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

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

A hedge fund reports a Sharpe ratio of 1.2, placing it comfortably in the top quartile of its peer group. An allocator runs the same returns through the Sortino ratio and gets 1.8. Another fund, marketed as a volatility-selling strategy, shows a Sharpe of 1.4 but a Sortino of only 0.9. The rankings reverse. Which metric is telling the truth?

The answer depends on the shape of the return distribution. The Sharpe ratio, introduced by William Sharpe (1966) and later revised in Sharpe (1994), penalizes all volatility equally, treating upside surprises the same as downside losses. The Sortino ratio, developed by Sortino and van der Meer (1991) and refined in Sortino and Price (1994), penalizes only downside deviation below a minimum acceptable return. For strategies with symmetric return distributions, both metrics produce similar rankings. For strategies with skewed returns, they can disagree sharply, and the disagreement reveals something important about the true risk profile.

The Sharpe Ratio: Assumptions and Limitations

The Sharpe ratio is defined as:

Sharpe Ratio = (R_p - R_f) / sigma_p

where R_p is the portfolio return, R_f is the risk-free rate, and sigma_p is the standard deviation of portfolio returns. The metric captures return per unit of total volatility.

The elegance of the Sharpe ratio lies in its simplicity. Standard deviation is easy to compute, well-understood, and directly tied to the normal distribution assumption that underpins much of modern portfolio theory. If returns are normally distributed, standard deviation captures the full picture of risk; upside and downside deviations are mirror images, and penalizing both equally is mathematically equivalent to penalizing just the downside.

The limitation emerges when returns are not normally distributed, which in practice is most of the time. Financial returns exhibit skewness (asymmetric tails) and excess kurtosis (fat tails). A strategy that generates frequent small gains and occasional large losses (negative skew) will have its volatility dominated by those large loss events, but its standard deviation also captures the volatility of the gains, partially offsetting the risk signal. Conversely, a strategy that generates frequent small losses and occasional large gains (positive skew) gets penalized for the upside volatility that investors actually welcome.

Sharpe (1994) himself acknowledged this limitation, noting that the ratio is most appropriate when comparing portfolios that serve as a candidate for the overall investment portfolio. When applied to individual strategies with non-normal return distributions, the Sharpe ratio can produce misleading rankings.

The Sortino Ratio: Penalizing Only What Hurts

The Sortino ratio addresses the asymmetry problem directly:

Sortino Ratio = (R_p - MAR) / sigma_d

where MAR is the minimum acceptable return (often set to zero or the risk-free rate), and sigma_d is the downside deviation, calculated using only returns below the MAR.

The key innovation is the denominator. Instead of measuring dispersion around the mean in both directions, downside deviation measures dispersion only below the threshold that defines acceptable performance. Returns above the MAR contribute zero to the risk measure, regardless of how volatile they are.

Sortino and van der Meer (1991) argued that investors do not experience upside volatility as risk. A portfolio manager who delivers 2% in one month and 8% the next has high return volatility, but no rational investor would consider those fluctuations problematic. The Sharpe ratio penalizes this manager; the Sortino ratio does not.

The MAR parameter gives the Sortino ratio additional flexibility. An endowment fund with a 5% annual spending rate might set MAR at 5%, defining risk as the probability of failing to meet obligations. A retirement portfolio might use an inflation-adjusted target. By allowing the risk threshold to reflect the investor's actual objectives, the Sortino ratio connects risk measurement to economic outcomes rather than statistical abstractions.

When the Two Metrics Agree

For strategies with approximately symmetric return distributions, the Sharpe and Sortino ratios produce similar relative rankings. This is the case for most long-only equity portfolios, diversified bond portfolios, and balanced allocations. The return distributions of these strategies approximate normality closely enough that upside and downside deviations are roughly equal.

In these contexts, the Sortino ratio will typically be approximately 1.4 times the Sharpe ratio (since downside deviation for a symmetric distribution is approximately sigma / sqrt(2)), but the ranking of multiple strategies will be preserved. If Strategy A has a higher Sharpe ratio than Strategy B, it will generally also have a higher Sortino ratio.

When They Disagree: The Skewness Signal

The informative case is when the rankings diverge. This happens with skewed return distributions, which are characteristic of several common strategy types.

Consider two hypothetical strategies, each with a 10% annualized return, a risk-free rate of 4%, and 12% annualized standard deviation:

Strategy A (Trend Following): generates a positively skewed return distribution. Most months produce small losses or small gains, but occasional large gains occur during extended market trends. Monthly return profile: median +0.3%, mean +0.83%, skewness +1.2.

Strategy B (Volatility Selling): generates a negatively skewed return distribution. Most months produce steady premium income, but occasional large losses occur during volatility spikes. Monthly return profile: median +1.1%, mean +0.83%, skewness -1.4.

Both strategies have identical mean returns and identical standard deviations, so they share the same Sharpe ratio:

Sharpe = (10% - 4%) / 12% = 0.50

But their Sortino ratios diverge. Strategy A, with positive skew, has smaller and less frequent downside deviations, yielding a Sortino ratio of approximately 0.85. Strategy B, with negative skew, has larger and more concentrated downside deviations, yielding a Sortino ratio of approximately 0.35.

The Sortino ratio reveals what the Sharpe ratio conceals: Strategy A's volatility comes primarily from upside moves that benefit the investor, while Strategy B's volatility is dominated by downside events that cause real losses. A rational investor should prefer Strategy A, and the Sortino ratio correctly reflects this preference.

Empirical Evidence: Strategy Rankings by Both Metrics

The following table presents stylized risk-return profiles for six common strategy types, illustrating how Sharpe and Sortino rankings can diverge.

The most notable ranking reversals occur with negatively skewed strategies. Volatility selling ranks second by Sharpe (0.52) but drops to fourth by Sortino (0.38). This strategy earns consistent premium income that inflates its mean return relative to total volatility, but its downside deviation is severe. Covered call writing exhibits a similar pattern, dropping from fourth (Sharpe) to last (Sortino).

Conversely, managed futures (trend following) ranks last by Sharpe (0.37) but jumps to second by Sortino (0.58). The strategy's positive skewness means its total volatility overstates its true downside risk. The Sortino ratio correctly identifies this as a more favorable risk profile than the Sharpe ratio suggests.

A Practical Comparison: Trend Following vs Volatility Selling

The ranking reversal between trend following and volatility selling deserves closer examination because it illustrates a real-world allocation decision that hinges on metric choice.

A volatility-selling strategy (systematically selling out-of-the-money put options on the S&P 500) typically exhibits a Sharpe ratio of 0.45 to 0.55 during normal market conditions. The strategy collects option premium monthly, generating a steady income stream with a high win rate (often 80-90% of months are profitable). Total standard deviation appears moderate because the many small gains partially offset the occasional large losses in the overall dispersion calculation.

A trend-following strategy typically exhibits a Sharpe ratio of 0.30 to 0.45. The strategy has a lower win rate (often 40-45% of months), with many small losses during trendless periods offset by occasional large gains during sustained trends. Total standard deviation is elevated because those large gains create upside volatility.

By Sharpe ratio alone, the volatility seller appears superior. This is precisely the ranking that prevailed through much of the 2010s, a period characterized by low volatility and steadily rising equity markets that favored option-selling strategies.

However, the Sortino ratio tells a different story. The volatility seller's downside deviation is substantially higher than its upside deviation, as losses are concentrated and large (the left tail of the distribution). The trend follower's downside deviation is substantially lower than its upside deviation, as gains are concentrated and large (the right tail). After adjusting for skewness, trend following typically ranks higher by Sortino ratio than by Sharpe ratio, and the volatility seller ranks lower.

The events of March 2020 provided a real-time stress test. The CBOE PutWrite Index (a benchmark for systematic put selling) lost approximately 16% in a single month. The SG Trend Index lost approximately 2%. Over longer crises such as 2008, the divergence was even more dramatic. The Sortino ratio's emphasis on downside risk proved to be the more informative signal.

Beyond Sharpe and Sortino: Alternative Metrics

The Sharpe-Sortino comparison is the most common, but other risk-adjusted metrics address similar concerns from different angles.

The Omega ratio, introduced by Keating and Shadwick (2002), captures the entire return distribution by computing the ratio of gains above a threshold to losses below it, without assuming any particular distributional form. It incorporates all moments of the distribution (mean, variance, skewness, kurtosis, and higher) into a single number. For normally distributed returns, the Omega ratio is a monotonic transformation of the Sharpe ratio; for non-normal returns, it provides additional information.

The Calmar ratio divides annualized return by maximum drawdown, providing a direct measure of return per unit of worst-case loss. It is particularly relevant for strategies where drawdown magnitude and duration are the primary risk concerns (such as CTAs and hedge funds). The Calmar ratio is highly sensitive to sample period, however, as a single extreme event can dominate the denominator.

Each metric illuminates a different facet of the risk-return tradeoff:

Institutional Practice: How Allocators Actually Use These Metrics

In practice, institutional investors rarely rely on a single metric. The typical due diligence process for evaluating a hedge fund or strategy involves computing multiple risk-adjusted return measures and examining where they agree and disagree.

A common institutional framework evaluates strategies across three dimensions. First, the Sharpe ratio serves as the baseline comparability measure, since nearly every strategy reports it and it enables cross-strategy ranking on a common scale. Second, the Sortino ratio (or a downside-deviation-based variant) serves as the skewness check; when the Sortino ratio diverges significantly from the expected 1.4x multiple of the Sharpe ratio, it signals non-normal returns that require further investigation. Third, the Calmar ratio or maximum-drawdown analysis serves as the tail-risk assessment, capturing the magnitude of worst-case outcomes that neither the Sharpe nor Sortino ratio fully reflects.

Rollinger and Hoffman (2013) documented that many practitioners compute the Sharpe ratio incorrectly, particularly regarding annualization of monthly data. The standard practice of multiplying a monthly Sharpe by sqrt(12) assumes returns are independent and identically distributed, which is violated by most strategy returns due to serial correlation, time-varying volatility, and regime dependence. The Sortino ratio is subject to similar annualization challenges.

Which Metric to Use When

The choice between Sharpe and Sortino depends on the strategy's return characteristics and the investor's risk priorities.

Use the Sharpe ratio as the primary metric when evaluating long-only equity portfolios, diversified balanced portfolios, or any strategy with approximately symmetric returns. In these cases, the Sharpe ratio's simplicity, comparability, and widespread acceptance make it the most practical choice. The penalty for upside volatility is minimal because upside and downside deviations are roughly equal.

Use the Sortino ratio as the primary metric (or as a critical supplement) when evaluating strategies with meaningfully skewed return distributions. This includes options-based strategies (both buying and selling), managed futures and trend following, event-driven strategies (merger arbitrage, distressed debt), leveraged strategies with non-linear payoffs, and any strategy where the investor's primary concern is avoiding losses below a specific threshold.

The strongest practice is to compute both, examine the ratio between them, and use the divergence as a diagnostic signal. When Sortino/Sharpe is notably above 1.4, the strategy has favorable positive skewness; when it is notably below 1.4, the strategy carries hidden negative skewness that the Sharpe ratio alone does not reveal.