The Sortino ratio is a risk-adjusted performance measure that tells you how much excess return an investment generates per unit of “bad” volatility. Unlike the Sharpe ratio, which uses total standard deviation (both upside and downside volatility), the Sortino ratio uses only downside deviation — the variability of returns that fall below a chosen threshold (often the risk-free rate or a minimum acceptable return). It was named after Frank A. Sortino and is widely used by investors and portfolio managers who want to reward upside volatility while penalizing only downside moves.
Quick definition and formula
Sortino ratio = (Rp − Rtarget) / σd
– Rp = actual or expected portfolio return (annualized if comparing annualized figures)
– Rtarget = target return (commonly the risk-free rate or a minimum acceptable return, MAR)
– σd = downside deviation (standard deviation of returns that fall below Rtarget)
Why use the Sortino ratio?
– Focus on “bad” risk: Investors care more about losses than variability that comes from unexpectedly large gains. Sortino isolates downside risk.
– Better for asymmetric/ skewed return distributions: When returns are not symmetrically distributed, Sharpe’s total standard deviation may over-penalize investments that have frequent positive outliers.
– More intuitive for many investment decisions: A higher Sortino ratio means more excess return per unit of downside risk.
How to interpret the Sortino ratio
– Higher is better. A larger Sortino ratio means an investment is generating more excess return for each unit of downside volatility.
– Rule-of-thumb ranges (context-dependent):
• 1.0 : good
• > 2.0 : excellent
These thresholds are only rough guides — compare like-to-like funds, return horizons, and investment styles.
Detailed steps to calculate the Sortino ratio (practical)
1. Select your return series (periodicity must match later steps)
• Use monthly returns for long-term mutual funds, daily returns for short-term strategies, etc.
• Ensure consistent frequency for Rp, Rtarget, and σd.
2. Choose the target return (Rtarget)
• Common choices: risk-free rate matching the return period (e.g., 1‑month T-bill for monthly returns, a 3‑month T-bill, or an annualized Treasury yield for yearly data).
• Alternatively use a Minimum Acceptable Return (MAR) such as 0% or an investor’s required hurdle.
3. Compute excess returns relative to that target
• For each period i: excess_i = Ri − Rtarget_period
• If Rtarget is annual and returns are monthly, convert Rtarget to monthly or annualize returns consistently.
4. Calculate downside deviation (σd)
• For each period where Ri < Rtarget, compute (Ri − Rtarget)^2; for Ri ≥ Rtarget use 0.
• Sum those squared shortfalls, divide by N (the number of observations or sometimes by the number of periods — using N gives the population downside deviation), then take the square root:
σd = sqrt( (1/N) * Σ [min(0, Ri − Rtarget)]^2 )
• Note: some implementations use the count of negative observations in the denominator; be consistent and state your convention.
5. Compute Rp − Rtarget
• Use the average (or annualized) portfolio return minus Rtarget (in the same periodic basis as σd).
6. Compute Sortino ratio
• Sortino = (Rp − Rtarget) / σd
Worked numerical example
Assume:
– Annualized return Rp = 12%
– Risk-free rate (Rtarget) = 2.5%
– Annual downside deviation σd = 10%
Sortino = (0.12 − 0.025) / 0.10 = 0.095 / 0.10 = 0.95
Compare with another fund:
– Rp = 10%, σd = 7% → Sortino = (0.10 − 0.025) / 0.07 = 0.075 / 0.07 ≈ 1.07
Interpretation: Although the first fund had higher raw return (12% vs 10%), the second fund produced higher risk‑adjusted returns when considering only downside volatility — therefore it may be preferable for downside-sensitive investors.
Practical implementation: Excel and Python examples
Excel (monthly returns in range A2:A61, target in B1 as monthly Rtarget):
– Downside deviation (array formula): =SQRT(AVERAGE(IF(A2:A61<$B$1,(A2:A61-$B$1)^2,0)))
(Enter as an array formula in older Excel versions or use dynamic array formulas.)
– Average excess return (monthly): =AVERAGE(A2:A61) − $B$1
– Annualize returns if needed (convert monthly to annual: (1+avg_monthly)^12 − 1). Be consistent.
Python / pandas (monthly returns as a Series `r`, target `target` as decimal monthly):
“`
import numpy as np
excess = r – target
downside_sq = np.where(excess < 0, excess**2, 0)
downside_deviation = np.sqrt(downside_sq.mean()) # population version
avg_return = r.mean()
sortino = (avg_return – target) / downside_deviation
# Annualize if needed.
“`
Choosing the risk-free rate and time horizon
– Match periodicity: use a risk-free rate that matches the return frequency (monthly T-bill yield for monthly returns, etc.).
– If you use a long-duration Treasury yield as Rtarget while computing daily downside deviations, you introduce inconsistency — convert to the same basis (e.g., daily risk-free rate).
– Alternatively, set Rtarget to a minimum acceptable return (MAR) which better represents investor goals (e.g., 0%, inflation+2%, or a fixed hurdle).
Limitations and cautions
– Sensitive to choice of Rtarget/MAR: changing the target changes both excess returns and which returns count toward downside deviation.
– Uses second moment of downside only: it still assumes downside variability is captured by variance; it won’t fully capture tail risk beyond variance (e.g., extreme fat tails).
– Data frequency and sample size matter: short samples or mismatched periodicity can give misleading values.
– Past performance bias: Sortino relies on historical returns if used for expected return — it doesn’t guarantee future performance.
– Comparison caveats: compare only similar strategies (same return horizon, frequency, and MAR) and be cautious when comparing across asset classes with different return distributions.
– Not a standalone decision tool: use alongside Sharpe, alpha, drawdown statistics, VaR, Omega ratio, and qualitative analysis.
When to use Sortino versus Sharpe
– Use Sortino if your priority is measuring downside exposure only — for loss-averse investors or strategies with large positive skew (hedge funds, option sellers, etc.).
– Use Sharpe to evaluate total volatility when upside variability is not differentiated from downside variability or when you need to compare across strategies where total volatility is relevant.
– Often best practice: report both Sharpe and Sortino, plus drawdown statistics.
Practical checklist for investors and portfolio managers
1. Define objective: Choose MAR (risk-free rate vs investor target return).
2. Choose data frequency and time window consistent with goals (e.g., 3–5 years of monthly returns).
3. Compute downside deviation consistently (document whether you divide by N or by the count of negative periods).
4. Annualize returns and downside deviation if you need annual Sortino ratios (convert both consistently).
5. Compare like-to-like investments and use Sortino along with other risk measures (max drawdown, Sharpe, beta).
6. Perform sensitivity checks: test alternative MARs, time windows, and sampling frequencies.
7. Monitor rolling Sortino (e.g., 12‑month rolling) to spot changing risk-adjusted performance over time.
8. Supplement with stress-testing and tail-risk metrics if concerned about extreme events.
Bottom line
The Sortino ratio is a practical, investor-friendly metric that measures reward per unit of downside risk only, making it attractive to loss-averse decision-makers. It improves on Sharpe by ignoring upside volatility, but it is sensitive to the choice of target return and still has limitations around tail risk and sample dependency. Use it alongside other performance and risk measures, be consistent with return periodicity and target selection, and document calculation choices when comparing investments.
Sources and further reading
– Investopedia, “Sortino Ratio” (Michela Buttignol)
– CME Group, “Sortino: A ‘Sharper’ Ratio”
– CFI Education, “Downside Risk” and “Risk-Free Rate”
– Indeed, “What Is Expected Return? (Plus How to Calculate It)”
(1) compute the Sortino ratio for your data if you paste returns, 2) provide an Excel workbook template, or 3) give a short script that computes rolling Sortino ratios for a CSV of returns.)
(continuation)
ADDITIONAL SECTIONS, EXAMPLES, AND PRACTICAL STEPS
LIMITATIONS AND POTENTIAL PITFALLS
– Choice of target/threshold (T) matters. The Sortino ratio depends on what you define as “downside.” Using the risk-free rate, zero, or a Minimum Acceptable Return (MAR) will produce different Sortino values and can change ranking among investments.
– Different downside-deviation conventions. Some practitioners divide the summed squared shortfalls by the total number of observations (N); others divide by the number of returns below the target (n−). This choice materially affects the computed downside deviation and therefore the Sortino ratio.
– Sensitivity to sample size and return frequency. With small samples or infrequent observations, the downside estimate can be noisy.
– Ignores higher moments beyond downside variance. The Sortino ratio does not capture skewness or kurtosis, which can matter if return distributions are non‑normal.
– Doesn’t capture tail dependence or extreme drawdowns explicitly. Two investments can have similar Sortino ratios but different extreme-loss behavior.
– Not a complete performance measure. Use it in conjunction with other metrics (Sharpe ratio, maximum drawdown, alpha, information ratio) and qualitative considerations (strategy, liquidity, fees).
– Survivorship and look-ahead bias may distort historical Sortino calculations if data are not clean.
HOW TO CALCULATE DOWNSIDE DEVIATION (σd)
Definition (intuitive): downside deviation measures variability of returns that fall below a specified target T (often the risk-free rate or a Minimum Acceptable Return).
Two common formulas (for a sample of returns Ri, i = 1..N):
– Convention A (divide by total observations):
1. For each period compute shortfall: shortfall_i = max(0, T − Ri).
2. Downside variance = (1/N) * Σ shortfall_i^2.
3. Downside deviation σd = sqrt(downside variance).
– Convention B (divide by number of failures):
1. Only include periods where Ri < T. Let n− be the count of those periods.
2. Downside variance = (1/n−) * Σ (T − Ri)^2 over Ri < T.
3. σd = sqrt(downside variance).
Be explicit about which convention you use when comparing funds or backtests.
ANNUALIZATION RULES
– If returns are periodic (e.g., monthly), annualize consistently:
• Annualized return = average periodic return × periods per year (or use geometric return for longer horizons).
• Annualized downside deviation ≈ σd_periodic × sqrt(periods per year).
• Use a risk-free rate expressed on the same periodic basis (or convert consistently).
– Do not mix annual returns with non-annual downside deviation without converting.
PRACTICAL STEP-BY-STEP (MANUAL / EXCEL)
1. Choose the target T (risk-free, MAR, zero).
2. Gather periodic returns (monthly, weekly, daily). Ensure consistent frequency.
3. Compute shortfalls: Shortfall = IF(return < T, T − return, 0).
4. Square shortfalls and average:
• Using total N: DownsVar = AVERAGE(squared_shortfalls)
• Using only failures: DownsVar = SUM(squared_shortfalls) / COUNTIF(returns, "<"&T)
5. σd = SQRT(DownsVar).
6. Compute excess return = (average return on same periodic basis converted as needed) − rf (on same basis).
7. Sortino = ExcessReturn / σd.
Excel examples:
– If returns in A2:A7 and target in B1:
• Shortfall squares: (B1-A2)^2 when A2<B1, otherwise 0.
• Downside dev (divide by N): =SQRT( SUMPRODUCT( (B1-A2:A7)^2 * (A2:A7 < B1) ) / COUNT(A2:A7) )
• Downside dev (divide by failures): =SQRT( SUMPRODUCT( (B1-A2:A7)^2 * (A2:A7 < B1) ) / COUNTIF(A2:A7, "<"&B1) )
• Sortino: = (AVERAGE(A2:A7)*periodsPerYear – rfAnnual) / (σd_annualized)
PRACTICAL STEP-BY-STEP (PYTHON / PANDAS)
– Example code sketch:
import numpy as np
import pandas as pd
returns = pd.Series([…]) # periodic returns, e.g., monthly decimals
T = 0.005 # target per period (e.g., 0.5% monthly)
shortfalls = np.maximum(0, T – returns)
# Convention A: divide by total N
downside_dev = np.sqrt(np.mean(shortfalls**2))
# Convention B: divide by number of shortfalls
failures = (returns annual)
downside_dev_annual = downside_dev * np.sqrt(12)
annual_return = returns.mean() * 12
rf_annual = 0.06 # 6%
sortino = (annual_return – rf_annual) / downside_dev_annual
EXAMPLE — SMALL TIME-SERIES (STEP-BY-STEP NUMERIC)
Monthly returns (6 months): 2%, −1%, 4%, −3%, 0.5%, 3% (expressed as decimals: 0.02, −0.01, 0.04, −0.03, 0.005, 0.03)
Target T = 0.005 (0.5% monthly) — equivalent to 6% annual.
1. Shortfalls (T − Ri) only for Ri T → 0 shortfall
• −0.01 → shortfall = 0.005 − (−0.01) = 0.015
• 0.04 > T → 0
• −0.03 → shortfall = 0.005 − (−0.03) = 0.035
• 0.005 = T → 0
• 0.03 > T → 0
2. Squared shortfalls: 0.015^2 = 0.000225; 0.035^2 = 0.001225; sum = 0.00145.
3. Downside variance (divide by N = 6): 0.00145 / 6 = 0.000241667 → σd_monthly = sqrt = 0.015539 = 1.5539% monthly.
• Annualize: σd_annual ≈ 0.015539 * sqrt(12) ≈ 5.394% annual.
4. Average monthly return = (0.02 − 0.01 + 0.04 − 0.03 + 0.005 + 0.03) / 6 = 0.0091667 ≈ 0.91667% monthly → annual ≈ 11.0%.
5. Excess annual return = 11.0% − 6.0% = 5.0%.
6. Sortino ≈ 5.0% / 5.394% ≈ 0.93.
If instead you divide by the number of failures (n− = 2):
– Downside variance = 0.00145 / 2 = 0.000725 → σd_monthly = sqrt = 0.02695 (2.695%) monthly.
– Annualized σd ≈ 0.02695 * sqrt(12) ≈ 9.33% annual.
– Sortino ≈ 5.0% / 9.33% ≈ 0.54.
This illustrates how the denominator choice changes the absolute Sortino value—be consistent and report the convention used.
INTERPRETING SORTINO VALUES
– Higher is better: more excess return per unit of downside risk.
– No universal cutoff. Loose guidelines sometimes used:
• >2.0: excellent
• 1.0–2.0: good
• 0.0–1.0: acceptable to poor (depends on asset class)
• <0: returns below the target (bad).
– Compare within asset classes and peers; absolute numbers depend on period, target, and convention.
SORTINO VS SHARPE (KEY DIFFERENCES)
– Sharpe uses total standard deviation of returns (penalizes both upside and downside volatility).
– Sortino uses only downside deviation (penalizes only returns below a target).
– Choose Sharpe if you are concerned about total volatility. Choose Sortino when downside risk is the main concern (many investors focus on downside).
– Both measures assume the chosen statistical risk measure is meaningful for your goals; neither tells the whole story.
USING SORTINO IN PORTFOLIO CONSTRUCTION
– Can be an objective in optimization: maximize Sortino instead of Sharpe (nonlinear objective because downside deviation depends on target and downside only).
– Use alongside constraints (turnover, exposure, max drawdown) and other performance metrics.
– When backtesting, ensure out-of-sample testing and avoid look-ahead/survivorship bias.
PRACTICAL USE CASES
– Income investors who prioritize protecting capital vs. upside may prefer Sortino.
– Fund comparisons within an objective: rank funds targeting similar strategies with identical target definitions.
– Risk-managed strategies where downside control is explicitly targeted (e.g., tail hedged, options overlays).
ADDITIONAL EXAMPLES (QUICK)
– Two funds with same average return: The one with fewer or smaller shortfalls (lower σd) will have a higher Sortino.
– A strategy with lumpy large losses and long stretches of small positive returns can have a comparable average but a much worse Sortino than a smoother but lower-mean strategy.
PRACTICAL ADVICE FOR INVESTORS AND ANALYSTS
– Be explicit about target choice (risk-free, MAR, zero) and denominator convention used for σd.
– Use consistent periodicity; annualize consistently for comparison.
– Combine Sortino with Sharpe, maximum drawdown, Calmar ratio, and qualitative analysis.
– Check sensitivity: recompute Sortino with different targets (e.g., rf, 0, investor hurdle) to see robustness.
– Watch for small-sample artifacts and extreme returns—robust statistics or bootstrapping can help.
REFERENCES AND SOURCES
– Investopedia. “Sortino Ratio.” (Michela Buttignol) — explanation, formula, and example.
– CME Group. “Sortino: A ‘Sharper’ Ratio.” (discussion of concept and practical notes).
– CFI Education. “Downside Risk”; “Risk-Free Rate.” (background definitions).
– Indeed. “What Is Expected Return?” (background).
(When preparing formal reports cite the full articles and dates accessed.)
CONCLUSION / SUMMARY
The Sortino ratio refines the Sharpe approach by focusing only on negative volatility relative to a chosen target (most commonly the risk-free rate or an investor’s minimum acceptable return). It is useful when downside risk—losses below a threshold—is the primary concern. While a higher Sortino indicates more attractive risk-adjusted performance on the downside, the metric is sensitive to the definition of the target and the method used to compute downside deviation. Use the Sortino ratio alongside other performance and risk measures, be explicit about calculation conventions, and test sensitivity to ensure robust conclusions.