Do Investors Care About Risk-Adjusted Returns? What Their Money Says

A retired couple I know well has never looked up the Sharpe ratio of their portfolio. They follow their investments loosely, in dollars, and I have never heard them ask whether they beat a benchmark. They work with a full-service advisor, and I have heard advisors say that clients like them don’t care about risk-adjusted returns.

Their record says they care a great deal. During the 2008–09 crash they sold some of their stocks and moved the money to cash. In March 2020 they did it again. By 2022 roughly half of their portfolio was in bonds, and when stocks and bonds fell together that year, they held, anxious the whole way and unhappy about losing money.

Two crashes in which they sold and one in which they held say more about their preferences than a risk questionnaire could. They will act, at a cost, to limit a loss measured in dollars. That is a preference about risk and return, even though neither of them would describe it with a ratio.

The same pattern runs through the research. Market prices, fund flows and answers to explicit gambles all show investors paying for lower risk, mostly without computing anything. What they pay to avoid is loss, counted in dollars from where they were, which the Sharpe ratio measures poorly. Whether a better Sharpe ratio turns into money you can spend depends on whether you can scale your exposure up and down. Cautious investors like this couple can, which makes the ratio more useful to them than to the aggressive investors who most often dismiss it.

Where “you can’t eat a Sharpe ratio” comes from

We could not find who coined the line. ReSolve Asset Management called it “the popular investment meme” in a paper titled “Yes – You Can Eat Sharpe Ratio.”¹ An older Wall Street saying makes the same complaint about benchmarks: “you can’t eat relative performance.” Barry Popik found it in a 1974 New York Times article quoting Roger G. Kennedy, then the Ford Foundation’s vice president for finance, and by 1995 an AP columnist called it “a time-honored Wall Street aphorism.”² The couple above would agree with both. They spend dollars, and neither a ratio nor a benchmark comparison pays for anything.

The complaint has a real basis. Using annual returns from 1928 to 2025, we combined the S&P 500 with 10-year Treasury bonds in every proportion from 0% to 100% stocks, rebalanced once a year.³ The mix with the highest Sharpe ratio held 50% stocks. It returned 4.64% a year after inflation, against 6.78% for the S&P 500 alone. At those rates a dollar grows to 3.9 real dollars over 30 years in the max-Sharpe mix and to 7.1 in stocks. An investor who picked the portfolio with the best ratio, and didn’t borrow to scale it up, would have finished with a little over half the wealth.

The winning mix also depends on the years you pick. From 1928 to 1980 the highest Sharpe ratio belonged to 100% stocks. From 1981 to 2025, a stretch that began with four decades of falling interest rates, it belonged to a 51% stock mix. With 98 annual observations, each Sharpe estimate carries a standard error of about 0.11,⁴ several times the gap between the 50% mix (0.46) and all stocks (0.43).

What investors’ money says

The equity premium is a price on risk

The largest piece of evidence is the stock market itself. Mehra and Prescott measured an average equity premium of 6.18% a year from 1889 to 1978, and found that a standard model with relative risk aversion anywhere from 0 to 10 could produce a premium of at most 0.35%.⁵ Investors as a group have demanded far more pay for holding stocks than textbook risk aversion explains. Economists still disagree about why. One leading explanation, loss aversion combined with checking portfolios often, is covered in our guide to panic selling. Whatever the cause, the premium is a price that investors set without computing a Sharpe ratio, and Stocks Are Always Risky explains why it has to come with risk attached.

Fund flows adjust for market risk

Fund flows are a second, noisier record. Barber, Huang and Odean found that mutual fund investors “attend most to market risk (beta) when evaluating funds,” while treating returns from size, value, momentum and industry tilts as manager skill.⁶ Berk and van Binsbergen ranked asset-pricing models by how well they explain flows and found that the CAPM, which adjusts returns for market risk, came closest.⁷ Investors move money toward funds that beat what their market exposure alone would have delivered. That is a risk adjustment, if a crude one.

Ben-David, Li, Rossi and Song read the same kind of data differently: “Simplistic performance chasing best explains aggregate flows.”⁸ An earlier working-paper version attributed much of the apparent risk adjustment to investors following Morningstar ratings. A Boston Fed working paper by Victoria Liu adds a downside angle: funds with asymmetric exposure to downside risk “experience more than double the performance induced fund flow than those without,” and the effect grows during market stress.⁹

The star rating is built on a utility function

Investors who follow Morningstar ratings are following a risk-adjusted number. The star rating ranks funds by Morningstar Risk-Adjusted Return, which the methodology defines as “the guaranteed riskless return that provides the same level of utility to the investor as the variable excess returns of the risky portfolio.”¹⁰ Economists call that a certainty-equivalent return. Morningstar sets its risk-aversion parameter to 2 because, in its words, that value “results in fund rankings that are consistent with the risk tolerances of typical retail investors.” Given how Morningstar writes its utility function, a parameter of 2 corresponds to relative risk aversion of 3. The methodology also says investors “dislike downside variation more than upside variation,” and its formula gives downside variation more weight.

An investor who picks a five-star fund is choosing by a ranking built on a utility function with relative risk aversion of 3, whether or not they have heard of one.

What investors say when the tradeoff is explicit

The Health and Retirement Study, which began in 1992 with a cohort then aged 51 to 61 and their spouses, asked respondents whether they would trade a job guaranteed to pay their current family income for life for an equally good job with a 50-50 chance of doubling that income or cutting it by a third. Those who declined were offered a smaller cut, a fifth. In the analysis by Barsky, Juster, Kimball and Shapiro, 64.6% of respondents turned down both, which implies relative risk aversion above 3.76.¹¹ Broader surveys that put a risk-return menu in front of people get answers in the same direction.

Each comes with a caveat. The ICI is the fund industry’s trade group, and its split shows selection as much as preference: 20% of households that own mutual funds chose below-average or no risk, against 57% of households that don’t. Natixis is an asset manager surveying wealthy investors. All four record what people say, and risk questionnaires predict little about what people do in a crash.

When Choi and Robertson asked 1,013 Americans what drives how much of their money they keep in stocks, the factors rated very or extremely important most often were years left until retirement (48% of employed respondents), the risk of illness or injury (47%), needing cash on hand for routine expenses (47%) and the risk of a rare economic disaster (45%). Worry about “even small losses” on stock investments, the survey’s loss-aversion item, was rated that important by 28%. Most of the top answers concern needing the money or fearing a collapse, which are both about when a loss would hurt.¹⁵

Experience moves these preferences. Using the Survey of Consumer Finances from 1964 to 2004, Malmendier and Nagel found that people who had lived through low stock returns reported less willingness to take financial risk, were less likely to own stocks, and held a smaller share in stocks when they did own them. Recent years counted most, but early experiences still showed up decades later.¹⁶ Among clients of one Italian bank, the share choosing “low return and no risk” rose from 16% in 2007 to 43% in 2009.¹⁷

The risk investors pay to avoid is loss

Across 607 estimates from 150 studies, Brown, Imai, Vieider and Camerer put the mean loss-aversion coefficient at 1.955: a loss weighs about twice as much as a gain of the same size.¹⁸ Loss aversion explains something ordinary risk aversion can’t. Matthew Rabin showed that if a person turns down a 50-50 bet to lose $100 or win $110 at every level of wealth, standard expected-utility theory implies they would also turn down a 50-50 bet to lose $1,000 or win any amount at all.¹⁹ Almost nobody would turn down that second bet. Caution over small stakes has to come from how people weigh losses against a reference point, and the couple measure their portfolio that way: in dollars, against what it was worth before the decline.

Volatility is a poor yardstick for that preference. The Sharpe ratio divides by standard deviation, which counts a 30% gain as just as risky as a 30% loss; our guide to portfolio risk metrics covers drawdown-based alternatives. Investors don’t mind upside. Kumar found that individual investors “prefer stocks with lottery features,”²⁰ and Shefrin and Statman’s behavioral portfolio theory describes portfolios built in layers, a low layer meant to avoid poverty and a high layer meant to offer a shot at riches.²¹ The same person can want protection from a large loss and a chance at a large gain. A measure that penalizes both equally describes neither.

When you can eat a Sharpe ratio

William Sharpe described the setting where his ratio is the right tool: an investor splitting money between a riskless asset and a single risky fund.²² Holding more or less of the fund, or borrowing to hold more than 100%, moves expected return and risk in proportion, so the fund with the higher ratio offers more return at every level of risk. The investor’s risk aversion then decides how much of it to hold.

A certainty-equivalent return, the guaranteed return an investor would accept in place of an uncertain one, puts a value on the ratio. Hold a fraction $w$ of wealth in a risky portfolio with expected excess return $\mu$ and volatility $\sigma$, and the rest in Treasury bills paying $r_f$. With relative risk aversion $\gamma$, the certainty-equivalent return is approximately:

The best fraction is the Merton share, and substituting it back gives the gain over holding only bills:

where SR is the Sharpe ratio $\mu/\sigma$. The formula above is an approximation, but in Merton’s continuous-time model with constant relative risk aversion the result is exact; as Paolo Guasoni’s lecture notes put it, “Only total Sharpe ratio matters.”²³

At the defaults in our Merton share calculator (4% expected excess return, 18% volatility, relative risk aversion of 3), the Sharpe ratio is 0.22, the investor holds 41% in stocks, and the certainty-equivalent gain is about 0.8% a year. Raise the Sharpe ratio to 0.30 at the same volatility and the gain becomes 1.5% a year, with 56% in stocks. These are outputs of assumed inputs, not forecasts.

An investor whose best mix is below 100% stocks collects all of that gain, because reaching it only takes holding cash or bonds alongside the risky portfolio. For that investor, a higher Sharpe ratio is worth the same as a higher guaranteed yield.

An investor already at 100% stocks can’t scale a higher-ratio, lower-return portfolio up to the risk they want without borrowing. Frazzini and Pedersen built a theory of how that constraint shapes prices, summed up in one line: “More constrained investors hold riskier assets.”²⁴ For these investors the borrowing rate decides whether a better ratio is worth anything. We levered the 50/50 mix from the 1928–2025 test 1.84 times, enough to match the S&P 500’s volatility, and charged the borrowed portion the Treasury bill rate plus a spread.

The levered mix and the S&P 500 finished even at a spread of about 0.77 points over bills. In October 2026, published margin rates on small balances ran from about 1.3 to 8 points over the three-month bill, as our personal leverage guide documents; box spread loans can come closer to Treasury rates. The levered mix’s edge also rests on bond returns from 1981 onward. For an aggressive investor who can’t borrow cheaply, expected return matters more than the ratio, and “you can’t eat a Sharpe ratio” describes their situation correctly.

Without leverage, the best Sharpe ratio picks a risk aversion for you

For an investor who won’t borrow, choosing the max-Sharpe portfolio amounts to choosing a level of risk aversion. The table below uses the same 1928–2025 data and lists the stock share that would have maximized each objective, with certainty-equivalent returns computed on the historical annual real returns.

S&P 500 and 10-year Treasury bond total returns, rebalanced annually, deflated by CPI. Source: Damodaran.³ No taxes or costs. Our calculation.

In this history the max-Sharpe mix sits between the choices of investors with relative risk aversion of 3 and 5. Morningstar’s typical retail investor would have held about two-thirds in stocks; the HRS majority, with risk aversion above 3.76, would have held less. An investor with risk aversion of 2 would have held 90%, and for that investor the max-Sharpe portfolio would have been a poor choice.

The table has limits worth keeping in view:

• It is one US sample, and the bond results lean on the 1981–2020 decline in interest rates.
• Year-end data hide declines within a year, so the drawdown column understates what holders lived through.
• The certainty equivalents treat each historical year as an equally likely draw, which ignores the order of returns that matters to a retiree drawing income.

The Merton share guide works through the same tradeoff with forward-looking inputs, and the portfolio risk metrics guide has a calculator for historical drawdowns.

Back to the couple

The table below shows the size of the three declines the couple lived through, for an S&P 500 index fund and for a 50/50 split with a total bond market index fund. Their actual holdings differed, so read it as the approximate scale of the losses they faced.

Vanguard 500 Index Fund (VFINX) and Vanguard Total Bond Market Index Fund (VBMFX), daily prices adjusted for distributions; the 50/50 split is rebalanced on the first trading day of each month, and its peak and trough dates differ from the stock fund’s by up to a few weeks. Our calculation.²⁵

In 2020 the S&P 500 fund regained its February 19 high on August 10. Money moved to cash in March and left there through the summer missed that recovery, which is the usual cost of selling in a crash. In 2022 the bond half fell too: the bond fund lost 13.2% for the year. The 50/50 portfolio’s decline reached 20.0%, and it regained its December 2021 high on February 29, 2024, about 11 weeks after the stock fund recovered. The couple held through all of it.

Their history shows strong aversion to losses, the same trait the HRS majority showed. Their current mix sits near what an investor with relative risk aversion of 4 to 5 would have chosen in the 1928–2025 data, and near the max-Sharpe mix. One couple can’t show whether the lower stock share is why they held in 2022. It does make them the kind of investor for whom the Sharpe ratio of the stock half matters most: they scale with bonds, so a better return per unit of risk in the stock portfolio reaches them in full.

When drawdown aversion costs too much

Loss aversion becomes expensive when it drives selling after a decline. The cost lands in the recovery, and in 2020 the recovery took less than six months. The panic-selling guide covers how often investors sell in crashes and what it has cost them.

Safety in dollars can also hide a real loss. In the same Damodaran data, 10-year Treasury bonds lost 55.5% after inflation from the end of 1940 to the end of 1981, a deeper real decline than the worst one for stocks in that series. Nominal statements showed little of it. A retiree’s spending may need to last 30 years, and a portfolio that never shows a large loss can still fail to fund it; funding an income floor first and taking risk with the rest is one way to hold both goals, and sequence of returns risk explains why the timing of losses matters once withdrawals start.

A cautious portfolio isn’t always a preference, either. Using how 401(k) savers responded to different default funds, Choukhmane and de Silva estimate that “absent participation frictions, 94% of investors prefer holding stocks,” and attribute low stock ownership to frictions rather than to loss aversion.²⁶ Inertia can look like caution.

What we recommend

Choose how much risk to take by the dollar loss you would hold through. Multiply the declines in the crash table by your balance: a 30% decline on $1 million is $300,000. If that number would make you sell, hold less stock now, when changing costs little, instead of during the next decline.

Use the Sharpe ratio to choose among portfolios with similar risk. Diversifying, removing uncompensated concentration, and lowering costs raise the ratio without adding drawdown. If you hold less than 100% stocks, those improvements reach you in full. If you are at 100% and want more return, a higher-ratio, lower-return portfolio only helps if you can borrow close to the Treasury bill rate.

If you work with an advisor, ask for the plan’s worst historical decline in dollars, and for what the plan does if you need to sell during one.

Frequently Asked Questions

Do investors care about risk-adjusted returns?

Most don’t compute them, but their choices adjust for risk. The equity premium, fund flows that track market-risk-adjusted performance, and survey majorities that choose cautious options all point the same way. What investors adjust for is mostly loss, which volatility measures poorly.

What does “you can’t eat a Sharpe ratio” mean?

You spend dollars, and a portfolio with a better ratio can leave you with fewer of them. From 1928 to 2025, the stock and Treasury mix with the highest Sharpe ratio (50% stocks) grew a dollar to $3.9 in real terms over 30 years at its average rate, against $7.1 for the S&P 500. The saying holds for investors at 100% stocks who can’t borrow cheaply. Investors who hold less can turn a better ratio into a higher certainty-equivalent return.

Is loss aversion irrational?

Not by itself. Weighing a loss about twice as heavily as an equal gain is the norm in experiments; a 2024 meta-analysis put the average coefficient at 1.955. It becomes costly when it leads to selling after a decline, or to holding so little growth that inflation erodes the portfolio. Sizing the portfolio so you won’t sell turns the preference into a plan.

What is a certainty-equivalent return?

The guaranteed return you would accept in place of a risky portfolio’s uncertain one. Morningstar’s star rating is built on one. For an investor who can scale exposure with cash, the certainty-equivalent gain from investing is approximately the Sharpe ratio squared divided by twice the investor’s relative risk aversion.

Does Morningstar’s star rating account for risk?

Yes. Stars rank funds within a category by Morningstar Risk-Adjusted Return, a certainty-equivalent return computed with a risk-aversion parameter of 2, which Morningstar says matches typical retail investors and which corresponds to relative risk aversion of 3. The rating describes past returns only.

Should retirees who hate drawdowns hold more bonds?

Often, if the alternative is selling in the next crash. A 50/50 split roughly halved the decline in 2008–09 and 2020, but cut it by less than five points in 2022, when the bond fund fell 13.2%. Retirees also need growth for spending that may last decades, so set the bond share from the loss you would hold through and the spending floor you need to protect.

Key Takeaways

• Investors price risk without ratios. The equity premium, flows that track market-risk-adjusted performance, and the 64.6% of HRS respondents who rejected both income gambles all show it.
• The star rating is built on a utility function. Morningstar ranks funds by a certainty-equivalent return with relative risk aversion of 3, which it describes as typical of retail investors.
• Loss is what investors price. A loss weighs about twice an equal gain, and the Sharpe ratio counts upside and downside volatility the same.
• Cautious investors can eat a Sharpe ratio. Below 100% stocks, the certainty-equivalent gain is about $\text{SR}^{2}/2\gamma$. At 100% stocks, a better ratio pays only with cheap borrowing; in our 1928–2025 test the break-even was about 0.77 points over bills.
• Size risk in dollars. A 50/50 index portfolio fell 30.0% in 2007–09, 18.4% in 2020 and 20.0% in 2022. Pick the decline you would hold through before the next one starts.

Related Guides

• Portfolio Risk Is More Than Volatility on drawdowns, the Ulcer Index and downside-adjusted ratios.
• The Merton Share on sizing stock exposure from risk aversion, with a calculator.
• Is Never Panic-Selling an Investing Superpower? on who sells in crashes and what it costs.
• Yes, an Unrealized Loss Is Still a Loss on why a decline matters before you sell.
• Personal Finance Gurus vs. Economists on rules of thumb against textbook portfolio models.
• Compensated vs. Uncompensated Risk on which risks pay.
• Retirement Income Floor on protecting essential spending before taking market risk.
• Personal Leverage on when borrowing to invest can make sense.

Sources

1. Adam Butler, Michael Philbrick and Rodrigo Gordillo, “Yes – You Can Eat Sharpe Ratio”, ReSolve Asset Management, October 2017. The PDF is undated; the date is from ReSolve’s landing page.
2. Barry Popik, “You can’t eat relative performance (Wall Street proverb)”, The Big Apple. Secondary source for the 1974 New York Times quotation, which we did not see directly.
3. Aswath Damodaran, “Historical Returns on Stocks, Bonds and Bills: 1928–2025”, NYU Stern. Annual S&P 500, 10-year Treasury bond, 3-month Treasury bill and CPI series. Mixes, Sharpe ratios (excess return over bills), leverage and certainty-equivalent returns are our calculations.
4. Andrew W. Lo, “The Statistics of Sharpe Ratios”, Financial Analysts Journal 58, no. 4 (2002): 36–52. The standard error uses the independent-returns approximation, the square root of (1 + SR²/2)/T.
5. Rajnish Mehra and Edward C. Prescott, “The Equity Premium: A Puzzle”, Journal of Monetary Economics 15, no. 2 (1985): 145–161.
6. Brad M. Barber, Xing Huang and Terrance Odean, “Which Factors Matter to Investors? Evidence from Mutual Fund Flows”, Review of Financial Studies 29, no. 10 (2016): 2600–2642.
7. Jonathan B. Berk and Jules H. van Binsbergen, “Assessing Asset Pricing Models Using Revealed Preference”, Journal of Financial Economics 119, no. 1 (2016): 1–23. The CAPM result is stated in the NBER working paper abstract.
8. Itzhak Ben-David, Jiacui Li, Andrea Rossi and Yang Song, “What Do Mutual Fund Investors Really Care About?”, Review of Financial Studies 35, no. 4 (2022): 1723–1774. The Morningstar-ratings finding is from the 2019 working-paper version.
9. Victoria Liu, “Downside Risk and Fund Flow”, Federal Reserve Bank of Boston Supervisory Research and Analysis Working Paper SRA 22-01 (2022). Not peer-reviewed.
10. Morningstar, “The Morningstar Rating for Funds”, Morningstar methodology document, April 2023, listed in Morningstar’s methodology library. The same text, including “RRA(.) = γ + 1,” appears in the 2006 version.
11. Robert B. Barsky, F. Thomas Juster, Miles S. Kimball and Matthew D. Shapiro, “Preference Parameters and Behavioral Heterogeneity: An Experimental Approach in the Health and Retirement Study”, Quarterly Journal of Economics 112, no. 2 (1997): 537–579. Figures from Table 3-A of NBER Working Paper 5213.
12. Sarah Holden, Daniel Schrass and Michael Bogdan, “Ownership of Mutual Funds and Shareholder Sentiment, 2025”, ICI Research Perspective 31, no. 8 (November 2025), Figure 8. The ICI is the US fund industry’s trade association.
13. Judy T. Lin and Christopher Bumcrot, “Investors in the United States: A Report of the National Financial Capability Study”, FINRA Investor Education Foundation, December 2025. The survey offered four risk choices, so its categories differ from the ICI’s.
14. Natixis Investment Managers, 2025 Global Survey of Individual Investors, US press release, June 10, 2025. The sponsor is an asset manager.
15. James J. Choi and Adriana Z. Robertson, “What Matters to Individual Investors? Evidence from the Horse’s Mouth”, Journal of Finance 75, no. 4 (2020): 1965–2020.
16. Ulrike Malmendier and Stefan Nagel, “Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?”, Quarterly Journal of Economics 126, no. 1 (2011): 373–416.
17. Luigi Guiso, Paola Sapienza and Luigi Zingales, “Time Varying Risk Aversion”, Journal of Financial Economics 128, no. 3 (2018): 403–421. The 16% and 43% figures are from the NBER working paper.
18. Alexander L. Brown, Taisuke Imai, Ferdinand M. Vieider and Colin F. Camerer, “Meta-Analysis of Empirical Estimates of Loss Aversion”, Journal of Economic Literature 62, no. 2 (2024): 485–516.
19. Matthew Rabin, “Risk Aversion and Expected-Utility Theory: A Calibration Theorem”, Econometrica 68, no. 5 (2000): 1281–1292.
20. Alok Kumar, “Who Gambles in the Stock Market?”, Journal of Finance 64, no. 4 (2009): 1889–1933.
21. Hersh Shefrin and Meir Statman, “Behavioral Portfolio Theory”, Journal of Financial and Quantitative Analysis 35, no. 2 (2000): 127–151.
22. William F. Sharpe, “The Sharpe Ratio”, Journal of Portfolio Management, Fall 1994. Author’s copy.
23. Paolo Guasoni, “Long Run Investment”, lecture slides, Winter School on Mathematical Finance, Lunteren, January 2019.
24. Andrea Frazzini and Lasse Heje Pedersen, “Betting Against Beta”, Journal of Financial Economics 111, no. 1 (2014): 1–25.
25. Daily adjusted closing prices for VFINX and VBMFX from Yahoo Finance, retrieved November 2026. Drawdowns and recovery dates are our calculations.
26. Taha Choukhmane and Tim de Silva, “What Drives Investors’ Portfolio Choices? Separating Risk Preferences from Frictions”, NBER Working Paper 32476 (2024, revised November 2025). Based on 401(k) data from one large US financial institution.

Editor’s note

Educational content, not investment advice. The historical tables are backtests on one US sample with no taxes or costs; they describe the past and are not expected returns. The couple’s holdings differed from the index portfolios in the crash table, and the anecdote is shared with identifying details removed. Survey results are self-reported.