Portfolio and risk · Measuring return and risk
Sortino ratio: judging a return by its downside only
How downside deviation and the Sortino ratio are calculated, a worked example on SPY, and why for most funds it is close to the Sharpe ratio times the square root of two.
On this page
The Sortino ratio is a return above a target, divided by the downside deviation: a measure of risk that counts only the periods that fell short of the target. It is the Sharpe ratio with one change. Where Sharpe divides by the standard deviation of all returns, Sortino divides by the spread of the bad ones.
The change answers an obvious complaint. Standard deviation treats a month of +10% and a month of −10% as the same amount of risk, and no investor does. Frank Sortino's measure, developed in the early 1990s, leaves gains out of the denominator. This lesson shows the calculation, applies it to nine portfolios, and reports a result that is less exciting than the theory: for ordinary funds the two ratios almost always agree.
The idea
Line up every month's return against a target. Here the target is the Treasury bill return: did the investment beat a risk-free asset this month or not? Months above the target are successes of varying size. Months below it are shortfalls.
The standard deviation measures the spread of all of them. The downside deviation looks only at the shortfalls, the red bars in the figure, and asks how large they typically were.
Of SPY's 262 months, 96 fell short of Treasury bills. The other 166 contribute nothing to downside risk, however large they were. A fund with big gains and small losses keeps a small denominator and gets a high ratio; under the Sharpe ratio its big gains would count against it.
How it works
Two details cause most of the mistakes made with this measure.
- Divide by all the months. A month above the target has a shortfall of zero, and those zeros are part of the average. Dividing only by the number of bad months inflates the downside deviation and makes results incomparable with anyone else's.
- Name the target. Sortino called it the minimum acceptable return. Common choices are the risk-free rate, zero, or a required return such as 5% a year. A higher target produces more shortfalls and a lower ratio. This page uses the Treasury bill return throughout, so that the numbers line up with the Sharpe ratios in the previous lesson.
A calculation by hand
Six months of returns above the target, illustrative numbers: +3%, −2%, +5%, −4%, +1%, +3%.
- The average is (3 − 2 + 5 − 4 + 1 + 3) ÷ 6 = 1.0%.
- The shortfalls are 0, −2, 0, −4, 0, 0.
- Squared and summed: 4 + 16 = 20.
- Divided by all six months: 3.33. The square root is 1.83. Downside deviation is 1.83% a month.
- Monthly Sortino ratio: 1.0 ÷ 1.83 = 0.55. Annualized: 0.55 × √12 = 1.90.
For comparison, the standard deviation of the same six numbers is 3.41%, so the monthly Sharpe ratio is 0.29 and the annual one 1.02. The Sortino ratio is almost twice as large because the two gains of 3% and the gain of 5% widened the standard deviation and left the downside deviation alone.
A worked example on SPY
The data: SPY's 262 monthly total returns, January 2005 to October 2026, each less that month's Treasury bill return.
- Average monthly excess return: 0.797%.
- Months with a shortfall: 96. The sum of their squared shortfalls is 0.2109 (in decimal terms).
- Divided by 262 months and square-rooted: a monthly downside deviation of 2.84%.
- Annualized, 2.84% × √12 = 9.8%, against a standard deviation of 14.8%.
- Sortino ratio: 9.56% ÷ 9.83% = 0.97. The Sharpe ratio over the same months is 0.65.
Reading it on real funds
| Portfolio | Volatility | Downside deviation | Sharpe | Sortino | Sortino ÷ Sharpe |
|---|---|---|---|---|---|
| QQQ | 18.4% | 11.6% | 0.75 | 1.19 | 1.59 |
| 60% SPY, 40% AGG | 9.4% | 6.3% | 0.66 | 0.99 | 1.50 |
| SPY | 14.8% | 9.8% | 0.65 | 0.97 | 1.50 |
| GLD | 17.4% | 10.6% | 0.54 | 0.88 | 1.64 |
| IWM | 19.8% | 13.5% | 0.42 | 0.61 | 1.47 |
| EFA | 16.7% | 11.7% | 0.33 | 0.47 | 1.43 |
| VNQ | 21.5% | 15.2% | 0.33 | 0.47 | 1.42 |
| AGG | 4.5% | 3.0% | 0.27 | 0.39 | 1.48 |
| TLT | 13.6% | 8.8% | 0.15 | 0.24 | 1.55 |
The Sortino ratio is larger than the Sharpe ratio for every fund, and that is arithmetic, not good news. If returns were perfectly symmetric around the target, half of the squared distance would lie on each side, the downside deviation would be the standard deviation divided by √2, and the Sortino ratio would be the Sharpe ratio times √2, or 1.41. With a positive average return, fewer than half the months are shortfalls and the multiple is a little higher. A Sortino ratio should be compared with other Sortino ratios, never with a Sharpe ratio.
The informative column is the last one. It runs from 1.42 to 1.64, a narrow band, and the ranking of the nine portfolios is identical under both measures. The fund that gains most is gold: its multiple of 1.64 says that more of its volatility came from sharp rises than is true of stocks. Real estate and foreign stocks, at 1.42 and 1.43, are almost exactly symmetric: their volatility was as much downside as upside.
So for diversified funds built from stocks and bonds, the Sortino ratio tells nearly the same story as the Sharpe ratio. The measure matters for return patterns that are strongly lopsided:
- Strategies with capped gains and open losses, such as selling options for income. Their standard deviation looks modest, their rare losses are large, and the Sortino ratio is less generous to them than the Sharpe ratio.
- Strategies with capped losses and open gains, such as trend following with stops or holding options for protection. Sharpe penalizes their occasional large gains; Sortino does not.
Where it fails or misleads
It needs many shortfalls to mean anything
Only the bad months inform the denominator. Over three years of monthly data a stock fund may have 12 shortfalls; the estimate rests on those 12 numbers. A period with no real decline produces a tiny downside deviation and an enormous ratio that describes the absence of a test, not the presence of skill. A fund launched after 2009 and measured before 2020, a stretch with few deep monthly losses, could show a very large Sortino ratio for this reason.
It cannot see a loss that has not happened
Like every measure built from past returns, the Sortino ratio reflects only the shortfalls in the sample. A strategy that loses rarely and heavily looks excellent until the first such loss arrives. It is a better description of observed downside than the Sharpe ratio; it is no better at anticipating unobserved downside. See Fat tails: why extreme market moves are more common than models say.
The target changes the answer
With a target of zero instead of the Treasury bill return, fewer months count as shortfalls and the numerator is larger, so every ratio rises. Two published Sortino ratios are comparable only if they use the same target and the same data frequency.
Negative values do not rank
TAIL, a fund that holds Treasuries and out-of-the-money put options on the S&P 500 as protection, compounded at −7.4% a year from June 2017 to October 2026 and has a Sortino ratio of −1.12. As with the Sharpe ratio, a negative value says the fund lost to Treasury bills and nothing useful about how much risk it took. A fund like that is held for what it does in a crash, which no stand-alone ratio measures; Tail-risk protection: insuring a portfolio against a crash looks at it in a portfolio.
It still ignores the order of losses
Downside deviation treats each shortfall separately. Ten losing months in a row and ten losing months scattered over a decade give the same number. Maximum drawdown (MDD): the worst fall from a peak, and the way back is the measure that cares about sequence.
Compared with its neighbours
| Measure | Numerator | Denominator | Sensitive to |
|---|---|---|---|
| Sharpe ratio | Return over the risk-free rate | Standard deviation of all returns | Any volatility, up or down |
| Sortino ratio | Return over a chosen target | Downside deviation | Size and frequency of shortfalls |
| Calmar ratio | Compound annual return | Maximum drawdown | The single worst cumulative loss |
| Value at risk | (not a ratio) | A loss threshold at a probability | How bad a bad period typically is |
How Finabase does it
Finabase does not calculate a Sortino ratio or a downside deviation, for symbols or for portfolios. Its return figures are price changes over fixed periods, shown on symbol pages, in the screener and in comparison tables, and its portfolio report lists profit and return by period without any risk measure.
The calculation is short in a spreadsheet. Put the monthly returns of a portfolio in column
A and the monthly target in column B. In column C compute =MIN(A2-B2, 0). The downside
deviation is =SQRT(SUMSQ(C:C)/COUNT(C:C)), and the ratio is the average of A minus B,
divided by that, times SQRT(12). Use at least five years of months and a period that
includes a real decline.
Frequently asked questions
Is the Sortino ratio better than the Sharpe ratio?
It is better matched to how investors feel about risk and it treats lopsided strategies more fairly. It is also noisier and less standardized. For a plain stock and bond portfolio the two rank investments the same way.
What is a good Sortino ratio?
Using the Treasury bill rate as the target, SPY scored 0.97 over 2005 to 2026 and a 60/40 mix 0.99. Because the ratio runs about 1.4 to 1.6 times the Sharpe ratio for ordinary funds, "above 1" for Sortino is roughly what "above 0.65" is for Sharpe.
What target return should I use?
The risk-free rate if the goal is to compare with Sharpe ratios; zero if the question is simply "how bad are the losing months"; a required return if the portfolio has to fund a specific spending rate. State which one was used.
Why is my downside deviation different from someone else's?
The usual causes are the divisor (all periods against only losing periods), the target, and whether the result was annualized with √12. Data providers do not all make the same choices, so check the definitions before comparing two published figures.
Summary
The Sortino ratio replaces the standard deviation in the Sharpe ratio with the downside deviation, the root-mean-square of shortfalls below a target, averaged over all periods. SPY scored 0.97 from 2005 to 2026 against a Sharpe ratio of 0.65, and nearly every ordinary fund shows a similar multiple of about 1.5, close to the √2 expected of symmetric returns. The ratio changes rankings only for lopsided return patterns, and for those it is the fairer measure. It remains a summary of losses already seen, it depends on the target chosen, and with few losing periods in the sample it can be very large for no good reason.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
A fund’s returns above its target over four months are +6%, −2%, +4% and 0%. What is its monthly downside deviation?
- 2%, because only one month fell short
- 1%, the square root of (4 ÷ 4)
- 3.7%, the standard deviation of the four numbers
Show the answer
B. 1%, the square root of (4 ÷ 4) Only the −2% month is a shortfall. Its square is 4, the sum is divided by all four months, and the square root of 1 is 1%. Dividing by the one losing month alone would give 2% and overstate the risk.
A fund reports a Sharpe ratio of 0.6 and a Sortino ratio of 0.85. What does the gap tell you?
- The fund controls its downside unusually well
- Very little: symmetric returns give a Sortino ratio about 1.4 times the Sharpe ratio
- The Sortino ratio was calculated incorrectly
Show the answer
B. Very little: symmetric returns give a Sortino ratio about 1.4 times the Sharpe ratio For a symmetric return pattern the downside deviation is the standard deviation divided by √2, so the Sortino ratio is roughly 1.41 times the Sharpe ratio. 0.85 ÷ 0.6 = 1.42 is exactly what an ordinary fund shows.
Which strategy is the Sortino ratio likely to rate much more favourably than the Sharpe ratio does?
- One with many small losses and occasional very large gains
- One with many small gains and occasional very large losses
- An index fund
Show the answer
A. One with many small losses and occasional very large gains Large gains raise the standard deviation and leave the downside deviation unchanged, so the Sharpe ratio penalizes them and the Sortino ratio does not. The opposite pattern, rare large losses, is treated more harshly by Sortino.
A fund launched four years ago, during a steady rise, shows a Sortino ratio of 3. What is the sensible conclusion?
- It has very little downside risk
- The sample contains too few shortfalls to measure its downside
- It will outperform funds with lower ratios
Show the answer
B. The sample contains too few shortfalls to measure its downside The denominator is built only from losing periods. With almost none in the sample the downside deviation is tiny and the ratio is large, which describes the period, not the fund.
Sources and further reading
- Frank A. Sortino and Robert van der Meer, “Downside Risk,” Journal of Portfolio Management, vol. 17, no. 4 (1991)
- Frank A. Sortino and Lee N. Price, “Performance Measurement in a Downside Risk Framework,” Journal of Investing, vol. 3, no. 3 (1994)
- Board of Governors of the Federal Reserve System — 3-month Treasury bill secondary market rate (TB3MS), via FRED
- Cambria Investments, Cambria Tail Risk ETF (TAIL) fund page — holds U.S. Treasuries and out-of-the-money S&P 500 put options; inception April 6, 2017
- Finabase chart data — monthly bars and dividends of SPY, QQQ, EFA, IWM, AGG, TLT, GLD, VNQ, USMV, SPLV and TAIL through October 2026; ratios computed by Finabase
