Sortino Ratio Calculator

Measure return against downside deviation only, ignoring the upside volatility the Sharpe ratio penalises.

The Sortino ratio begins from an objection to the Sharpe ratio: that an unexpectedly good month is being counted as risk. This Sortino ratio calculator keeps the same shape — a return divided by a dispersion — but builds the dispersion only from periods that finished below a target you choose. Returns above the target move the numerator and leave the denominator exactly where it was, so only downside risk ever reaches the divisor.

Formula: sortino = (mean return - target) / downside deviation below the target

Your figures

Return series

One row per period, oldest first, as a percentage of the balance at the start of that period. Enter losses as negative figures. Any period length works, as long as every row is the same length and the count below matches it.

Return 1

Return 2

Return 3

Return 4

Return 5

Return 6

Target return per period

The minimum acceptable return for one period, on the same basis as the series above. Zero treats any loss as downside; the periodic risk-free rate is the other common choice.

Periods in a year

The cadence of the returns above. It is used only to annualise the ratio.

Custom periods in a year

Only used when the frequency is set to Custom.

Results

Annualised Sortino ratio

3.41

0.983 per period, multiplied by the square root of 12

Sortino ratio per periodReturn above the target divided by the downside deviation, both measured over a single period.: 0.983
Downside deviation per periodThe root mean square shortfall below the target, divided by every period counted rather than by the losing ones alone.: 1.153%
Periods below the targetThe shortfalls come from these periods, but the average is taken over all of them. Dividing by the losing count instead would give a larger deviation and a smaller ratio from the same record.: 2 of 6
Mean return per periodThe arithmetic mean of every return entered, winners and losers alike.: +1.133%
Target return per periodEchoed back as entered. A period that lands exactly on it has met the requirement, so it counts as neither downside nor shortfall.: 0.0000%
Return above the targetMean return less the target. This is the numerator of the ratio.: +1.133%
Periods countedBlank rows are not counted. Everything above is estimated from this many observations, and a short record moves a long way on one more of them.: 6
Periods in a yearUsed for nothing but the annualisation, which multiplies the periodic ratio by the square root of this figure.: 12

What this means

Of the 6 periods entered, 2 finished below the 0.0000% target. Squaring each of those shortfalls, averaging over all 6 periods rather than over the 2 that fell short, and taking the square root gives a downside deviation of 1.153%. The mean return sits +1.133% from the target, and dividing that by 1.153% gives 0.983 per period, or 3.41 annualised at 12 periods a year. Because only shortfalls enter the denominator, a period that beat the target by a wide margin moves the numerator and leaves the denominator exactly where it was — which is the whole of the difference between this and the Sharpe ratio.

The downside deviation divides the summed shortfalls by every period counted, not by the number of losing ones. That is deliberate: the measure answers how much downside the record produced per period, and a strategy that lost ground in one month out of twelve produced less of it per period than one that lost ground in eleven.
It is also what keeps this consistent with the Sharpe ratio's n − 1. A degree of freedom is spent estimating a mean from the data; none is spent on a target you supplied, so there is none to give back.
The target is a periodic figure. Entering an annual rate beside monthly returns compares the two on different scales and understates the ratio heavily.
The deviation is estimated from the periods that fell short. The fewer of those a record contains, the more one of them moves the figure.
Annualising multiplies the periodic ratio by the square root of the periods in a year, because the mean grows with time while the standard deviation grows with the square root of it. That assumes each period is independent of the one before. A trending or mean-reverting record breaks the assumption, and the annualised figure flatters or punishes it accordingly.
Nothing here is fetched. The returns, the rate and the period count are all as entered, so the calculation works the same on a backtest as on a live account. The risk-free rate a field opens with is a round illustration, not a current market rate — take yours from the instrument you would actually have held.
This describes a record that has already happened. It is a summary of the figures entered, not a statement about what follows them, and no particular level of it carries a meaning beyond the arithmetic that produced it.

The link reopens this calculator with the same numbers. Nothing you enter is sent to a server.

Continue with these figures

About the Sortino Ratio calculation

This downside deviation calculator takes the same return series the Sharpe ratio uses, a target return for one period, and how many periods make a year. It reports the ratio per period and annualised, the target downside deviation, the mean return, the return above the target, and how many of the periods entered actually finished below it. That last figure is shown because the denominator is built from those periods alone and is usually estimated from far fewer observations than the record contains.

One detail of the denominator is misquoted more often than anything else in performance statistics, so this calculator is explicit about it. The squared shortfalls are summed over the losing periods but divided by every period in the record, winners included. Dividing by the count of losing periods instead produces a larger deviation and a much smaller ratio from identical data, and two figures computed those two ways are not comparable.

Sortino Ratio formula

shortfallᵢ = min(0, rᵢ − target)
downside σ = √( Σ shortfallᵢ² ÷ n )
excess     = mean − target

sortino    = excess ÷ downside σ
annualised = sortino × √P

target: The minimum acceptable return for one period, as a percentage, on the same basis as the series.
shortfallᵢ: How far period i fell below the target. Exactly zero for any period that met or beat it.
n: Every period counted, winners included. This is the denominator of the downside deviation.
downside σ: The target downside deviation, or root mean square shortfall: square the misses, average over all periods, take the square root.
P: Periods in a year, used for nothing but the annualisation factor.

Dividing by n rather than by the number of losing periods is deliberate. Sortino and Price defined the measure to answer how much downside a strategy produces per period, and a record that lost ground in one month out of twelve has produced less downside per period than one that lost ground in eleven of them. Averaging over the losing months alone would score those two records identically, which is precisely the distinction the measure exists to draw.

There is a second reason, and it is what keeps this consistent with the n − 1 the Sharpe ratio uses. A degree of freedom is given back when the centre of the distribution was estimated from the data itself. Here the shortfalls are measured against a target you supplied, which cost the data nothing, so there is no degree of freedom to return and n is the whole record.

Shortfalls are squared before they are averaged, which makes one deep miss count for far more than several shallow ones: a single −8% month contributes as much to the sum as sixteen −2% months. Periods that met the target contribute zero rather than a negative amount, so a strong month cannot cancel a weak one inside the denominator the way it does inside a standard deviation. That asymmetry is the whole of the difference between this ratio and the Sharpe ratio.

The target is a figure for one period, never an annual one, so that it and the series are always on the same scale. Entering an annual rate beside monthly returns measures every shortfall against a number roughly twelve times too large: almost every period is suddenly counted as downside, the deviation swells, and the ratio collapses while looking entirely plausible. Annualising the finished ratio uses the same √P factor as the Sharpe ratio, and rests on the same assumption that one period says nothing about the next.

Sortino Ratio: a worked example

The same six monthly returns — +2.4%, −1.1%, +3.2%, +0.8%, −2.6% and +4.1% — scored against a target of 0% for each period, with twelve periods in a year.

Periods below the targettwo of six: −1.1% and −2.6%
Sum of squared shortfalls1.1² + 2.6² = 1.21 + 6.76 = 7.97
Downside deviation√(7.97 ÷ 6) = 1.153% — divided by six, not by two
Return above the target1.1333 − 0 = +1.133%
Sortino ratio per period1.1333 ÷ 1.1525 = 0.983
Annualised0.983 × √12 = 3.41

An annualised Sortino ratio of 3.41, against a Sharpe ratio of 1.08 on the identical six months. Two things separate them: the numerator here is measured against a 0% target rather than against the periodic risk-free rate, and the denominator counts only the two losing months, which were shallower than the swings the standard deviation had to absorb. Dividing 7.97 by those two periods instead of by all six would give a downside deviation of 1.996% and an annualised ratio of 1.97 — the same record, one convention apart.

How to use the Sortino Ratio calculator

Enter the return series exactly as you would for the Sharpe ratio — one row per period, oldest first, losses negative — so that the two ratios are describing the same record.
Set the target as a figure for one period. Zero treats any loss as downside; the periodic risk-free rate is the other common choice, and it makes the numerator match the Sharpe ratio's.
Match the periods in a year to the length of your rows, since that number touches nothing except the annualisation.
Read the periods below the target line in the results. It shows how many observations the denominator was actually built from, which is often only two or three.
Compare the answer with the Sharpe ratio on the same series. What a Sortino vs Sharpe comparison shows is how asymmetric the record was, rather than a second opinion about it.

Limitations of the Sortino Ratio calculator

Every calculation makes assumptions. These are the ones that matter here.

The downside deviation is estimated from the losing periods alone. In the worked example above that is two observations, and a statistic resting on two numbers moves a long way when a third arrives.
Squaring the shortfalls makes the figure acutely sensitive to the single worst period. Remove the deepest loss from a short record and the ratio can roughly double without the strategy having changed at all.
A record with no period below the target has no downside deviation and therefore no ratio. The calculator says so rather than reporting an infinity, but the measure is least informative exactly when a record looks best.
The target is a choice rather than a property of the data. Moving it from 0% to the periodic risk-free rate changes both the numerator and which periods count as downside, so two quoted Sortino ratios are incomparable unless both targets are stated.
Annualising by √P assumes serial independence, exactly as the Sharpe ratio does. A downside deviation is no more immune to a trending record than a standard deviation is.
Nothing here can separate a real edge from a fortunate sample. Over a record of a few dozen periods the ratio is mostly a description of which periods happened to be included in it.

Frequently asked questions about Sortino Ratio

Should the downside deviation be divided by the number of losing periods?

No, and this is where published figures most often diverge. The shortfalls are summed over the losing periods but averaged over every period in the record. Dividing by the losing count gives a larger deviation and a smaller ratio from identical data — 1.97 rather than 3.41 in the example above. When comparing a Sortino ratio from another source, that convention is the first thing worth checking.

Why is the Sortino ratio usually higher than the Sharpe ratio?

Because its denominator ignores upside variability and, on most records, ignores the majority of the observations as well. Whenever the losses are shallower or rarer than the gains, the downside deviation is smaller than the standard deviation, and dividing by a smaller number gives a larger answer. A record whose losses are deeper than its gains reverses that comparison.

What target should be entered?

Zero and the periodic risk-free rate are the two conventional choices, and the field starts at zero rather than assuming either. Zero asks how much the record fell into outright loss; the risk-free rate asks how often it failed to beat cash. Whichever is used has to be quoted alongside the ratio, because it changes both halves of the calculation.

Why was my series refused?

Most often because no period finished below the target, which leaves the denominator empty: a downside risk calculator has nothing to divide by when the record contains no downside. Raising the target or extending the record to include a losing period gives the measure something to work with. The other cause is a series of fewer than two returns, which has no dispersion of any kind to measure.

Does a period that lands exactly on the target count as downside?

No. It met the requirement rather than failing it, so it contributes zero to the sum of squared shortfalls and is not counted among the periods below the target. It still counts towards n, which means it lowers the downside deviation slightly — exactly as a winning period does.

Can the Sortino ratio be negative?

Yes, whenever the mean return falls below the target, and it is reported rather than hidden. As with the Sharpe ratio the ordering of negative values is counter-intuitive: a larger downside deviation pushes a negative ratio closer to zero, so a record that lost ground erratically can produce a less alarming number than one that lost ground steadily.

Last reviewed 29 August 2026. See our methodology and editorial policy.

More performance calculators