Skip to content

Sharpe and Sortino ratios, explained

By Investor Wisdom Daily Team·

Imagine two hypothetical funds. Over the same year, Fund X returned 12% and Fund Y returned 9%. X looks better. Now add one fact: X’s monthly returns swung wildly, with several sharp drops, while Y climbed in small, steady steps. Which one did a better job for the risk it took?

That is the question risk-adjusted ratios try to answer. Suppose a hypothetical savings account paid 3% with no risk. X earned 12 − 3 = 9 percentage points more than that, Y earned 9 − 3 = 6 points more. If X’s returns bounced around three times as much as Y’s (say a typical swing of 18 against 6), X earned 9 ÷ 18 = 0.5 point of extra return per point of swing, while Y earned 6 ÷ 6 = 1. On that measure Y did twice the work per unit of risk.

This lesson shows how to compute that “return per unit of swing” properly, in two versions: the Sharpe ratio and the Sortino ratio.

A return is the percentage gain or loss over a period. The risk-free rate is what you could have earned with (almost) no risk over the same period; analysts usually use short-term government bills, such as US Treasury bills, which TreasuryDirect sells in terms from four weeks to 52 weeks and pay their face value at maturity. Excess return is the return minus the risk-free rate. Standard deviation measures how widely returns scatter around their average. The Sharpe ratio, named after economist William F. Sharpe, divides average excess return by standard deviation. The Sortino ratio keeps the same numerator but divides by downside deviation, which counts only returns that fell below a target. Both answer: how much extra return did you get for each unit of risk, as each one defines risk?

Item Sharpe ratio Sortino ratio
Numerator Mean return − risk-free rate Mean return − target (here, the risk-free rate)
Denominator Standard deviation of all returns Downside deviation (only shortfalls below the target)
Units Pure number (percent ÷ percent) Pure number
Treats a big gain as Risk (it widens the spread) Not risk

Ratio values change with conventions, so here are ours. Use the same ones whenever you compare two portfolios.

  • Returns are hypothetical monthly returns in percent. No real fund or security is used.
  • Risk-free rate: a hypothetical 3% a year, or 3 ÷ 12 = 0.25% a month. This is an assumption for teaching, not a current rate.
  • Standard deviation: the sample version, dividing by (n − 1), because six months is a sample of a longer history. Spreadsheets call this STDEV.S.
  • Downside deviation: the target is the risk-free rate (0.25% a month); shortfalls below it are squared, summed and divided by all n months, not just the losing ones. Some sources use a target of 0 or divide differently, which gives different numbers.

Worked example 1: standard deviation and the Sharpe ratio

Section titled “Worked example 1: standard deviation and the Sharpe ratio”

Portfolio A’s six monthly returns are: 2, −1, 3, 1, −2, 3 (percent).

  1. Mean return. Sum: 2 − 1 + 3 + 1 − 2 + 3 = 6. Mean: 6 ÷ 6 = 1% a month.
  2. Deviations from the mean. Subtract 1 from each: 1, −2, 2, 0, −3, 2.
  3. Square them. 1² + (−2)² + 2² + 0² + (−3)² + 2² = 1 + 4 + 4 + 0 + 9 + 4 = 22. Squaring makes every deviation positive and weights big misses more.
  4. Variance (sample). 22 ÷ (6 − 1) = 4.4.
  5. Standard deviation. Square root of 4.4 = 4.4^0.5 ≈ 2.0976% a month.
  6. Excess return. 1 − 0.25 = 0.75% a month.
  7. Sharpe ratio (monthly). 0.75 ÷ 2.0976 ≈ 0.3575.
Month Return Minus mean (1) Squared
1 2 1 1
2 −1 −2 4
3 3 2 4
4 1 0 0
5 −2 −3 9
6 3 2 4
Total 6 0 22

What changes with the population version? Dividing by n instead of n − 1 gives 22 ÷ 6 and a standard deviation of about 1.9149. The Sharpe ratio would look higher only because of the formula. With short histories the choice matters; with many years of data it barely does. Pick one and keep it.

Worked example 2: downside deviation and the Sortino ratio

Section titled “Worked example 2: downside deviation and the Sortino ratio”

Same portfolio, same 0.25% target.

  1. Shortfalls below the target. For each month, take return − 0.25 and keep it only if it is negative; otherwise use 0. Month 2: −1 − 0.25 = −1.25. Month 5: −2 − 0.25 = −2.25. The other four months beat 0.25, so they contribute 0.
  2. Square and sum. 1.25² + 2.25² = 1.5625 + 5.0625 = 6.625.
  3. Divide by all six months. 6.625 ÷ 6 ≈ 1.1042.
  4. Downside deviation. (6.625 ÷ 6)^0.5 ≈ 1.0508% a month.
  5. Sortino ratio (monthly). 0.75 ÷ 1.0508 ≈ 0.7137.

The Sortino ratio is about twice the Sharpe ratio here because two of the months that widened the standard deviation (the 3% gains) were good news. Sortino ignores them.

Monthly ratios look small, so most reports annualise them. The usual shortcut:

  • Annual excess return ≈ monthly excess × 12: 0.75 × 12 = 9%.
  • Annual standard deviation ≈ monthly × √12: 2.0976 × 12^0.5 ≈ 2.0976 × 3.4641 ≈ 7.2664%.
  • Annual Sharpe ≈ 9 ÷ 7.2664 ≈ 1.2386, which is the same as the monthly Sharpe × √12: 0.3575 × 3.4641 ≈ 1.2386.
  • Annual Sortino ≈ 0.7137 × 3.4641 ≈ 2.4725.

Why √12 and not 12? Averages grow with time, but if months are independent, the variance of a sum grows with the number of months, so the standard deviation grows with its square root. The rule therefore assumes monthly returns are independent (this month tells you nothing about next month) and come from the same distribution each month. It also ignores compounding. If returns trend or cluster (calm stretches then turbulent ones), the √12 figure can understate or overstate true annual risk. Use √52 for weekly data and roughly √252 for daily trading data, with the same caveats.

Worked example 3: when Sharpe and Sortino disagree

Section titled “Worked example 3: when Sharpe and Sortino disagree”

Portfolio B’s monthly returns are: 4, 0, 0, 4, 0, −2. It has the same 6 ÷ 6 = 1% mean as A and the same 0.75% excess return.

Sharpe for B. Sum check: 4 + 0 + 0 + 4 + 0 − 2 = 6. Deviations from 1 are 3, −1, −1, 3, −1, −3. Squares: 3² + (−1)² + (−1)² + 3² + (−1)² + (−3)² = 30. Variance: 30 ÷ 5 = 6. Standard deviation: 6^0.5 ≈ 2.4495. Sharpe: 0.75 ÷ 2.4495 ≈ 0.3062.

Sortino for B. Three months of 0 each fall 0.25 short; the −2 month falls 2.25 short. Squares: 3 × 0.25² + 2.25² = 0.1875 + 5.0625 = 5.25. Divide by six: 5.25 ÷ 6 = 0.875. Downside deviation: 0.875^0.5 ≈ 0.9354. Sortino: 0.75 ÷ 0.9354 ≈ 0.8018.

Measure (monthly unless stated) Portfolio A Portfolio B Higher
Mean return 1% 1% Tie
Standard deviation ≈ 2.0976 ≈ 2.4495 A is steadier
Downside deviation ≈ 1.0508 ≈ 0.9354 B has milder shortfalls
Sharpe ratio ≈ 0.3575 ≈ 0.3062 A
Sortino ratio ≈ 0.7137 ≈ 0.8018 B
Sharpe, annualised ≈ 1.2386 ≈ 1.0607 A
Sortino, annualised ≈ 2.4725 ≈ 2.7775 B

The rankings flip. B’s big swings were mostly upward (two +4% months), which Sharpe counts as risk and Sortino does not. A’s worst month (−2) matches B’s, but A also had a −1. Which ratio “wins” depends on what you mean by risk: total unpredictability, or the chance of falling short. Neither answer tells you which portfolio will do better next. With six months of data, the difference is also well within what chance alone could produce.

  • They look backward. Every input is a past return. As the SEC publication Invest Wisely puts it, “The past success of a particular investment is no guarantee of future performance.”
  • Non-normal returns. Standard deviation describes a bell-shaped spread well. Strategies that collect small steady gains and occasionally suffer a large loss (for example, selling options) can show a high Sharpe ratio right up to the bad month. Check the worst month and the drawdown, not just the ratio.
  • Period choice. A window that starts after a crash and ends before the next one flatters almost anything. Compare portfolios over the same dates, frequency and risk-free rate.
  • Leverage. Borrowing to double a position roughly doubles both the excess return and the standard deviation, so the ratio barely moves: (2 × 0.75) ÷ (2 × 2.0976) is still about 0.3575 (ignoring borrowing costs). A good Sharpe ratio says nothing about whether the size of the losses is bearable.
  • Survivorship. Lists of funds with great ratios usually leave out funds that closed after bad runs, so the survivors look better than the average manager really was.

Work it with a pencil or a spreadsheet first, then check your answers with the Sharpe and Sortino calculator. Take a hypothetical five-month series: 1, 2, −1, 0, 3.

  1. Mean: (1 + 2 − 1 + 0 + 3) ÷ 5 = 1.
  2. Sample variance: (0² + 1² + (−2)² + (−1)² + 2²) ÷ 4 = 2.5; standard deviation 2.5^0.5 ≈ 1.5811. In a spreadsheet, =STDEV.S(A1:A5) should match.
  3. Sharpe with the 0.25 monthly rate: 0.75 ÷ 1.5811 ≈ 0.4743.
  4. Downside: the 0 month falls 0.25 short and the −1 month 1.25 short, so ((−0.25)² + (−1.25)²) ÷ 5 = 0.325. Sortino: 0.75 ÷ 0.325^0.5 ≈ 1.3156.

What to notice: change the 3 to a 6 and recompute. The mean rises, the standard deviation also rises, but the downside deviation does not move at all.

  1. Mixing periods. Dividing an annual excess return by a monthly standard deviation produces a meaningless number. Put both on the same period, then annualise.
  2. Comparing ratios built with different conventions. Sample vs population, target 0 vs risk-free, and daily vs monthly data all change the result. Only compare like with like.
  3. Treating a higher ratio as a forecast. It summarises one past window. A different window can rank the same portfolios the other way.
  4. Forgetting that the ratio can be negative. If the mean return is below the risk-free rate, the ratio is negative, and comparing two negative ratios is tricky: more volatility can make a negative ratio look “better”.
  5. Reading Sortino with too few losing months. With one or two shortfalls, downside deviation rests on almost no data and the ratio can look spectacular by accident.

You will see these ratios on fund fact sheets, in brokerage portfolio reports and in research screens, usually annualised and labelled with a period such as “3 years”. Look for the fine print: the data frequency, the risk-free proxy (often a short-term Treasury bill index) and whether the figure is before or after fees. The SEC’s Beginners’ Guide to Asset Allocation reminds readers that “risk and reward are inextricably entwined”; these ratios are one way to put both on a single scale, not a way to remove the risk.

  1. A portfolio returned 8% over a year with a 2% risk-free rate and a 12% annual standard deviation. What is its Sharpe ratio?
Answer to question 1

(8 − 2) ÷ 12 = 0.5.

  1. A monthly Sharpe ratio is 0.2. Using the √12 rule, what is the annual figure, and what does that rule assume?
Answer to question 2

0.2 × 12^0.5 ≈ 0.6928. It assumes monthly returns are independent and identically distributed, and it ignores compounding.

  1. Using a 0 target and dividing by all four months, what is the downside deviation of the returns 2, −1, 5, −3?
Answer to question 3

Only −1 and −3 fall below 0: 1² + 3² = 10, then 10 ÷ 4 = 2.5, and 2.5^0.5 ≈ 1.5811.

  1. Portfolio B has a lower Sharpe ratio than A but a higher Sortino ratio. What does that suggest?
Answer to question 4

More of B’s variability comes from gains rather than shortfalls. Whether that makes B “better” depends on whether you count upside swings as risk, and six months is too short to be sure.

  • Sharpe = (mean return − risk-free rate) ÷ standard deviation; Sortino swaps in downside deviation.
  • State your conventions (sample or population, target, period) and keep them identical when comparing.
  • Annualise with √12 for monthly data, remembering it assumes independent months.
  • Sharpe and Sortino can rank the same portfolios differently, as A and B show.
  • Both are backward-looking summaries that leverage, short windows, non-normal returns and survivorship can distort. All figures here are hypothetical.

For educational purposes only. This material does not provide investment advice.