By Keel Research Team · Updated May 13, 2026
Sortino and Sharpe ratios both measure risk-adjusted return as (excess return / risk). They differ in how risk is measured: Sharpe uses standard deviation of all returns; Sortino uses standard deviation of only the losing periods. The choice matters when your strategy has asymmetric upside.
Both ratios answer the same question: "How much return did I get per unit of risk I took?" The numerator is the same — excess return over the risk-free rate. The difference is entirely in the denominator: how we define "risk."
Sharpe defines risk as standard deviation of all returns — upside spikes and downside losses are treated symmetrically. Sortino defines risk as standard deviation of only the losing periods (or returns below a configurable threshold). For strategies where the upside and downside are roughly mirror images, both ratios give similar answers. For asymmetric strategies, they diverge meaningfully.
Both ratios share the same structure:
ratio = (mean_return − risk_free_rate) / risk_denominatorSharpe uses standard deviation of all returns:
sharpe_risk = stddev(returns)Sortino uses downside deviation — the standard deviation calculated using only returns below a "minimum acceptable return" (MAR), which is conventionally zero:
sortino_risk = sqrt(mean of (r − MAR)² where r < MAR)Annualization for both is the same: multiply by sqrt(periods_per_year). Daily returns scale by sqrt(252), monthly by sqrt(12), and so on.
The choice between ratios depends on what kind of risk you care about and what shape your strategy's returns are.
Use Sharpe when:
Use Sortino when:
For Hyperliquid carry strategies — the funding-carry pattern with mostly stable hourly P&L — Sharpe and Sortino tend to be similar. For trend-following on volatile altcoins, Sortino is consistently higher and more honestly represents the trade's risk character.
Compute Sortino / Sharpe. Three regimes:
The Sortino/Sharpe ratio is itself a strategy-character signal. Highly skewed strategies are typically more psychologically difficult to hold through losing streaks (you stop, then miss the next big winner), so a high Sortino/Sharpe sometimes reflects strategies that look great in backtest but are hard to actually run live.
The Sortino vs Sharpe calculator computes both ratios from a returns series with proper annualization and a configurable risk-free rate. It also surfaces the Sortino/Sharpe ratio so you can read strategy asymmetry directly.
For a real Keel strategy example, the funding carry strategy reports Sharpe 2.17 on a 20-month backtest — that's institutional-grade. Sortino would typically be similar on a carry trade (symmetric P&L), confirming Sharpe is the right metric to lead with for that strategy.
Keel is a Strategy OS for AI-assisted systematic trading on Hyperliquid. Backtest, optimize, and run live strategies across single-stock perps, indices, and crypto majors — realistic fees, slippage, and funding modeled.
Free to start — connect a Hyperliquid wallet when you’re ready to go live.
Sharpe divides excess return by all volatility (upside + downside); Sortino divides by downside volatility only. Sortino is more forgiving for strategies that have big winners and small losers.
For strategies with symmetric P&L (carry, market-making, mean-reversion at scale), Sharpe and Sortino are similar — pick Sharpe for convention. For strategies with positive skew (trend-following, breakout, options) where big winners drive returns, Sortino is more honest about real risk. For directional or asymmetric crypto strategies, Sortino is often the better lens.
Rough trading-strategy benchmarks. Under 1: not worth the operational complexity over buy-and-hold. 1-2: real edge, but path-dependent. 2-3: institutional-grade — most pro traders aim here. Over 3: usually too good to be true on a long sample. Check for survivorship bias, look-ahead bias, or in-sample overfitting before committing capital.
Multiply the per-period ratio by sqrt(N) where N is periods-per-year. Daily returns → sqrt(252). Weekly → sqrt(52). Monthly → sqrt(12). This standard scaling assumes returns are roughly independent across periods. For highly autocorrelated returns (e.g. monthly hedge-fund-of-funds), sqrt-scaling overstates the annualized ratio.
It means upside volatility dominates downside volatility — your strategy has positive skew. The ratio between Sortino and Sharpe is a useful signal: greater than 1.3 indicates meaningful asymmetry; greater than 2.0 is rare and worth investigating (could be real edge from a trend-friendly regime, could be sample-size artifact).
Sortino's downside-deviation calculation only includes returns below a threshold (usually 0 — i.e. only periods where you lost money). The MAR is configurable. For most use cases, set MAR = 0 (treat any loss as downside). Setting MAR = risk-free rate makes Sortino more comparable to Sharpe under risk-adjusted-excess-return framing.