Two strategies can post the same annual return and still deserve completely different trust. The Sharpe ratio is how that gets separated into a number — run through the exact formula our calculator uses, on three real trade profiles with the actual arithmetic shown.
Returns look great until you adjust for risk.Negocie no Bybit — grande liquidez, taxas baixas, bônus de boas-vindas de até US$ 30.000.Abra a conta Bybit →Not "how much did this make" — "how much risk did it take to make that much." A strategy that returns 40% a year with huge swings can be objectively worse than one that returns 15% smoothly, because the first one is mostly luck riding volatility and the second is a repeatable edge. The Sharpe ratio turns that comparison into a single number: excess return per unit of volatility.
Sharpe = (annualized return − risk-free rate) ÷ annualized standard deviation
The risk-free rate (roughly 4-5% from short-term T-bills right now) is subtracted because that's the return available for doing absolutely nothing. A Sharpe of 1.0 means 1% of excess return for every 1% of volatility taken on; above 1.5 is generally considered solid, above 2 is rare outside of overfit backtests or short lucky samples.
Our Sharpe & Sortino calculator builds the ratio from four trading inputs: win rate, average win %, average loss %, and trades per year. Here's the full derivation at the calculator's own default inputs — 50% win rate, +3% average win, −2% average loss, 250 trades/year, 4% risk-free rate:
| Etapa | Fórmula | Valor |
|---|---|---|
| Per-trade mean | 0.5×3 − 0.5×2 | 0,50% |
| Per-trade variance | 0.5×(3−0.5)² + 0.5×(−2−0.5)² | 6.25 |
| Per-trade std. dev. | √6.25 | 2.50% |
| Annualized mean | 0.50% × 250 | 125.0% |
| Annualized std. dev. | 2.50% × √250 | 39.53% |
| Proporção de Sharpe | (125.0 − 4.0) ÷ 39.53 | 3.06 |
| Downside dev. (annualized) | √(0.5×2²) × √250 | 22.36% |
| Proporção de Sortino | (125.0 − 4.0) ÷ 22.36 | 5.41 |
Notice Sortino (5.41) comes out far higher than Sharpe (3.06) on the exact same trades. That's not a rounding quirk — it's the whole point of the second metric, covered next.
Now compare a trend-style setup: 35% win rate, but wins average +5% against losses of only −1.5%, running 150 trades/year at the same 4% risk-free rate.
| Métrica | Symmetric (50% WR, 3%/2%) | Asymmetric (35% WR, 5%/1.5%) |
|---|---|---|
| Per-trade expectancy | 0,50% | 0.775% |
| Annualized return | 125.0% | 116.25% |
| Annualized volatility | 39.53% | 37.97% |
| Proporção de Sharpe | 3.06 | 2.96 |
| Annualized downside deviation | 22.36% | 14.81% |
| Proporção de Sortino | 5.41 | 7.58 |
Almost identical Sharpe — the second strategy actually scores slightly lower on Sharpe despite having a much better risk profile on paper. Why? Sharpe's standard deviation treats a big +5% win exactly like it would treat a −5% loss: both count as "volatility" and both inflate the denominator. The asymmetric strategy's large wins are punishing its own Sharpe score. Sortino, which only measures deviation on the downside, correctly rewards it instead — 7.58 versus 5.41. This is the single biggest reason to look at both numbers together: a strategy with lumpy, large wins and small, controlled losses will often look mediocre on Sharpe and excellent on Sortino, and only one of those two readings is telling you something you'd actually want to know before sizing a position with the Critério de Kelly.
Marketing pages love to show Sharpe ratios above 2. Here's a more typical account: 52% win rate, a plain symmetric 2%/2% payoff — a thin but genuine edge — at 300 trades/year:
| Etapa | Valor |
|---|---|
| Per-trade expectancy | 0,08% |
| Annualized return | 24.0% |
| Annualized volatility | 34.61% |
| Proporção de Sharpe | 0.58 |
| Proporção de Sortino | 0.83 |
A 24% annual return sounds appealing in isolation. Run through the exact same formula as the two examples above, it scores a Sharpe of 0.58 — below the "most of your return is noise" line. This is the far more common outcome for a real, unoptimized edge than the 3+ figures in the earlier examples, and it's a useful gut check: if a strategy's backtest reports a Sharpe north of 2.5 on a meaningful sample size, the more likely explanation is overfitting or a short lucky stretch, not a genuinely exceptional system. Cross-check any such number against the calculadora de expectativa comercial and a full Monte Carlo simulation before trusting it.
Every example above subtracted a 4% risk-free rate — the rough yield on short-term T-bills. That choice is uncontroversial for a traditional equity strategy. For crypto trading it's genuinely contested: some analysts use 0% (there's no truly "risk-free" crypto asset), some use the T-bill rate as a nod to opportunity cost, and some use BTC buy-and-hold as the real benchmark being beaten. Swap the benchmark and the same strategy's Sharpe moves — at the 125% annualized return from the first example, using 0% instead of 4% risk-free barely changes anything (3.16 vs 3.06), but at the thin-edge 24%/year example it matters a lot more (0.69 vs 0.58). Always check which benchmark a quoted Sharpe ratio is actually using before comparing it to your own.
Per-trade expectancy = (win rate × avg win) − (loss rate × avg loss). Variance is the win-rate-weighted squared deviation of the win and loss outcomes from that mean, and standard deviation is its square root. Annualize the mean by multiplying by trades per year, and annualize the standard deviation by multiplying by the square root of trades per year. Sharpe = (annualized mean − risk-free rate) ÷ annualized standard deviation.
Sharpe penalizes all volatility, upside and downside, equally. Sortino only penalizes downside deviation below a target. A trend-following setup with big wins and small, consistent losses can have the same Sharpe as a symmetric win/loss system — because the big wins inflate Sharpe's denominator just as much as they'd inflate it if they were losses — while its Sortino is far higher, since the large swings are all on the good side. In a worked example with a 35% win rate, +5%/−1.5% payoff, Sharpe (2.96) was close to a symmetric 50%-win-rate system (3.06), but Sortino was 7.58 versus 5.41.
The widely cited "above 2 is excellent" threshold describes institutional multi-asset portfolios measured over years, not the output of plugging optimistic backtest numbers into a trade-level formula. A thin but genuine edge — 52% win rate, symmetric 2%/2% payoff, 300 trades a year — annualizes to a Sharpe near 0.58, which is a far more typical number for an active discretionary or systematic trader than the 2-3+ figures usually seen in marketing material.
The risk-free rate represents the return available for doing nothing — currently around 4-5% from short-term T-bills. Subtracting it means the Sharpe ratio measures excess return per unit of risk, not just raw return per unit of risk. For crypto specifically the benchmark is genuinely debatable: using 0%, the T-bill rate, or BTC buy-and-hold as the baseline all produce different Sharpe numbers for the same strategy, so it matters which one a comparison is actually using.
⚠️ Apenas educacional – não aconselhamento financeiro. A negociação de alavancagem pode perder todo o seu depósito.