourable sample they usually beat a fixed target, and they add parameters: how far the trail sits, when it activates, what counts as structure. Each of those is a trial.
The honest comparison holds everything else constant and tests the exits against each other on the same entry set, out of sample, with the number of variants recorded. Anything else compares an exit that had five attempts at fitting against one that had none.
A defensible starting point is to establish the fixed-target baseline first, then see which adaptive exit beats it out of sample. If none does, the baseline was the answer — and knowing that is worth more than an unvalidated trail.
The same discipline applies to partial exits, which are usually adopted for psychological reasons and rarely tested as the parameter change they are. Taking half off at 1R changes the expectancy, the variance and the win rate all at once, and whether the trade is worth making is measurable on the same trade set that produced the baseline.