Pretty much every fund fact sheet, strategy tearsheet, and platform leaderboard opens with the same number: average excess return divided by the standard deviation of returns. That is the Sharpe ratio, and it is everywhere for two reasons. It is easy to compute and everyone else uses it. Neither of those is a good reason to lean on it as your main measure of whether a strategy is any good.
What the Sharpe ratio quietly assumes
The first assumption is that returns are normally distributed. They are not. Real returns have fat tails, meaning extreme moves show up far more often than a normal curve predicts, and most strategies carry negative skew, meaning the big losses are bigger and more likely than the big wins. A strategy that grinds out small consistent gains and then eats one enormous loss can post a lovely Sharpe ratio right up until the day it blows up. The metric never saw it coming because the math assumes that kind of event basically doesn't happen.
The second assumption is that standard deviation is the right way to measure risk in the first place. Standard deviation treats upside and downside the same. So a strategy that returns 1% most months and occasionally spikes 10% gets punished with a high standard deviation and a lower Sharpe ratio, even though all that volatility is in your favor. Getting dinged for making too much money in a good month isn't risk in any sense a trader cares about.
Then there is time-period sensitivity. Run the Sharpe ratio over a bull market and it looks one way. Run the same strategy over a bear market or a full cycle and it looks completely different. Most backtested Sharpe numbers get calculated over the friendliest stretch available, which inflates them relative to what you'll actually live through once conditions turn.
Metrics that hold up better
The Sortino ratio is the obvious first upgrade. It swaps standard deviation for downside deviation, so only negative returns count as risk. That lines up with how traders actually think. A strategy with plenty of upside volatility and very little downside volatility will score higher on Sortino than on Sharpe, which is the correct read on it.
The Calmar ratio takes annualized return and divides it by maximum drawdown. This one captures the worst-case experience of holding the thing, which is usually more useful than average volatility. Two strategies can both show a 2.0 Sharpe ratio while one has a 15% max drawdown and the other has a 50% max drawdown. Those are wildly different things to hold through, and the Sharpe ratio alone can't tell them apart. Calmar can.
The Omega ratio goes further and looks at the whole distribution instead of just the mean and variance. You pick a threshold return, then take the probability-weighted gains above it over the probability-weighted losses below it. Because it uses the full shape of the distribution, Omega picks up skew, kurtosis, and every higher moment that mean-variance metrics throw away. It is more work to compute and harder to explain, but it is the most complete single number of the bunch.
The drawdown reality check
Maximum drawdown deserves its own paragraph, because it is the number that describes the worst moment you'll actually feel. And here is the uncomfortable part: in live trading, your realized max drawdown will almost always be worse than the one in your backtest. Backtests are finite samples. Give the strategy enough live time and it will eventually find a drawdown deeper than anything in the historical record, simply because it has more chances to.
The heuristic I use at Blockcircle when we evaluate a signal is to assume the future max drawdown will run 1.5x to 2x the worst one in the backtest. Then I ask a blunt question: if it actually got that bad, would I still be holding? If the answer is no, the strategy is too risky for me no matter how pretty its Sharpe ratio is. That single gut check has killed more attractive-looking strategies for me than any statistical test.
Read them together, not one at a time
No single metric tells the whole story, so we never rank on one. The setup that has held up best is looking at several at once:
- Sharpe ratio for a rough sense of risk-adjusted return
- Sortino ratio to isolate downside risk from the noise
- Calmar ratio for the worst-case view
- Win rate and average win versus average loss for what individual trades look like
- Max drawdown duration for the part nobody models, which is how many months you'll sit underwater before it comes back
The interesting information shows up when these disagree. When they do, the conservative reading is usually the right one. A strategy with a great Sharpe ratio but a poor Calmar ratio almost certainly has tail risk that the Sharpe number is hiding from you. Flip it around and a strategy with a great Calmar ratio but only a so-so Sharpe ratio is often sturdier than it looks, because it clearly survives the ugly periods even if its average-condition performance is nothing special.
None of this means you have to throw the Sharpe ratio out. It is a fine starting point and a common language everyone already speaks. Just don't let it be the last word. Pull up the Sortino, the Calmar, and the drawdown alongside it, sit with the case where the worst month happens twice as hard as your backtest, and decide from there.