Home / Performance Metrics / Sharpe Ratio
Risk-adjustedIntermediate

Sharpe Ratio

The Sharpe ratio is a risk-adjusted performance measure equal to the portfolio's excess return over the risk-free rate divided by the standard deviation of its returns, expressing how much reward the strategy earned per unit of total volatility.

Quick answer: The Sharpe ratio is a risk-adjusted performance measure equal to the portfolio's excess return over the risk-free rate divided by the standard deviation of its returns, expressing how much reward the strategy earned per unit of total volatility.

In simple words

The Sharpe ratio scores a strategy by how much return it earns for the amount of bounce in its returns. Two strategies might both make 15 percent a year, but the one whose equity curve is smoother has the higher Sharpe and is the better risk-adjusted performer. It is the most widely quoted risk-adjusted metric, but it treats upside swings as “risk” just like downside ones, which is its central quirk.

Purpose

The Sharpe ratio exists so that returns can be compared on a level playing field after accounting for the volatility taken to earn them, preventing a strategy from looking good merely because it took large risks.

Visual explanation

Sharpe Ratio

A bell-shaped distribution of returns whose centre sits above the risk-free rate and whose width is the volatility; Sharpe is the distance of the centre above the risk-free line measured in units of that width.

Sharpe Ratio — Reward per Unit of RiskReturn − Risk-freeVolatility σSharpe ratioreturn →mean−σ+σspread = σ

Professional explanation

The definition and what excess return means

The Sharpe ratio is the mean excess return divided by the standard deviation of returns, where excess return is the strategy's return minus the risk-free rate over the same interval. The risk-free rate matters: in India a short-dated T-bill or overnight rate is the usual proxy, and subtracting it ensures you only reward the strategy for return above what idle cash would have earned. Using zero for the risk-free rate, common in casual backtests, silently inflates the Sharpe, especially when rates are high.

Annualising the ratio correctly

Sharpe is usually computed on periodic returns and then annualised. Because mean return scales with the number of periods p while standard deviation scales with the square root of p, the annualised Sharpe is the periodic Sharpe multiplied by the square root of p. For daily returns that means multiplying by the square root of about 252. This square-root scaling assumes returns are independent across periods; if they are autocorrelated, as trend or mean-reversion strategies often are, the naive annualisation is biased.

Why standard deviation is the risk proxy, and its cost

Using total standard deviation treats every deviation from the mean, up or down, as risk. This is mathematically convenient and correct if returns are roughly normal, but real strategy returns are often skewed and fat-tailed. An option-selling strategy that grinds out small gains and occasionally suffers a huge loss can show a flattering Sharpe because its many small positive deviations dominate the volatility estimate while the rare catastrophic tail is under-weighted. Sharpe's assumption of symmetric, well-behaved risk is precisely where it misleads.

The deflated Sharpe and multiple-testing inflation

A Sharpe computed from the single best of many tested strategies is upward-biased, because trying many configurations and keeping the winner is a form of data snooping. The deflated Sharpe ratio adjusts the observed Sharpe for the number of independent trials, the length of the track record, and the skew and kurtosis of returns, giving a probability that the true Sharpe exceeds zero. A high in-sample Sharpe from a wide parameter search should always be discounted; the more configurations you tried, the more a high Sharpe is expected by chance alone.

Sample size and the stability of the estimate

Sharpe is a ratio of two sample statistics and is itself noisy, with a standard error that shrinks only slowly with track-record length. A Sharpe estimated from a few months of daily data has such a wide confidence interval that it is nearly uninformative; distinguishing a true Sharpe of 0.5 from 1.5 can require years of data. Reporting a Sharpe without acknowledging its sampling uncertainty, or annualising one from a tiny sample, is a common and serious error.

What a given Sharpe does and does not tell you

As a loose convention, an annualised Sharpe below 1 is often called modest, around 1 to 2 respectable, and above 2 excellent, but these are heuristics, not laws, and they are conditional on honest, out-of-sample, cost-inclusive returns. A high Sharpe says nothing about maximum drawdown, tail risk, capacity or whether the edge survives live. A strategy can have a superb Sharpe and still be ruined by a single event its return distribution never sampled.

Formula

Sharpe = (Rp − Rf) ÷ σp ; annualised Sharpe = periodic Sharpe × √p

Rp = the portfolio (strategy) mean return over the interval, Rf = the risk-free rate over the same interval, σp = the standard deviation of the portfolio's returns. p = periods per year (about 252 for daily). The √p factor annualises and assumes returns are independent across periods; autocorrelation biases it. Standard deviation counts upside and downside deviations equally.

Sharpe ratio vs Sortino ratio

AspectSharpe ratioSortino ratio
Risk measureTotal standard deviationDownside deviation only
Penalises upside volatilityYesNo
Best when returns areRoughly symmetricSkewed or asymmetric
Reference returnRisk-free rateA target or minimum acceptable return
Shared blind spotAssumes stable distribution, ignores tail events beyond varianceSame, though downside-focused

Practical example

Illustrative example (Indian market)

A Nifty futures strategy produces a mean daily return of 0.06 percent with a daily standard deviation of 0.9 percent, while the risk-free rate is about 6 percent a year, or roughly 0.0238 percent per trading day. The daily excess return is 0.06 minus 0.0238 = 0.0362 percent, so the daily Sharpe is 0.000362 ÷ 0.009 ≈ 0.0402. Annualising by the square root of 252 (about 15.87) gives an annualised Sharpe of about 0.64. This is a modest figure, and it would fall further once realistic STT, brokerage and slippage reduce the mean return, which is why costs must be modelled before Sharpe is quoted.

Deep out-of-the-money index option selling on NSE frequently shows a seductively high backtest Sharpe because the strategy books many small premiums and rarely realises its tail, so standard deviation understates the true risk; a single gap event like a budget-day or global-shock move can erase months of grind that the Sharpe never reflected.

Advantages

  • Standardises return comparison by adjusting for volatility taken
  • Universally understood, enabling comparison across strategies and asset classes
  • Simple to compute from a return series and a risk-free rate
  • Annualises cleanly under the independence assumption
  • Its deflated variant can correct for multiple-testing inflation

Limitations

  • Treats upside volatility as risk, penalising favourable large moves
  • Assumes roughly symmetric returns, so it flatters skewed, fat-tailed strategies whose blind spot is the rare tail
  • Naive annualisation is biased when returns are autocorrelated
  • Noisy in small samples, with a wide confidence interval
  • Inflated by multiple testing unless deflated
  • Says nothing about maximum drawdown, capacity or tail risk

Why it matters in practice

  • It is the default risk-adjusted number allocators screen on, so its blind spots must be disclosed
  • The deflated Sharpe is one of the clearest defences against data-snooped results

Common mistakes

  • Using a risk-free rate of zero, which inflates the Sharpe when rates are high
  • Annualising a Sharpe from a few weeks of data as if it were stable
  • Reporting the single best Sharpe from a wide parameter search without deflating it
  • Trusting a high Sharpe on an option-selling strategy whose tail risk it hides
  • Ignoring autocorrelation when annualising a trend or mean-reversion strategy
  • Quoting Sharpe without max drawdown, treating one number as a full risk picture

Professional usage

Institutional researchers report Sharpe alongside Sortino, maximum drawdown and, crucially, the deflated Sharpe when a parameter search was involved, and they always specify the risk-free proxy and the annualisation convention. They are sceptical of very high Sharpes from skewed strategies, checking the return distribution for negative skew and excess kurtosis before believing the number. They also confront the Sharpe with its sampling error, refusing to distinguish two strategies whose confidence intervals overlap, and they insist the returns be net of realistic costs.

Key takeaways

  • Sharpe is excess return over the risk-free rate divided by total return volatility
  • Annualise by multiplying the periodic Sharpe by the square root of periods per year
  • It counts upside and downside deviations equally, flattering skewed strategies
  • It is noisy in small samples and inflated by multiple testing unless deflated
  • Never read Sharpe without max drawdown and the return distribution beside it

Frequently asked questions

What is the Sharpe ratio?
The Sharpe ratio is a strategy's excess return over the risk-free rate divided by the standard deviation of its returns. It measures how much reward was earned per unit of total volatility, letting strategies be compared on a risk-adjusted basis.
How do I annualize the Sharpe ratio?
Multiply the periodic Sharpe by the square root of the number of periods per year, about the square root of 252 for daily returns. This assumes returns are independent across periods; autocorrelation makes the naive annualization biased.
What risk-free rate should I use in India?
A short-dated Treasury bill or overnight rate is the usual proxy. Using zero, as many casual backtests do, silently inflates the Sharpe, an error that grows when interest rates are high.
What is a good Sharpe ratio?
As a loose convention, below 1 is modest, 1 to 2 is respectable and above 2 is excellent, but these are heuristics conditional on honest, out-of-sample, cost-inclusive returns. A high in-sample Sharpe from a wide search is not the same thing.
Why does Sharpe penalize upside volatility?
Because it uses total standard deviation, which counts every deviation from the mean equally, whether the return was above or below average. A strategy with large favourable swings is penalized as if that upside were risk, which many argue is a flaw the Sortino ratio addresses.
What is the deflated Sharpe ratio?
It adjusts an observed Sharpe for the number of strategy configurations tested, the track-record length, and the skew and kurtosis of returns, yielding the probability that the true Sharpe exceeds zero. It corrects the upward bias created by keeping the best of many trials.
Why can option-selling strategies show a misleadingly high Sharpe?
Because they book many small premiums and rarely realize their tail, so standard deviation understates the true risk. The Sharpe looks excellent until a single gap event delivers the catastrophic loss the volatility estimate never captured.
How much data do I need for a reliable Sharpe?
More than most people assume. Sharpe is a noisy ratio with a slowly shrinking standard error, and distinguishing a true Sharpe of 0.5 from 1.5 can require years of returns. A Sharpe from a few months is nearly uninformative.
Does a high Sharpe mean low drawdown?
Not necessarily. Sharpe measures reward per unit of volatility, not the depth of the worst loss. A strategy can have a high Sharpe and still suffer a severe drawdown, so the two must be read together.
How do trading costs affect the Sharpe ratio?
Costs lower the mean return and therefore the Sharpe, and the effect is largest for high-turnover strategies. A Sharpe computed on gross returns can look far better than the achievable net figure once STT, brokerage and slippage are applied.
Is the Sharpe ratio affected by autocorrelation?
Yes. If returns are positively autocorrelated, the square-root-of-time annualization understates volatility and overstates the annual Sharpe; negative autocorrelation does the reverse. Trend and mean-reversion strategies often violate the independence assumption.
What is the difference between Sharpe and information ratio?
Sharpe measures excess return over the risk-free rate per unit of total volatility, while the information ratio measures excess return over a benchmark per unit of tracking error. Sharpe judges standalone risk-adjusted return; the information ratio judges skill relative to a benchmark.
Can the Sharpe ratio be negative?
Yes. If the strategy's mean return is below the risk-free rate, the excess return is negative and so is the Sharpe, indicating it underperformed cash on a risk-adjusted basis.
Should I rank strategies purely by Sharpe?
No. Sharpe ignores tail risk, drawdown depth, capacity and skew, and it is inflated by multiple testing. Use it as one input alongside Sortino, drawdown, the deflated Sharpe and out-of-sample evidence.
Why did my Sharpe change when I switched from daily to monthly returns?
Because the estimate and its annualization depend on the sampling frequency and on any autocorrelation in returns. Different frequencies capture different structure, so the annualized Sharpe from daily and monthly data need not match.

Voice search & related questions

Natural-language questions people ask about Sharpe Ratio.

What is the Sharpe ratio in simple terms?
It is how much return a strategy earns for each unit of bounce in its returns, so a smoother curve at the same return scores higher.
How do I make a Sharpe ratio annual?
Multiply the periodic Sharpe by the square root of the number of periods in a year, which is about the square root of 252 for daily data.
What counts as a good Sharpe ratio?
Roughly, above one is respectable and above two is excellent, but only if the returns are honest, out-of-sample and after costs.
Why do people say Sharpe is flawed?
Because it treats big upside moves as risk and it can flatter strategies that hide a rare, huge loss behind lots of small gains.
Does a high Sharpe mean I won't have big drawdowns?
No, Sharpe measures reward per unit of volatility, not the worst loss, so you still have to check the drawdown separately.
What is the deflated Sharpe ratio?
It is a corrected Sharpe that accounts for how many strategies you tried, so a lucky best-of-many result gets discounted.

Sources & references

    Last reviewed 11 July 2026. Educational content only — not investment advice. Markets and rules change; verify current conventions with SEBI, NSE/BSE and your broker.

    ← Previous
    Annualized Return
    Educational content only — not investment advice. Examples use illustrative numbers and simplified models. Backtested results are hypothetical and trading derivatives involves substantial risk. See our Risk Disclosure and SEBI Disclaimer.