Missing the Market's Best Days: What the Famous Chart Leaves Out

Every market selloff revives the same chart: a bar graph showing that an investor who missed the market’s 10 or 20 best days ended up with a fraction of the buy-and-hold outcome. Advisors post it, fund companies print it, and influencers caption it “time in the market beats timing the market.” We raised a question about it on the Engineer Investor feed¹ that the chart never answers: what happens if you also remove the worst days?

The answer is documented in the same research the chart comes from. In Bank of America’s version of the exercise, deleting each decade’s 10 best days since 1930 collapses the S&P 500’s cumulative return from 19,975% to 45%. Deleting each decade’s 10 worst days instead lifts it to 4,275,143%. And deleting both sets, the honest version of the thought experiment, produces 30,678%, which beats staying fully invested.² The chart everyone shares is one row of a four-row table. This guide walks through the whole table: why the both-tails result happens, why extreme days cluster, what trend following can and cannot do about it, and why the case for buy-and-hold index investing survives, and even strengthens, once the chart’s weakness is admitted.

What the viral chart gets right

The arithmetic is real and the numbers are large. In the BofA analysis (price returns, 1930 through March 2022), sitting out the 10 best days per decade left an investor with 45% cumulative growth against 19,975% for staying put.² Merrill’s update through the end of 2024 shows the same shape: 27,320% fully invested versus 93% without each decade’s 10 best days.³ J.P. Morgan’s widely used 20-year version (2005 through 2024) finds that missing just the 10 best days roughly halved the annualized return, and missing the best 40 turned it negative.⁴

Two details of the BofA table are routinely dropped when the chart circulates. First, it removes the 10 best days from every decade, roughly 90 days across the sample, which is a different and larger intervention than removing 10 days from a 30-year window. Second, it uses price returns without dividends, which changes the cumulative figures substantially over 90 years. Charts built on different windows, day counts, and return definitions get quoted interchangeably, and they should not be.

The rows that never get shown

The same table contains the mirror image. Avoiding each decade’s 10 worst days helped far more than missing the best days hurt: 4,275,143% versus the 45% disaster case. And the symmetric experiment, deleting both sets, beat buy-and-hold, 30,678% to 19,975%.²

Cliff Asness made this the centerpiece of an AQR piece first written in 1999 and updated since. Using monthly data from 1970 to 1996, he found that missing the 12 best months cost about 5.1 percentage points of annualized return, while dodging the 12 worst months added about 5.8 points: near mirror images. He then showed that returns simulated from a plain normal distribution produce the same “extreme observations dominate” pattern, so the finding reflects compounding itself rather than anything special about stock markets. His conclusion was blunt: there are good reasons to avoid market timing, but this argument is not one of them.⁵

One more piece of honesty the cumulative numbers obscure: 30,678% versus 19,975% sounds enormous, but spread over 92 years it is about 0.47 percentage points of annualized outperformance. Real, and far smaller than the six-digit percentages suggest. The tails come close to canceling.

Why deleting both tails can win

Percentage gains and losses are not symmetric in wealth terms. A +10% day followed by a -10% day leaves you at 99% of where you started, because 1.10 × 0.90 = 0.99. A 10% loss needs an 11.11% gain to break even; a 50% loss needs 100%. Every paired extreme up-and-down episode therefore grinds wealth down a little, and deleting the pair removes the grind. When the worst days are larger in magnitude than the best days, or when there are enough such pairs, the compounded return of the deleted days is negative, and removing them raises terminal wealth.

There is no theorem guaranteeing the both-tails version wins. The outcome depends on the window, the day count, whole-period versus per-decade selection, price versus total returns, and what the excluded days earn. The calculator below lets you test windows where it wins and windows where it loses, which is itself the point: a result this fragile is a property of the sample, and it carries no instruction about what to do next.

Extreme days cluster in the same crises

The best and worst days are neighbors. In J.P. Morgan’s 2005-2024 sample, seven of the 10 best days occurred within two weeks of one of the 10 worst days.⁴ The pattern is general. Malkiel, Saha, and Grecu studied extreme daily moves (three standard deviations or more) across decades of U.S. data and found they cluster tightly and are typically preceded by unusually large moves in the prior three sessions. Volatility arrives in regimes: crashes, margin calls, forced selling, and the violent rebounds that follow all inhabit the same few weeks.⁶

Their second finding deserves equal billing. Using the clustering, they could build simulated rules that sidestepped roughly 80% of the extreme down days. Those rules cut portfolio volatility meaningfully, and they still failed to raise long-run returns above simple buy-and-hold, because the same rules sat out the recoveries.⁶ Javier Estrada reached a similar dead end from another direction: across 15 international markets and more than 160,000 daily returns, missing the 10 best days cut terminal wealth by about half, and avoiding the 10 worst days multiplied it 2.5 times, but the decisive days are fewer than 0.1% of all trading days. Betting a portfolio on identifying one day in a thousand in advance is, in his words, a goose chase.⁷

Four questions the chart cannot answer

Keeping four questions separate clears up most of the confusion in this debate:

The viral chart answers the first question and is silently marketed as an answer to the second. The both-tails rebuttal makes the same leap in the other direction. Everything that follows is about questions two through four.

What trend following does with the tails

Trend following is the most serious systematic attempt to hold equities in good regimes and step aside in bad ones. The rules are simple in outline: measure whether an asset has been rising or falling over recent months, hold it when the trend is up, reduce or short it when the trend is down, size positions inversely to volatility, and diversify across dozens of futures markets. Because extreme days cluster inside downtrends, a trend rule that is out of the market during a prolonged bear market skips some of the worst days as a byproduct.

It is an approximation of the oracle, with structural gaps. A trend rule is fully invested when a crash begins from a calm uptrend; it turns defensive only after prices have already fallen; and it is still defensive when the sharpest rebound days fire near the bottom. A fast V-shaped reversal hits it twice, once on the way down and again re-entering late, and sideways markets generate whipsaw losses from repeated false signals.⁸ The spring of 2020 is the clean example: the S&P 500 fell about a third and recovered within months, and the SG Trend index of large trend-following managers finished the year modestly positive, useful, but nothing like the crisis protection the strategy showed in the long grind of 2008.⁹ Trend following protects against persistent drawdowns. It is not crash insurance, and it does not identify tomorrow’s best or worst day.

The evidence on trend and volatility timing, both sides

The supportive literature is substantial. Moskowitz, Ooi, and Pedersen documented time-series momentum across 58 futures and forward markets, with returns persisting over roughly 1 to 12 months and a diversified trend portfolio performing best in the most extreme up and down markets.¹⁰ Hurst, Ooi, and Pedersen extended the simulation across 67 markets back to 1880 and found positive returns in every decade, net of estimated transaction costs and a hypothetical 2-and-20 fee, with low correlation to stocks and bonds. Those are backtested results, not a live track record, and the authors flag real uncertainty in the historical cost estimates.¹¹

The critiques deserve the same airtime. Kim, Tse, and Wald found that stripping the volatility-scaling overlay out of the original time-series-momentum construction left performance similar to buy-and-hold, with statistically indistinguishable alphas: much of the apparent edge came from risk-based position sizing rather than directional prediction.¹² Huang, Li, Wang, and Zhou re-tested the 12-month predictability rule asset by asset and found the statistical evidence weak, with part of the performance reflecting persistent differences in average returns across assets.¹³

Volatility management tells the same story in miniature. Moreira and Muir showed that cutting exposure when recent volatility is high raised Sharpe ratios across many factors in-sample, because volatility is persistent while expected returns do not rise one-for-one with it.¹⁴ Cederburg, O’Doherty, Wang, and Yan then tested the idea across 103 strategies and found no systematic outperformance versus unmanaged versions, with real-time implementable versions frequently underperforming.¹⁵ And for the retail favorite, Faber’s 10-month moving-average rule historically matched market returns with smaller drawdowns,¹⁶ while Zakamulin’s out-of-sample re-examination found the advantage largely evaporates once data-mining bias and realistic trading frictions are counted.¹⁷ A pattern can be convincing in hindsight and still fail to convert into a reliable real-time improvement. That sentence summarizes half of empirical finance.

Why buy-and-hold beta stays the default

The strong case for buy-and-hold index investing never depended on the scary chart. It rests on four properties.

It requires no forecasts. An index holder never has to decide whether a decline has begun, whether volatility will persist, or when to re-enter. William Sharpe estimated in 1975 that a stocks-versus-cash timer needs to be right roughly 70% of the time just to keep pace with buy-and-hold, because the penalty for wrong exits compounds.¹⁸

It minimizes leakage. Low turnover means low costs, low taxes, and few decisions. Barber and Odean’s study of 66,465 brokerage households found the most active traders earned 11.4% annually while the market returned 17.9%, about 6.5 points a year surrendered to trading.¹⁹

Demonstrated timing skill is rare. Graham and Harvey examined 237 investment-newsletter timing strategies over 1980-1992 and found no evidence of any ability to shift into equities before rallies or out before declines.²⁰ Professionals fare little better against the index: in the SPIVA year-end 2025 scorecard, 79% of active U.S. large-cap funds trailed the S&P 500 that year, roughly nine in ten trail over 15 to 20 years, and the companion persistence scorecard finds that outperformance rarely repeats.²¹

And it is behaviorally durable. Trend strategies lag badly through long bull markets, sideways chop, and sharp reversals; an investor who abandons one after three disappointing years converts tracking error into permanent loss. A strategy’s theoretical Sharpe ratio is irrelevant if its owner cannot hold it. Buy-and-hold is easy to automate, easy to benchmark, and easy to keep.

Who a trend sleeve fits

None of this makes trend following useless. As a modest sleeve, on the order of 10-20% of a portfolio, diversified across markets and run systematically, it has a credible century of simulated evidence as a diversifier that tends to earn its keep in prolonged bear markets. The fit depends on the investor: it suits someone who understands it is a diversifier rather than a hedge, tolerates years of lagging equities, and evaluates it at the portfolio level over full cycles. It does not suit an investor who wants a three-fund portfolio, chases last year’s winner, or expects protection in every selloff. Our managed futures guide covers the implementation details: fund choices, leverage, taxes, and the 2022-2023 performance-chasing lesson. Anyone tempted by signal-following in retirement specifically should read the Bengen tactical-risk guide first.

Run the exercise yourself

The calculator below runs the whole four-row table, plus the one implementable comparison, on real daily U.S. market total returns starting July 1926.²² Try the default 30-year window, then flip to the per-decade BofA convention over the full sample. Watch three things: whether “miss both” beats buy-and-hold in your window, how many best days sit within 10 trading days of a worst day, and what the 10-month moving-average rule captured, both the worst days it dodged and the best days it paid for that with.

A few windows worth trying. 1996 to 2026 reproduces the modern influencer chart. 1930 to 1940 shows a decade where the tails were so violent that hindsight deletion changes everything. 2010 to 2020 shows a calm decade where it barely matters and the SMA rule mostly just paid whipsaw costs. The lesson is the same one Asness drew: the exercise measures compounding and the sample, and it cannot tell you what to do tomorrow.⁵

Key Takeaways

• The missing-best-days chart is arithmetic, and it is one row of a four-row table. Avoiding the worst days helped more than missing the best days hurt, and deleting both beat buy-and-hold in the BofA sample, by about 0.47 percentage points a year.
• Compounding explains it: a 10% loss needs an 11.11% recovery, so paired extremes grind wealth down and deleting them helps.
• Best and worst days cluster in the same crises. Rules that dodge the worst days also miss the rebounds, which is why simulated extreme-day avoidance cut volatility without raising returns.
• Trend following approximates persistent-drawdown protection with a lag; it is not crash insurance, and its edge net of frictions is contested in the academic literature.
• Buy-and-hold index beta stays the default because it needs no forecasts (Sharpe's timer needs ~70% accuracy), minimizes cost and tax leakage, and survives investor behavior. A 10-20% diversified trend sleeve is a reasonable option for investors who accept tracking error, never a replacement for the core.
• Any chart that deletes days you could only identify afterward is measuring hindsight. Judge strategies on real-time rules, net of costs, out of sample.

Frequently asked questions

What happens if you miss the 10 best days in the market?

In Bank of America's per-decade version, missing each decade's 10 best days since 1930 reduced the S&P 500's cumulative price return from 19,975% to 45% (through March 2022). In J.P. Morgan's 20-year version, missing the 10 best days roughly halved the annualized return. The figures are correct; the omission is that avoiding the worst days would have helped even more.

Is the missing-best-days chart misleading?

The numbers are accurate and the framing is selective. It shows the penalty of missing the best days without the symmetric benefit of avoiding the worst days, and it implies that exiting means missing only good days. Since extreme days cluster, a real exit during turmoil misses days of both kinds.

What if you missed both the best and worst days?

In the BofA sample, deleting each decade's 10 best and 10 worst days produced 30,678% versus 19,975% for buy-and-hold, about half a percentage point a year of outperformance. The result flips in some windows, and either way it requires perfect hindsight, so it is an illustration of compounding rather than a strategy.

Can trend following avoid the market's worst days?

Partially and with a lag. Trend rules reduce exposure after a downtrend establishes, so they skip some worst days inside prolonged bear markets, and they miss rebound days near bottoms for the same reason. They offer little protection in fast crashes that begin from calm uptrends, as in early 2020.

Does the 200-day moving average beat buy and hold?

Historically, simple moving-average rules like Faber's 10-month version roughly matched market returns with smaller drawdowns in backtests. Out-of-sample re-examinations, including Zakamulin's, find the advantage largely disappears after data-mining bias and trading costs. Lower drawdowns are real; higher returns are not reliable.

Should I just stay fully invested all the time?

For most DIY investors, yes, in a diversified low-cost portfolio sized so its worst historical drawdown is one you can hold through. The defense against crashes is an allocation that never depended on predicting them. Investors who cannot tolerate a 50% equity decline should hold fewer equities structurally rather than plan to exit in time.

Related guides

• Do You Need Managed Futures? An Evidence-Based Look at Trend Following
• Bengen vs. Bengen: Should Retirees Outsource Equity Risk to a Third-Party Signal?
• Should Long-Term Index Investors Use Stop-Loss Orders?
• The Problem With Buy the Dip
• Lump Sum vs. Dollar-Cost Averaging: 98 Years of Evidence
• The Problem With “Just Invest in Index Funds”

Author disclosure

This guide is educational and descriptive. The excluded-day scenarios are hindsight exercises, no allocation to any strategy or fund is being recommended, and the calculator's SMA comparison uses a fixed published specification precisely so it cannot be tuned into a sales pitch.

Sources

1. Engineer Investor (@egr_investor). Post on the missing-best-days chart. X, June 2023.
2. Subramanian, S., et al. (2022). “Equity Strategy Focus Point: History lessons for volatile markets.” BofA Global Research, March 9, 2022. PDF.
3. Merrill Chief Investment Office (2025). Steer the Course of Your Financial Future. Data through December 31, 2024.
4. J.P. Morgan Asset Management (2025). Guide to Retirement: impact of being out of the market.
5. Asness, C. (1999, updated 2025). (So) What If You Miss the Market’s N Best Days? AQR.
6. Malkiel, B. G., Saha, A., & Grecu, A. (2009). “The Clustering of Extreme Movements: Stock Prices and the Weather.” CEPS Working Paper No. 186 / Journal of Investment Management. PDF.
7. Estrada, J. (2008). “Black Swans and Market Timing: How Not to Generate Alpha.” The Journal of Investing, 17(3), 20-34. PDF.
8. Babu, A., et al. (2020). “You Can’t Always Trend When You Want.” Journal of Portfolio Management, 46(4). AQR.
9. Société Générale Prime Services. SG Trend Index.
10. Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). “Time Series Momentum.” Journal of Financial Economics, 104(2), 228-250. sciencedirect.com.
11. Hurst, B., Ooi, Y. H., & Pedersen, L. H. (2017). “A Century of Evidence on Trend-Following Investing.” Journal of Portfolio Management, 44(1), 15-29. AQR.
12. Kim, A. Y., Tse, Y., & Wald, J. K. (2016). “Time series momentum and volatility scaling.” Journal of Financial Markets, 30, 103-124. sciencedirect.com.
13. Huang, D., Li, J., Wang, L., & Zhou, G. (2020). “Time series momentum: Is it there?” Journal of Financial Economics, 135(3), 774-794. sciencedirect.com.
14. Moreira, A., & Muir, T. (2017). “Volatility-Managed Portfolios.” Journal of Finance, 72(4), 1611-1644. onlinelibrary.wiley.com.
15. Cederburg, S., O’Doherty, M. S., Wang, F., & Yan, X. S. (2020). “On the performance of volatility-managed portfolios.” Journal of Financial Economics, 138(1), 95-117. sciencedirect.com.
16. Faber, M. T. (2007). “A Quantitative Approach to Tactical Asset Allocation.” Journal of Wealth Management, 9(4), 69-79. SSRN.
17. Zakamulin, V. (2014). “The real-life performance of market timing with moving average and time-series momentum rules.” Journal of Asset Management, 15(4), 261-278. link.springer.com.
18. Sharpe, W. F. (1975). “Likely Gains from Market Timing.” Financial Analysts Journal, 31(2), 60-69. tandfonline.com.
19. Barber, B. M., & Odean, T. (2000). “Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors.” Journal of Finance, 55(2), 773-806. PDF.
20. Graham, J. R., & Harvey, C. R. (1996). “Market timing ability and volatility implied in investment newsletters’ asset allocation recommendations.” Journal of Financial Economics, 42(3), 397-421. sciencedirect.com.
21. S&P Dow Jones Indices (2026). SPIVA U.S. Scorecard, Year-End 2025 and U.S. Persistence Scorecard.
22. Kenneth R. French Data Library. Fama/French daily research factors (calculator data: value-weighted U.S. market total return and one-month T-bills, July 1926 onward).