Skip to content
Menu
Finabase Learn

Portfolio and risk · Measuring return and risk

Risk-adjusted return compared: Sharpe, Sortino, Treynor, Calmar and M²

What each risk-adjusted measure divides by, how nine portfolios rank on four of them over 2005 to 2026, and how to choose the measure that fits the question.

On this page
  1. The idea
  2. How it works
  3. Using it: which measure for which question
  4. Where it fails or misleads
  5. Compared with its neighbours
  6. How Finabase does it
  7. Frequently asked questions
  8. Summary
  9. Check your understanding

A risk-adjusted return is a return divided by, or reduced for, the risk taken to earn it. The purpose is to compare investments that took different amounts of risk on equal terms. A fund that earned 12% with wild swings and a fund that earned 9% with mild ones cannot be ranked by return alone, and a risk-adjusted measure is an attempt to rank them.

There are many such measures because "risk" has several reasonable definitions, and each measure picks one. The previous lessons in this section took them one at a time. This one puts them side by side on the same nine portfolios, shows where they agree and where they disagree, and gives a way to choose.

The idea

Two portfolios can be compared honestly only at the same level of risk. A risk-adjusted measure does this by dividing: return per unit of risk. The portfolios can then be ranked by that quotient.

The figure plots nine portfolios by return and by one kind of risk, the deepest fall from a peak. A portfolio is better the further up and the further left it sits. The Calmar ratio is the steepness of the line from the corner of the chart to each point.

Return against two kinds of risk, 2005–2026 0% 4% 8% 12% 16% 0% 10% 20% 30% 40% 50% 60% 70% QQQ 60/40 SPY GLD IWM EFA VNQ AGG TLT Deepest fall from a peak (maximum drawdown) Return per year (CAGR) Calmar ratio = CAGR ÷ maximum drawdown: the slope from the origin to each point.
Compound return against maximum drawdown, January 2005 to October 2026, monthly total returns (GLD: price only; it pays no income). 60/40 is 60% SPY and 40% AGG, rebalanced monthly.

Long-term Treasuries (TLT) and the aggregate bond fund (AGG) earned almost the same return, 3.0% and 2.9% a year. TLT's deepest fall was 47.3% and AGG's 17.1%. Any sensible measure should rank AGG above TLT, and all of them do. The harder cases are the ones where measures disagree, such as gold.

How it works

Measure Risk it divides by Unit Lesson
Sharpe ratio Standard deviation of all returns Pure ratio Sharpe ratio: return per unit of risk, calculated on real funds
Sortino ratio Deviation of returns below a target only Pure ratio Sortino ratio: judging a return by its downside only
Treynor ratio Beta: sensitivity to the market Percent Portfolio beta: how much of the market a portfolio carries
Calmar ratio The single deepest peak-to-trough fall Pure ratio Maximum drawdown (MDD): the worst fall from a peak, and the way back
M² (Modigliani) Volatility, rescaled to match a benchmark Percent This lesson

Two of these are not covered elsewhere.

Treynor ratio. Jack Treynor's 1965 measure divides excess return by beta in place of volatility. The reasoning: an investor who holds many assets has diversified away each one's independent wobble, and what is left is its share of the market's movement. For that investor, beta is the risk that matters. The result is a percentage: the excess return earned per 1.0 of beta.

M². Franco and Leah Modigliani's 1997 measure answers a complaint about the Sharpe ratio: that 0.66 against 0.65 means nothing to most people. M² mixes the portfolio with Treasury bills (or borrows) until its volatility equals the benchmark's, and reports the return of that adjusted portfolio. The ranking is identical to the Sharpe ratio's. The advantage is the unit: a percent a year that can be set directly beside the benchmark's return.

A worked example: a 60/40 portfolio against SPY

The data: monthly total returns from January 2005 to October 2026. Over that period Treasury bills averaged 1.8% a year.

SPY 60% SPY / 40% AGG
Compound return (CAGR) 10.7% 7.8%
Excess return, arithmetic 9.6% 6.2%
Volatility 14.8% 9.4%
Maximum drawdown −50.8% −32.3%
Beta to SPY 1.00 0.63
  1. Sharpe. SPY: 9.6 ÷ 14.8 = 0.65. 60/40: 6.2 ÷ 9.4 = 0.66.
  2. Treynor. SPY: 9.6 ÷ 1.00 = 9.6%. 60/40: 6.2 ÷ 0.63 = 9.9%.
  3. Calmar. SPY: 10.7 ÷ 50.8 = 0.21. 60/40: 7.8 ÷ 32.3 = 0.24.
  4. M² (as excess return). 60/40: 0.66 × 14.8% = 9.7%, against SPY's 9.6%.

SPY earned 2.9 points a year more. On every risk-adjusted measure the 60/40 portfolio was slightly ahead. The M² line states it most plainly: had the 60/40 portfolio been levered up to SPY's volatility, it would have earned 0.1 points a year more than SPY. That is a tie. The result says the two portfolios were about equally efficient, and that the investor's choice between them was a choice of how much risk to carry, not of which was better.

Nine portfolios, four rankings

How nine portfolios rank on four risk-adjusted measures, 2005–2026 Sharpe Sortino Calmar Treynor 3. SPY 3 1. QQQ 1 6. EFA 6 5. IWM 4 8. AGG 7 9. TLT 8 4. GLD 9 7. VNQ 5 2. 60/40 2 Rank 1 at the top. A line that crosses others: the measures disagree about that fund.
Rank of each portfolio on four measures, January 2005 to October 2026. Treynor ratios for holdings with a beta below 0.3 (GLD, AGG, TLT) are not meaningful and are placed last.
Portfolio CAGR Volatility Deepest fall Beta Sharpe Sortino Calmar Treynor
QQQ 14.8% 18.4% −49.7% 1.13 0.75 1.19 0.30 12.2%
60/40 7.8% 9.4% −32.3% 0.63 0.66 0.99 0.24 9.9%
SPY 10.7% 14.8% −50.8% 1.00 0.65 0.97 0.21 9.6%
GLD 10.1% 17.4% −42.9% 0.10 0.54 0.88 0.23 (91.9%)
IWM 8.4% 19.8% −52.5% 1.19 0.42 0.61 0.16 6.9%
EFA 6.0% 16.7% −57.4% 0.96 0.33 0.47 0.11 5.7%
VNQ 6.7% 21.5% −68.3% 1.07 0.33 0.47 0.10 6.7%
AGG 2.9% 4.5% −17.1% 0.07 0.27 0.39 0.17 (16.7%)
TLT 3.0% 13.6% −47.3% −0.08 0.15 0.24 0.06 (−26.9%)

Returns are after fund expenses; taxes and trading costs are not included. Treynor ratios in brackets are calculated but not meaningful.

What the table shows:

  • Sharpe and Sortino agreed completely. The two rankings are identical. With monthly returns of diversified funds, upside and downside volatility are similar enough that the refinement changes the numbers and not the order.
  • Calmar moved the bond fund up three places. AGG's Sharpe ratio is low because its return was low. Its Calmar ratio is respectable because its worst fall was only 17.1%. An investor who cares most about the depth of losses sees AGG differently from one who cares about monthly variability.
  • Calmar moved SPY below gold. Gold's deepest fall was 8 points shallower.
  • Treynor moved real estate up. VNQ was the most volatile fund in the table (21.5%), but much of that volatility was its own and its beta was only 1.07. Per unit of beta it ranked fifth; per unit of volatility, seventh.
  • Treynor failed on three holdings. Gold's beta to SPY was 0.10. Dividing its excess return by 0.10 gives 91.9%, which is arithmetic without meaning. TLT's beta was negative, which turns a positive excess return into a negative ratio. A measure that divides by beta cannot be used for assets that do not move with the benchmark.

Using it: which measure for which question

The question Measure that fits Why
Which of these could be my entire portfolio? Sharpe or M² All of the portfolio's volatility is borne by the holder
Which holding adds most to an already diversified portfolio? Treynor, or alpha Only the market-related part of its risk remains after diversifying
Did this manager beat the index it is paid to beat? Information ratio The alternative is the index, not cash
Returns are lopsided (options, trend following). Is the upside penalized? Sortino Counts only shortfalls as risk
I will need this money and cannot sit through a deep loss Calmar, and the drawdown itself Measures the loss actually experienced, in sequence
How does it behave in falling markets? Downside capture Separates falling months from rising ones

Two habits make any of them more useful.

Look at the parts, not only the quotient. A Sharpe ratio of 0.66 can be 6.2% over 9.4% or 20% over 30%. These are different investments. The ratio says which is more efficient; the parts say which one an investor could live with.

Use more than one. Agreement between measures, as with QQQ and the 60/40 portfolio above, is reassuring. Disagreement, as with gold and the bond fund, is information about the investment: it marks an asset whose risk looks different depending on how risk is defined.

Where it fails or misleads

  • All of them are backward-looking. The ranking above belongs to 2005 to 2026. QQQ ranked first in a period that started after the 2000 to 2002 collapse of the Nasdaq and included a long run of technology leadership. A window starting in early 2000 would put that collapse inside the sample and give a very different ranking.
  • Ratios hide the level. A high ratio on a tiny return may not be worth having unless the holder is willing to use leverage. A Treasury bill fund has almost no volatility and a ratio that means little.
  • They assume risk is one number. Volatility, beta and drawdown each leave things out: the chance of being unable to sell, of a fund closing, of inflation. See Liquidity risk: the cost of selling when you have to and Inflation and real returns: what an investment earns after prices.
  • Smooth returns flatter every ratio. Assets that are priced infrequently or by estimate (private funds, some real estate) show low volatility and shallow drawdowns that reflect the pricing, not the asset.
  • Strategies that sell insurance look best just before they fail. Collecting small premiums produces years of steady gains, a high Sharpe ratio and a small drawdown, until the event insured against happens. Fat tails: why extreme market moves are more common than models say explains why the usual measures miss it.
  • Small differences are noise. A Sharpe ratio of 0.66 against 0.65 over 22 years is no difference at all. Differences of less than about 0.2 over such a period are within the range of chance.
  • Negative ratios do not rank sensibly. When the excess return is negative, a higher volatility makes the Sharpe ratio less negative, which is the wrong direction.

Compared with its neighbours

Family Members Compared with Covered in
Return per total risk Sharpe, Sortino, M² Cash Sharpe ratio: return per unit of risk, calculated on real funds, Sortino ratio: judging a return by its downside only
Return per market risk Treynor, alpha A levered index Portfolio beta: how much of the market a portfolio carries, Alpha: the return that market exposure does not explain
Return per loss Calmar, return ÷ value at risk Nothing; the fund's own path Maximum drawdown (MDD): the worst fall from a peak, and the way back, Value at risk (VaR): the loss a portfolio should rarely exceed
Return per active risk Information ratio An index fund Tracking error and information ratio: a fund against its benchmark
Behaviour by regime Upside and downside capture An index, in two halves Upside and downside capture: a fund in rising and falling markets

How Finabase does it

Finabase does not calculate any risk-adjusted ratio. There is no Sharpe, Sortino, Treynor, Calmar or M² figure on symbol pages, in the screener, in portfolio reports or in backtest results.

What it shows are the parts that go into them:

  • Return. Symbol pages and the comparison table give price returns from one week to two years; a portfolio's report gives its gain by period. None includes a risk adjustment, and symbol returns exclude dividends.
  • Beta. The provider-supplied beta appears in a symbol's key statistics and in the comparison table.
  • Range. The 52-week high and low on a symbol's summary give a quick sense of how far the price has travelled, and the comparison chart shows the depth of recent falls for several symbols at once.
  • Backtest and strategy results report gain, win rate and number of trades without volatility or drawdown, so two tested strategies cannot be ranked on risk-adjusted terms from those screens alone.

To compute the ratios, export month-end prices from a symbol's History tab. Each lesson linked above gives the spreadsheet formula for its measure.

Frequently asked questions

Which risk-adjusted measure is best?

None in general. The Sharpe ratio is the default because it is simple and widely published, and it suits a whole portfolio. For a single holding within a diversified portfolio, beta- based measures fit better; for money with a deadline, drawdown-based ones do.

What is the difference between the Sharpe and Treynor ratios?

The denominator. Sharpe divides excess return by total volatility; Treynor divides it by beta. For a broadly diversified portfolio the two rank funds the same way. For a single stock or a narrow fund, Treynor ignores the risk specific to that holding.

What is M² and why use it?

It is the return a portfolio would have earned if it had been mixed with cash, or levered, to the benchmark's volatility. It ranks portfolios exactly as the Sharpe ratio does and expresses the result in percent a year, which is easier to interpret.

Why do different measures rank the same funds differently?

Because they define risk differently. A bond fund has low volatility, a shallow drawdown and almost no beta, so it ranks differently on each. The disagreement describes the fund.

Is a higher risk-adjusted return always better?

Higher is more efficient, not necessarily preferable. An investor who cannot or will not use leverage may rationally choose the portfolio with the lower ratio and the higher return, accepting more risk to get it.

Summary

Risk-adjusted measures divide return by a chosen definition of risk: total volatility (Sharpe, M²), downside volatility (Sortino), market sensitivity (Treynor) or the deepest loss (Calmar). On nine portfolios from 2005 to 2026, Sharpe and Sortino ranked them identically, while Calmar and Treynor reordered the bond, gold and real estate funds, because those are the assets whose risk depends on the definition. Treynor is unusable when beta is near zero. Pick the measure whose idea of risk matches the purpose of the money, read it alongside its numerator and denominator, and treat small differences as ties.

Check your understanding

5 questions on this lesson. Each answer comes with the reason, and nothing is scored.

  1. What do the Sharpe, Sortino, Treynor and Calmar ratios have in common, and how do they differ?

    1. All divide a return by a measure of risk; they differ in how risk is defined
    2. All divide risk by return; they differ in the period used
    3. They are the same number in different units
    Show the answer

    A. All divide a return by a measure of risk; they differ in how risk is defined Each has a return in the numerator. The denominators are total volatility, downside deviation, beta and maximum drawdown respectively.

  2. Gold had a beta of 0.10 to SPY and an excess return of about 9% a year, giving a Treynor ratio above 90%. How should that be read?

    1. Gold was nine times as efficient as the stock market
    2. The figure is not meaningful: dividing by a beta near zero inflates the ratio
    3. Gold had no risk
    Show the answer

    B. The figure is not meaningful: dividing by a beta near zero inflates the ratio Treynor measures return per unit of market sensitivity. An asset that barely moves with the benchmark has a tiny denominator, and the quotient says nothing about its actual risk, which was substantial.

  3. A 60/40 portfolio earned 7.8% a year with a Sharpe ratio of 0.66. SPY earned 10.7% with a Sharpe ratio of 0.65. What is the fairest conclusion?

    1. The 60/40 portfolio was clearly better
    2. SPY was clearly better
    3. They were about equally efficient; the difference between them was the amount of risk taken
    Show the answer

    C. They were about equally efficient; the difference between them was the amount of risk taken A gap of 0.01 in the Sharpe ratio over 22 years is noise. SPY’s higher return came with proportionally higher volatility, so the choice between them is a choice of risk level.

  4. An investor will need the money in four years and is most concerned about a deep loss before then. Which measure matches that concern most directly?

    1. Treynor ratio
    2. Calmar ratio, together with the maximum drawdown itself
    3. Information ratio
    Show the answer

    B. Calmar ratio, together with the maximum drawdown itself Calmar divides return by the deepest peak-to-trough fall, the loss an investor would have had to sit through. Beta and tracking error describe relative movement, not the size of the loss.

  5. Why can a strategy that collects small, steady premiums show an excellent Sharpe ratio for years?

    1. Because it has no risk
    2. Because its rare, large losses have not yet occurred in the period measured
    3. Because the Sharpe ratio ignores returns
    Show the answer

    B. Because its rare, large losses have not yet occurred in the period measured Risk-adjusted ratios describe the sample. A strategy with infrequent severe losses looks smooth until one arrives, so the ratio is highest just before the loss.

Sources and further reading

  1. William F. Sharpe, “The Sharpe Ratio,” Journal of Portfolio Management, vol. 21, no. 1 (1994)
  2. Jack L. Treynor, “How to Rate Management of Investment Funds,” Harvard Business Review, vol. 43, no. 1 (1965)
  3. Franco Modigliani and Leah Modigliani, “Risk-Adjusted Performance,” Journal of Portfolio Management, vol. 23, no. 2 (1997)
  4. Terry W. Young, “Calmar Ratio: A Smoother Tool,” Futures, vol. 20, no. 1 (1991)
  5. Board of Governors of the Federal Reserve System, 3-Month Treasury Bill Secondary Market Rate (TB3MS), via FRED
  6. Finabase chart data — monthly bars and dividends of SPY, QQQ, EFA, IWM, AGG, TLT and VNQ, and monthly bars of GLD, January 2005 to October 2026; statistics computed by Finabase