Hedge Fund Performance Metrics: How Allocators Measure Sharpe, Drawdown and Unsmoothed Risk

Hedge fund performance metrics are the statistics an allocator computes from a manager's monthly net returns: annualised return and volatility, the Sharpe and Sortino ratios, maximum drawdown and recovery, correlation, skew and kurtosis. Each is an estimate, and each can mislead. On 98 years of US equity data from 1928 to 2025, the annual Sharpe ratio was 0.43, yet ten-year windows of the same series ranged from minus 0.17 to 1.34. Smoothed marks, short samples, convenient start dates and gross figures move the numbers further. This article computes the core metrics on real data, shows where each one breaks, and sets out what a launching manager should present.

When an allocator receives a monthly return series, the analysis starts before anyone reads the commentary. They recompute the Sharpe ratio, test for serial correlation, unsmooth the series where the marks look too stable, and check drawdowns against their own book. We tell managers to run the same numbers first and publish them on the same basis. A Sharpe ratio of 2 on three years of smoothed, gross returns rarely survives that process. A Sharpe ratio of 0.8, net of fees, independently administered and honestly explained, often does.

David Lloyd, Chief Executive Officer at CV5 Capital

Executive Summary

Allocators rebuild a manager's headline statistics from the monthly return series and adjust for the known ways in which they flatter. A manager who understands the adjustments can prepare materials that hold up.

• The Sharpe ratio is an estimate with a standard error. On 98 years of US equity returns the 95 per cent interval runs from about 0.22 to 0.64; on ten years it runs from 0.14 to 1.60.
• Multiplying a monthly Sharpe ratio by the square root of 12 is correct only for independent returns. Andrew Lo showed that serial correlation can overstate a hedge fund's annual Sharpe ratio by as much as 65 per cent.
• Smoothed returns understate volatility and drawdown. Applying a simple three-year smoothing to real equity returns cut measured volatility from 19.2 to 11.4 per cent and lifted the Sharpe ratio from 0.43 to 0.71.
• Track record length matters. At an annualised Sharpe ratio of 1.0 and normal returns, about 35 months of monthly data are needed to be 95 per cent confident the true ratio is above zero.
• For SEC-registered advisers, the marketing rule requires net performance alongside any gross performance, with equal prominence and the same period and method.

The Metrics Allocators Compute From a Monthly Return Series

The starting point is the fund's monthly net returns from inception, ideally as calculated by the administrator. A short primer on the ratios themselves is in our guide to Sharpe, Sortino and risk-adjusted return metrics. This article goes further, into how each figure is computed and where it misleads.

William Sharpe introduced the measure in the Journal of Business in 1966 under the title Mutual Fund Performance. His 1994 article in the Journal of Portfolio Management restated it as the average differential return divided by the standard deviation of that differential return. The denominator is the volatility of excess returns, not raw returns, and Sharpe noted that the ratio is not independent of the measurement period.

A Worked Example on 98 Years of US Returns

The figures below use the annual return series for 1928 to 2025 published by Professor Aswath Damodaran at NYU Stern. They cover US large-capitalisation equities with dividends, the 10-year US Treasury bond and the 3-month Treasury bill as the risk-free rate. CV5 Capital computed every figure independently from the source file. A fund's monthly series is analysed the same way.

Definitions are stated so that the numbers can be reproduced. The Sharpe ratio is the mean annual return above the Treasury bill, divided by the sample standard deviation of those excess returns. The Sortino ratio uses the same numerator, divided by the downside deviation of excess returns below zero across all years. Maximum drawdown is measured on year-end cumulative values.

Figures current as at 14 October 2026, computed by CV5 Capital from the dataset as published. Over the full period, US equities returned an arithmetic mean of 11.85 per cent and a geometric mean of 10.02 per cent, with volatility of 19.40 per cent. The equity drawdown from the end of 1928 to the end of 1932 was 64.8 per cent, and year-end value did not regain its 1928 level until 1936.

Three lessons follow. First, the most recent ten years produced a Sharpe ratio twice the long-run figure. A record that began in 2016 would look exceptional on market exposure alone. Second, the Sortino ratio rises sharply when a sample contains few losing years, because the downside deviation has little to work with. Third, drawdown is path dependent. The Treasury bond, usually treated as the defensive asset, fell 21.5 per cent from the end of 2020 to the end of 2022. At the end of 2025 it remained 13.5 per cent below its 2020 level.

Annual sampling also hides losses that recover within a year, and monthly reporting hides intra-month losses in the same way. Allocators therefore read drawdown alongside sampling frequency. In 2022, US equities fell 18.0 per cent and the 10-year Treasury fell 17.8 per cent in the same calendar year.

Why a Sharpe Ratio Is an Estimate, Not a Fact

Andrew Lo set out the statistical properties of the Sharpe ratio in the Financial Analysts Journal in 2002, in The Statistics of Sharpe Ratios. For independent and identically distributed returns, the variance of the estimated ratio is one plus half the squared ratio, divided by the number of observations. The standard error is its square root. That gives a simple test of how much a track record can prove.

On the 98-year equity series, the Sharpe ratio of 0.43 has a standard error of 0.11. An approximate 95 per cent interval runs from 0.22 to 0.64. On the ten years from 2016 to 2025, the ratio of 0.87 has an interval of 0.14 to 1.60. Ten annual observations cannot distinguish a modest ratio from an exceptional one.

Across all 89 rolling ten-year windows from 1928 to 2025, the equity Sharpe ratio ranged from minus 0.17, for 1965 to 1974, to 1.34, for 2012 to 2021. The median was 0.45. Five-year windows ranged from minus 0.34 to 3.53. A manager's three-year record carries at least that degree of period sensitivity, before any question of skill.

Minimum track record length

David Bailey and Marcos López de Prado addressed the same problem in The Sharpe Ratio Efficient Frontier, published in the Journal of Risk in 2012. They introduced the probabilistic Sharpe ratio, the probability that the true ratio exceeds a chosen threshold. They also derived the minimum track record length needed to reject the hypothesis that the true ratio is below that threshold at a given confidence level. Their formula allows for skew and kurtosis. Negative skew and fat tails lengthen the record required.

Illustrative arithmetic only, with no implied performance, rounded up to whole months: monthly observations, a threshold of zero and 95 per cent one-sided confidence, applying the minimum track record length formula as published by Bailey and López de Prado. On the real 98-year equity series, with its measured skew and kurtosis, the same formula gives about 19 years of annual data. A manager with an annualised Sharpe ratio of 0.5 would need roughly 11 years of monthly returns to make the equivalent statistical case.

Their later paper, The Deflated Sharpe Ratio, in the Journal of Portfolio Management in 2014, adds a correction for selection bias. Where many strategy variants are tested and the best is reported, the best Sharpe ratio is inflated by the search itself. An allocator reviewing a backtested systematic strategy can reasonably ask how many variants were tried. These issues sit beside the portfolio question covered in our analysis of the efficient frontier and hedge fund portfolio construction.

Smoothing, Serial Correlation and Unsmoothed Risk

Getmansky, Lo and Makarov showed in the Journal of Financial Economics in 2004 that hedge fund returns are often highly serially correlated. They found illiquidity the most likely explanation. In their model, a reported return is a weighted average of the current and past true returns, with weights between zero and one that sum to one. Smoothing leaves the mean unchanged. It lowers volatility, inflates the Sharpe ratio and depresses measured correlation with markets.

The effect can be shown on real data. The illustration below applies a smoothing with weights of 0.5, 0.3 and 0.2 on the current and two prior years to the actual equity returns for 1930 to 2025. The weights are illustrative; the underlying returns are real.

• Measured volatility fell from 19.2 to 11.4 per cent, close to the model's predicted ratio of 0.62.
• The Sharpe ratio rose from 0.43 to 0.71, with almost no change in the average return (11.7 against 11.6 per cent).
• Maximum drawdown shrank from 61.6 to 50.0 per cent.
• First-order autocorrelation rose from minus 0.01 to 0.52, the statistical signature that allocators test for.

Given the weights, the model can be inverted to recover the true series exactly, as the code for this article confirmed. In practice the weights are estimated from the fund's own autocorrelations, and allocators commonly unsmooth a series where first-order autocorrelation looks material. The adjustment usually raises measured risk. Valuation policy is where smoothing begins, which is why a robust fund valuation policy matters to the performance record as well as to the NAV.

Drawdown, Recovery and the Shape of Losses

Volatility treats a large gain and a large loss as equal. Allocators do not, and drawdown measures address the difference. Maximum drawdown is the largest fall from a running peak to a subsequent trough. Recovery time is the period from that trough back to the previous peak. Some funds never recover within the sample, as the Treasury bond had not by the end of 2025.

A Calmar-style ratio divides annualised return by maximum drawdown. On the 98-year equity series, a geometric return of 10.02 per cent against a 64.8 per cent drawdown gives 0.15. For 2016 to 2025 the same ratio is 0.81. The difference reflects one event, the absence of a 1929-type loss from the recent window.

Skew and kurtosis describe the shape of the distribution. US equity excess returns from 1928 to 2025 showed skew of minus 0.37. In the 2016 to 2025 window, skew was minus 1.41. Strategies that sell options, provide liquidity or take credit risk commonly show negative skew, earning steadily until a large loss. A short sample may not contain that loss, so allocators ask for stress tests as well as historical statistics.

For a manager, the practical response is to present drawdown in full rather than in summary. Allocators expect:

• the maximum drawdown with its peak, trough and recovery dates, on monthly data;
• the five largest drawdowns, not only the worst one;
• the fund's return in identified stress periods, alongside the allocator's likely core holdings;
• the risk limits in force at the time, and any changes made afterwards.

Presenting Performance: Net, Gross and the SEC Marketing Rule

How performance is presented is partly a regulatory matter. For investment advisers registered with the SEC, rule 206(4)-1 under the Investment Advisers Act, the marketing rule, governs performance in advertisements. Paragraph (d)(1) prohibits any presentation of gross performance unless the advertisement also presents net performance. Net performance must have at least equal prominence, in a format designed to facilitate comparison. It must cover the same time period and use the same type of return and methodology.

Paragraph (d)(2) requires one, five and ten year performance results, ending no earlier than the most recent calendar year-end. Its text expressly excludes private funds, so it does not apply to a hedge fund's own performance. Allocators nonetheless expect consistent periods, and many will ask for them. Paragraph (d)(6) sets conditions for hypothetical performance, which the rule defines to include model portfolio and backtested performance. Its application to managers outside the United States is covered in our SEC marketing rule guide for offshore fund managers.

The Global Investment Performance Standards, maintained by CFA Institute, are a voluntary industry standard for calculating and presenting performance. They are not law. For managers selling through consultants they can function as a screening criterion, as discussed in our analysis of GIPS compliance for hedge fund managers.

Institutional reviewers typically ask for the full monthly net series from inception, the fee basis of the share class shown, the risk-free rate used and the method of each calculation. Requests of this kind are standard in the hedge fund due diligence questionnaire, and answers that are reproducible from the administrator's figures carry more weight.

Common Mistakes

• Annualising a monthly Sharpe ratio by the square root of 12 without testing for serial correlation.
• Reporting arithmetic mean returns where compounded returns are the relevant measure.
• Choosing a start date that excludes an early drawdown, or presenting only the strongest period.
• Showing gross returns more prominently than net, or omitting the fee basis of the share class.
• Quoting a Sortino ratio from a sample with very few losing months, without the sample size.
• Presenting backtested or model results alongside live results without separating them clearly.
• Calculating metrics in-house on figures that differ from the administrator's NAV.

The CV5 Implementation Perspective

CV5 Capital is the Cayman-headquartered institutional fund infrastructure platform for hedge fund and digital asset managers who need to launch quickly, operate properly and satisfy serious investors from day one. It operates CV5 SPC for traditional strategies and CV5 Digital SPC for digital asset strategies, both registered with CIMA under the Mutual Funds Act. Each manager receives a segregated portfolio within an SPC governed by an independent board, with coordinated fund administration, audit, banking, custody and compliance arrangements. CV5 Capital is not the investment manager of the strategies on its platforms.

For a manager building a record that allocators will test, the Cayman hedge fund platform for emerging managers provides the structure behind the return series. Independently administered NAV, regular investor reporting and annual audited financial statements give allocators figures they can reproduce. The manager retains the strategy and every investment decision. The process of building that record from launch is set out in our guide to building an institutional track record as an emerging manager.

Key Takeaways

• Compute the allocator's metrics before the allocator does, from the administrator's monthly net returns, and state every definition used.
• Report the Sharpe ratio with its sample length, and show how it varies across sub-periods rather than a single full-period figure.
• Test the return series for serial correlation and present unsmoothed volatility where autocorrelation is material.
• Present drawdowns in full, with dates, recovery times and returns in stress periods against core holdings.
• Show net performance with at least equal prominence to gross, over the same periods and on the same method.
• Separate live, model and backtested results, and disclose how many strategy variants a backtest selected from.

Frequently Asked Questions

Which hedge fund performance metrics do allocators look at first?

Most start with annualised net return and volatility, the Sharpe ratio, maximum drawdown with recovery time, and correlation with their existing holdings. They then look at the Sortino ratio, skew and kurtosis, and test for serial correlation. All are computed from the monthly net return series from inception.

How is the Sharpe ratio of a hedge fund calculated?

Subtract the risk-free rate from each period's return, then divide the average excess return by the standard deviation of excess returns. Monthly ratios are commonly annualised by the square root of 12, which is correct only for independent returns. Serially correlated returns need Lo's adjusted annualisation factor.

What is the difference between the Sharpe and Sortino ratios?

The Sharpe ratio divides excess return by total volatility. The Sortino ratio divides it by downside deviation, which counts only returns below a target. On US equity data for 1928 to 2025, the Sharpe ratio was 0.43 and the Sortino ratio 0.80, using the Treasury bill as both risk-free rate and target.

How long a track record does a hedge fund need?

It depends on the Sharpe ratio and the shape of returns. Using the Bailey and López de Prado minimum track record length formula, normal returns and 95 per cent confidence, an annualised Sharpe ratio of 1.0 needs about 35 months of monthly data. A ratio of 0.5 needs about 133 months. Negative skew lengthens both.

What does unsmoothed volatility mean?

Unsmoothed volatility is a fund's volatility after removing the serial correlation that stale or smoothed valuations introduce. Getmansky, Lo and Makarov modelled reported returns as a weighted average of current and past true returns. Inverting that model recovers a series with higher, more realistic volatility and a lower Sharpe ratio.

Does the SEC marketing rule require 1, 5 and 10 year returns for hedge funds?

No. Rule 206(4)-1(d)(2) requires one, five and ten year periods for portfolio performance but expressly excludes private funds. Paragraph (d)(1) does apply: an SEC-registered adviser that shows gross performance must also show net performance with at least equal prominence, over the same period and method.