Which is best: Sharpe, Sortino, or Calmar?

No single metric is best — they measure different things. Sharpe is the industry default and best for comparing across funds. Sortino is better for strategies with asymmetric (skewed) returns. Calmar is best for capital-allocation decisions where surviving large drawdowns matters more than smooth returns. Sophisticated allocators report all three.

Why is Sortino almost always higher than Sharpe?

Sortino only counts downside volatility in the denominator while Sharpe counts both upside and downside. Since most strategies have some upside volatility, removing it from the denominator shrinks it, which raises the ratio. The gap is largest for strategies with positive skew (occasional big wins) like trend-following.

When does Sharpe lie?

Sharpe lies when returns are non-normal — specifically when there is significant negative skew or fat tails. Strategies like option selling, carry trades, and short-volatility can show beautiful Sharpe ratios in calm periods because their few wins are small but consistent. The blowup that defines them is invisible until it happens. Calmar and Sortino partially correct this; the probabilistic Sharpe ratio (PSR) corrects it explicitly.

What Sharpe / Sortino / Calmar values are considered good?

Rough rules of thumb (annualized, after fees): Sharpe above 1.0 is good, above 2.0 is excellent, above 3.0 is suspect. Sortino above 1.5 is good (usually 1.3-1.7x the Sharpe). Calmar above 0.5 is acceptable for long-only equity, above 1.0 is good, above 3.0 is excellent and rare. These vary by asset class — bond strategies routinely show higher Sharpes than equity; crypto strategies often have lower Calmars due to deeper drawdowns.

What is the formula difference?

Sharpe = (mean return − risk-free rate) / standard deviation of returns. Sortino = (mean return − risk-free rate) / downside deviation (only returns below a threshold contribute). Calmar = annualized return / |maximum drawdown|. All three are annualized for comparison.

Why do some strategies have negative Sortino but positive Calmar?

Sortino measures recent return-per-downside-vol; Calmar measures lifetime return-per-worst-loss. A strategy that has been bleeding recently (negative Sortino) can still have a positive Calmar if its long-term return is positive — the drawdown denominator was set long ago and the cumulative return is still net positive. Watch when these diverge: it usually means the strategy has degraded but the historical track record is masking it.

Can I use these for crypto and FX, not just stocks?

Yes. The formulas are asset-class agnostic. The interpretive ranges differ: crypto strategies often show Sortino 0.8-1.5 and Calmar 0.5-1.0 due to higher tail risk, while a "good" equity long-short might show Sortino 1.5-2.0 and Calmar 1.5-2.5. Whatever asset, compute on at least 3 years of returns and use the same risk-free rate convention across strategies you compare.

Which one do allocators actually use?

Allocators report Sharpe for comparability (it is the industry default), but the metric that drives capital allocation decisions is usually Calmar — because drawdown tolerance, not volatility tolerance, is what kills funds. Many quant funds publish all three plus Sortino in monthly tearsheets. Modern shops also add the probabilistic Sharpe ratio (PSR) for statistical significance.

Where does Treynor / Information Ratio fit in?

Different family of metrics. Treynor and Information Ratio use systematic risk (beta) or active tracking error in the denominator instead of total volatility. They are best for relative-return strategies measured against a benchmark. Sharpe / Sortino / Calmar measure absolute return risk and are more useful for absolute-return strategies (most hedge funds, crypto, long-short).

Is there a calculator for all three?

Yes. The QuantOracle Sharpe Ratio Calculator computes Sharpe with a confidence interval. The Drawdown Calculator gives you max drawdown for Calmar. A standalone Sortino calculator is on the roadmap; in the meantime, the Probabilistic Sharpe Ratio Calculator surfaces the same higher-moment information (skewness, kurtosis) that motivates using Sortino.

Sharpe vs Sortino vs Calmar: Which Risk-Adjusted Return Metric Should You Use?

Use…	When…	Watch out for…
Sharpe	Comparing across funds / industry default / academic publication	Lies on non-normal returns (option selling, carry, short-vol)
Sortino	Strategies with asymmetric returns (trend, long-vol, crisis alpha)	Always higher than Sharpe — not directly comparable across managers
Calmar	Capital allocation / risk-of-ruin / surviving real drawdowns	Sensitive to one bad period — looks bad after a recent drawdown

What each metric actually measures

All three answer the same question — "how much return did this strategy generate per unit of risk?" — but they disagree about what counts as risk. That disagreement is the entire reason multiple metrics exist.

Sharpe: volatility is risk

Sharpe = (mean return − risk-free rate) / standard deviation of returns

The Sharpe ratio (Sharpe, 1966) treats any deviation from the mean as risk — including upside surprises. A strategy that returned +30% one month and +5% the next is penalized for the same total "volatility" as one that returned +5% and -20%. That matches the idea of risk used in mean-variance portfolio theory and CAPM, but it doesn't match how investors actually feel risk.

Sortino: only downside is risk

Sortino = (mean return − target) / downside deviation (returns below target only)

The Sortino ratio (Sortino & Price, 1994) fixes the symmetry problem. It only counts returns that fall below a target threshold (usually zero or the risk-free rate) toward the denominator. Upside volatility is no longer penalized. This matches the intuition that big winning months are not risk — they're the point.

Calmar: peak-to-trough loss is risk

Calmar = annualized return / |maximum drawdown|

The Calmar ratio (Young, 1991) goes further: it doesn't care about the shape of the return distribution at all. It cares about one number — the worst peak-to-trough loss ever observed. This matches the practical question allocators ask: "What is the worst experience this strategy has put investors through?"

A concrete example: three strategies, three winners

Imagine three hypothetical strategies, each running for 5 years with the same average annualized return of 12%:

Strategy	Profile	Vol	Max DD	Sharpe	Sortino	Calmar
A. Carry	Smooth → one blowup	8%	35%	1.50	2.10	0.34
B. Long-only equity	Normal-ish	15%	25%	0.80	1.10	0.48
C. Trend-following	Big wins, small losses	18%	12%	0.67	1.45	1.00

By Sharpe, Strategy A (carry) looks best at 1.50. By Sortino, A is still best at 2.10 — but B and C close the gap. By Calmar, the ranking inverts: A is worst (0.34) and C is best (1.00).

Same returns, same period, three different recommended strategies. The metric you pick determines the answer. That's why allocators report all three.

When each one lies

Sharpe lies on non-normal returns

Sharpe assumes returns are roughly normal. They're not, for most real strategies. Three pathological cases:

Negative skew (option selling, carry trades, short-volatility): the strategy has many small wins and rare large losses. Until a tail event hits, Sharpe looks great. Then it doesn't.
Fat tails (high kurtosis): the strategy occasionally has moves much larger than the normal distribution would predict. Sharpe under-estimates the risk because standard deviation under-weights tail observations.
Short sample sizes: with fewer than ~30 monthly observations, the sample Sharpe has a wide confidence interval. A 6-month strategy with Sharpe 3.0 could really be anywhere from -1 to +5. The probabilistic Sharpe ratio corrects for this explicitly.

Sortino lies by always looking better than Sharpe

Sortino mechanically produces a higher number than Sharpe (for the same strategy) because the denominator is smaller. The ratio of Sortino to Sharpe is typically 1.3-1.7x. That's not a feature — it just means you can't directly compare a Sortino from one fund to a Sharpe from another. Apples-to-apples comparison requires using the same metric.

Sortino also still uses a denominator computed from past observations. A strategy that had no big down months in-sample will have an artificially small downside deviation and therefore a sky-high Sortino. Recent crisis-alpha strategies (long-vol, tail hedges) can look terrible by Sortino during calm regimes for exactly this reason — they bleed steadily with no large up months either.

Calmar is dominated by one data point

Calmar's denominator is the single worst drawdown. That means it can change overnight if a new worst-ever drawdown happens. A strategy with a 5-year track record and Calmar 1.5 can drop to Calmar 0.6 after one bad quarter. It also means a strategy that happens not to have experienced a drawdown yet (because it's only 18 months old, or because the regime has been favorable) will show an inflated Calmar.

Calmar also doesn't care how frequently drawdowns happen. A strategy with one big drawdown and three years of recovery scores the same as a strategy that experiences the same drawdown depth annually. Some practitioners use the average drawdown (Pain Index) alongside Calmar to fix this.

What good values look like

Rough ranges for what's acceptable in different contexts. Annualized, after fees, computed on at least 3 years of monthly returns:

Strategy class	Good Sharpe	Good Sortino	Good Calmar
Long-only equity	0.5 – 0.8	0.7 – 1.2	0.3 – 0.6
60/40 balanced	0.6 – 1.0	0.9 – 1.4	0.5 – 0.9
Hedge fund (typical)	0.8 – 1.5	1.2 – 2.0	0.7 – 1.5
Quant CTA / trend	0.6 – 1.2	1.0 – 2.0	0.5 – 1.5
Market-neutral	1.0 – 2.0	1.5 – 2.5	1.0 – 3.0
Crypto strategy	0.5 – 1.5	0.8 – 1.8	0.3 – 1.0

Sharpe above 3.0 on a multi-year sample is rare and usually indicates a data error, look-ahead bias, or unaccounted-for transaction costs. Worth scrutinizing before allocating to.

The decision rule (use this, save the rest for context)

Always compute Sharpe. It's the industry default and the only metric an LP will compare to other managers without conversion. Use the Sharpe ratio calculator (with its 95% confidence interval) — and run the probabilistic Sharpe ratio to check whether the number is statistically meaningful.
If the strategy has skewed returns, also compute Sortino. Skew < -0.5 or > +0.5 is enough to make Sharpe misleading. The PSR output gives you skewness as a free side-effect; check it.
Always compute Calmar before allocating capital. The drawdown calculator gives you max drawdown, which is the Calmar denominator. If the strategy hasn't had a drawdown yet (under-2-year track records), assume Calmar will be lower than the backtest suggests.
Report all three in your tearsheet. Putting just one number in front of an allocator who knows finance is a red flag. Putting all three with one-sentence interpretations is what a competent shop does.

Related calculators

Sharpe Ratio Calculator — Sharpe with a 95% confidence interval (most calculators omit the CI; it matters a lot for short samples)
Probabilistic Sharpe Ratio Calculator — Lopez de Prado 2012 PSR adjusting for sample size, skew, and kurtosis
Drawdown Calculator — max drawdown (the Calmar denominator), average drawdown, recovery time
Value at Risk Calculator — parametric and historical VaR + CVaR for the same return series
Monte Carlo Simulation Calculator — forward-projects Calmar by simulating thousands of return paths

References

Sharpe, W. F. (1966). "Mutual fund performance." Journal of Business 39, 119-138.
Sortino, F. & Price, L. (1994). "Performance measurement in a downside risk framework." Journal of Investing 3(3), 59-64.
Young, T. (1991). "Calmar ratio: A smoother tool." Futures Magazine.
Bailey, D. H. & Lopez de Prado, M. (2012). "The Sharpe ratio efficient frontier." Journal of Risk 15(1).
Magdon-Ismail, M. & Atiya, A. (2004). "Maximum drawdown." Risk Magazine, October.