The Sharpe ratio relates the mean of your results to their irregularity, measured by standard deviation. It does not say how much you earn, it says how steadily you earn it, and that is a question neither P&L nor profit factor asks.
The mean, and the spread around it
Contents
The Sharpe ratio comes from portfolio management, where it serves to compare funds. The idea is that a return cannot be judged alone: making 12% a year calmly and making 12% a year through brutal falls are not the same performance, even if the final figure is identical.
It is computed by dividing the mean of your results by their standard deviation, that is by the measure of their spread around that mean. The more your results resemble each other from one period to the next, the smaller the standard deviation, and the higher the ratio. At equal gain, steadiness pays.
In trading journals it is almost always computed on daily results and left non-annualised, which is the FTMO and MetaStats convention. The academic version additionally subtracts a risk-free rate and multiplies by the square root of the number of periods; both versions carry the same name and do not give the same number, which makes comparing a Sharpe ratio across tools impossible without checking the definition.
Sharpe ratio = mean of the results ÷ standard deviation of the results
The reference period has to be the same for both terms: daily results give a daily Sharpe, monthly results a monthly one. These are not the same numbers.
The annualised version multiplies the result by the square root of the number of periods in a year, about 15.9 for trading days. That is why the same account shows 0.12 here and 1.9 elsewhere.
The academic version subtracts a risk-free rate from the numerator. Trading journals do not, and the difference becomes significant when rates rise.
| Annualised Sharpe ratio | What it describes |
|---|---|
| below 0 | The mean result is negative. The rest of the calculation has no object. |
| 0 to 1 | Irregular results. Most discretionary traders sit here. |
| 1 to 2 | Solid steadiness. This is already the domain of well-run systems. |
| 2 to 3 | Very steady, rare, and generally obtained over many short trades. |
| above 3 | Almost always too short a track record, or a calculation that is not the one you think. |
Before reading this table, you need to know which version of the calculation is displayed. The thresholds above apply to an annualised Sharpe, the one from financial literature. A trading journal showing a non-annualised daily Sharpe will give 0.09 where the same account is worth 1.4 annualised, and comparing the two makes no sense. That is the first thing to check, before even looking at the value.
The second point is that this ratio penalises volatility in both directions. A very good month moves your results away from their mean exactly as a bad one does, so it lowers the figure. That is logical for a fund manager selling predictability, and debatable for a trader whose job is precisely to catch exceptional moves.
The ratio is 0.56 over those twelve months. The point to see is this: the +4.0% month lowers the ratio, because it deviates from the mean as much as a bad month does. The Sharpe ratio does not tell pleasant surprises from unpleasant ones, and that is its best-known flaw.
This is the use the ratio was invented for, and it remains the right one. When two systems return the same over the year, the one with the higher Sharpe got there more calmly, so it is easier to hold and easier to fund.
That comparison only makes sense if both figures come from the same definition and the same reference period. A Sharpe read on a prop firm report and one read in a journal almost never compare directly.
A mediocre Sharpe despite a good return signals that the result rests on few periods. It is the same information as the check consisting of removing your biggest winner from profit factor, obtained another way, and it is worth seeing before drawing conclusions.
The defensive use is the most useful for a retail trader: the ratio serves less to congratulate yourself than to spot that a performance rests on too little.
The most discussed flaw is the symmetry of volatility. Standard deviation treats a +4% month and a −4% month exactly alike, because it only measures a distance from the mean. A trader who catches a big move therefore sees their ratio fall for having done well, and it is to answer that that other ratios exist which count only downside deviations.
The second is confusion between versions. Daily or annualised, with or without a risk-free rate, computed on trades or on days: the same account can show 0.09, 0.56 or 1.4 depending on the convention. Almost no tool writes down which one it applies, and comparing two figures without knowing compares nothing at all.
The third is the sample. Under sixty trading days or so, standard deviation is itself very unstable, so the ratio is too. A Sharpe computed over six weeks halves with two unusual days, which makes it a figure not to look at rather than one to interpret cautiously.
The classic mistake. Comparing one Sharpe ratio with another without checking the convention. The same account shows 0.09 daily non-annualised and 1.4 annualised, and almost no tool writes down which of the two it displays. The more flattering figure is often simply the one multiplied by the square root of the number of days.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| Sharpe ratio | mean ÷ standard deviation of results | Steadiness, in both directions | Comparing two methods at equal return |
| Maximum drawdown | (peak − trough) ÷ peak | The worst decline lived through | Sizing positions |
| Recovery factor | net P&L ÷ maximum drawdown | Gain per unit of decline endured | Comparing two systems at equal risk |
| Profit factor | Σ gains ÷ |Σ losses| | What you earn per $1 lost | Judging a whole system, or one setup |
Daily or annualised, with or without a risk-free rate: until you know, the figure compares to nothing and published thresholds do not apply. It is the one step to take before all the others, and it takes as long as reading a footnote.
The ratio rises when spread falls. Taking smaller positions on the setups with the most scattered results tightens the curve without removing anything from the method, and it is the most direct lever on this particular figure.
Below that order of magnitude, standard deviation moves so much that the ratio does not inform: it adds noise. Many traders adjust their method on a Sharpe computed over six weeks, that is on a number that did not yet mean anything.
It is the mean of your results divided by their standard deviation. It measures the steadiness of a performance rather than its size: at equal gain, the system whose results resemble each other most from one period to the next gets the higher ratio.
On an annualised basis, between 1 and 2 describes solid steadiness, and above 3 you should suspect too short a track record. Careful: those thresholds only hold for the annualised version. A journal showing a non-annualised daily Sharpe will give values fifteen times smaller for exactly the same performance.
Take the series of your results over a fixed period, daily or monthly, compute their mean, compute their standard deviation, and divide the first by the second. To annualise, multiply by the square root of the number of periods in a year, about 15.9 for trading days.
Because standard deviation measures a distance from the mean without caring about direction. A +4% month deviates from your mean as much as a −4% one, so it raises the spread and lowers the ratio. It is the most common criticism of this measure, and it is well founded for a trader whose job is to catch exceptional moves.
Sixty trading days or so is a minimum, and that is still few. Below that, standard deviation is itself very unstable: two unusual days can move the ratio by half. On a short history it is better not to look at it than to look at it cautiously.
Tradoshi computes the Sharpe ratio on your daily results and does not annualise it, which is the convention adopted by FTMO and MetaStats. That is important to know before comparing: the usual thresholds from financial literature, the ones talking about 1 or 2, describe an annualised Sharpe and therefore do not apply directly to the figure displayed.
It appears in the Reports screen, next to maximum drawdown and recovery factor, and the neighbourhood is the point: Sharpe describes spread, drawdown describes depth, and reading one without the other always leaves half the story out.
To go deeper on this: read the full guide. Every term is gathered in the glossary.