We measure. We do not promise.
Most people publishing an indicator announce a high win rate without saying how they calculated it, over how many trades, or what they count as a win. We measured the indicator with an internal research build, and we publish what it produced.
First: what R means
R is the distance between the entry price and the stop-loss. From a real signal:
- Hitting the stop = −1R
- Making 14.8 points = +1R
- Making 29.6 points = +2R
R is not a money amount. It is a relative unit that each trader translates into their own position size, which is what makes trades of different sizes comparable.
Second: why we do not lead with a win rate
Because it says nothing about profitability. Consider two systems:
| System A | System B | |
|---|---|---|
| Win rate | 80% | 25% |
| Gain when right | +0.5R | +5R |
| Loss when wrong | −1R | −1R |
| Expectancy | +0.2R | +0.5R |
System B is more profitable with a third of the win rate. The number that matters is average R per trade.
Third: the results
Gold · 15-minute chart · 782 closed signals. The management assumed in the measurement: a third of the position closes at the first target, a third at the second, and the remainder at the third target or the stop.
| Value | |
|---|---|
| Profitable trades | 489 |
| Losing trades | 293 |
| Breakeven trades | 0 |
| Best trade | +7.8R |
| Worst trade | −1R |
Fourth: the win rate depends on the management
62.5% is not a property of the indicator. It is a consequence of scaling out. Hold the full position for the third target alone and the win rate drops to 24%, while the total rises from 42R to 60R.
| Management | Win rate | Total | Best trade |
|---|---|---|---|
| Scale out in thirds | 62.5% | +42.3R | +7.8R |
| Full position to TP3 | 24.3% | +60.2R | +12.8R |
The higher win rate produces the lower return, because closing early truncates the rare large winners the system lives on. Choose what suits your temperament — but know the price of the choice.
Fifth: what we do not know
- Costs are not included. Spread, slippage and commission. At an average of 0.054R per trade these are not a footnote — they can consume a meaningful share of the total.
- Drawdown is unmeasured. The figures show the system is positive, but not how far the account fell from its peak. A system that is positive on paper can be impossible to hold in practice.
- One instrument, one period. Gold in a declining phase. It has not been tested in a strong uptrend or an extended sideways market.
- Dependence on rare trades. The single best trade is a meaningful share of the total. Miss it and the picture changes.
- 782 signals over a few months is roughly ten a day, and nobody takes all of them. Which ones you take changes your result entirely.
Sixth: we will not project returns for you
Anyone can multiply 42.3R by a risk size and arrive at an impressive figure. We will not, because the figure would be fiction: nobody takes every signal, costs are excluded, and the sample is from one market.
What we can offer is the tool itself, the numbers as they are, and their limits stated plainly. Then you decide.
The measurement ran on an internal research build, not the published script. It is the published engine at its default settings, so it fires the same signals, plus a tally that records each signal's result in R, shows the sample size, and warns explicitly when the count is too small to support a conclusion (under 200 trades). The full code is on the measurement code page for anyone who wants to reproduce the measurement. Nothing was hidden.