Sharpe Ratio — Return
Annualized Sharpe ratio computed on the Return equity curve (low-to-low log-returns).
- Computed from
- Equity curve
- Scope
- Single report
- Range
- Any real number
- Direction
- Higher is better
The Sharpe ratio answers one question: how much return did this account earn for each unit of risk it took on? It divides the average return by how much that return bounced around, so a steady 1%/month scores far higher than a wild ride that happens to end at the same place.
How it's calculated
Sharpe = mean(r) / stdev(r) × √(ann_factor)
- r
- per-period return of the equity curve (log-returns in % mode)
- mean(r)
- average return across all periods
- stdev(r)
- standard deviation of those returns — the "risk"
- ann_factor
- periods per year, derived from the data (~252 for forex, ~365 for crypto)
On the Return and TWR curves the returns are log-returns (percentages compound), so the Sharpe here is size-independent and comparable across accounts. The difference between the two is the cashflow treatment: Return is money-weighted (deposits/withdrawals move it), TWR strips them out — see the curve linked above.
There is no risk-free rate subtracted — the threshold is zero. For short-horizon trading accounts the risk-free drag is negligible, and leaving it out keeps the number comparable across records. The tradeoff: against a cash or benchmark yardstick this overstates Sharpe by roughly the risk-free rate divided by volatility, so in a high-rate environment treat sub-1 readings as a touch generous.
What it tells you
| Value | Reading | Notes |
|---|---|---|
| < 0 | Losing | Negative average return — risk taken for nothing. |
| 0 – 1 | Weak | Returns barely exceed their own noise. |
| 1 – 2 | Solid | A genuine, tradeable edge. |
| 2 – 3 | Excellent | Strong reward per unit of risk. |
| > 3 | Suspicious | Rare on real money — check for overfitting or a too-short record. |
A useful mental picture: the Sharpe ratio is the signal-to-noise of an equity curve. A high number means the upward drift is large compared to the day-to-day jitter around it. One handy property follows from this: Sharpe doesn't change when you scale leverage up or down — both the return and its volatility scale together — which is exactly why it works as a cross-account comparator where raw return doesn't. (The exception is leverage high enough to trigger margin stops or a blow-up; the fat-tail pitfall below covers that.)
Worked example
The series has one observation per trading day (the day's low-to-low return). Say the average daily return is 0.08% with a daily standard deviation of 0.6%, over a forex record where the data-driven annualization factor works out to ≈252. A daily number is tiny, so we scale it up to a full year — that's what annualized means, and the ×√252 stretches one day's reward-to-risk across the ~252 trading days in a year:
Sharpe = 0.08% / 0.6% × √252 = 0.133 × 15.87 ≈ 2.1
A daily reward-to-risk of 0.133 is modest, but scaled up across a full year of trading days it annualizes to an excellent ~2.1.
Pitfalls
- Volatility cuts both ways. A few huge winning days raise your standard deviation and lower your Sharpe — penalizing exactly the outcome you wanted. Sortino fixes this by measuring only downside deviation.
- Not comparable across trade frequencies. Sharpe here is the daily return-to-risk scaled up by √252. A fast strategy that trades many times a day annualizes to a much bigger number than a slow one with the same per-trade skill — so a high-frequency Sharpe and a swing-trading Sharpe aren't the same yardstick. Same √N caveat as SQN.
- Trending positions inflate it. The math assumes each day's return is independent of the last. Strategies that hold or pyramid into a trend produce smooth, autocorrelated returns that shrink the measured volatility and inflate Sharpe — the classic way a too-good Sharpe lies. Read it next to max drawdown, which serial correlation can't hide.
- Short records inflate it. A handful of lucky weeks can produce a Sharpe above 3 that won't survive. Always read it next to the track-record length.
- Fat tails hide here. Sharpe assumes returns are roughly normal. A strategy that sells options or martingales can show a great Sharpe right up until the blow-up — check kurtosis and max drawdown alongside it.
- Below zero it inverts. When the average return is negative, more volatility makes Sharpe look less bad — so don't rank losing accounts by Sharpe, the math runs backwards there.
A Sharpe above 3 on a record shorter than a year is a yellow flag, not a gold star. On real money, sustained Sharpe ratios above 2 are uncommon.
Sharpe vs Sortino vs Calmar
All three are "return per unit of risk" — they differ in how they define risk:
| Metric | Risk measure | Best when you care about… |
|---|---|---|
| Sharpe | total volatility (both sides) | overall smoothness of the curve |
| Sortino | downside deviation only | not penalizing upside spikes |
| Calmar | maximum drawdown | surviving the single worst stretch |