Investing at All-Time Highs: Better for Five Years, Worse for Twenty

“Don’t be afraid of all-time highs” is good advice. Markets spend a great deal of time near records, selling because an index hit one is a reliable way to sit in cash through a bull market, and the instinct to wait for a dip is the same instinct that keeps people uninvested for years.

The evidence usually offered for that advice says something narrower than the advice does, and the gap is worth seeing because it is a good example of a true statistic losing its qualifiers on the way to becoming a slogan.

The short version

Vanguard’s analysis of the S&P 500 from 1950 to 2025 finds higher average returns after all-time highs at one, three and five years, and lower average returns at ten and twenty. Over twenty years the cumulative figures are 243.1% after a high against 348.8% after all other days, roughly 6.4% a year against 7.8%. So the horizon decides the sign of the result. The behavioural advice survives this intact: staying invested still beats waiting for a dip. What does not survive is calling the data unequivocal.

The claim

On June 3, 2026, Barry Ritholtz wrote: “All-Time Highs: The data is unequivocal. Investing at all-time highs yields better returns than at all other dates.”¹ There is no horizon attached to it, and the supporting detail is outsourced to a quotation from another writer rather than presented.

The version of the statistic that circulates is the short-horizon one, and at one, three and five years it is correct. Extend the window and the same dataset reverses.

What Vanguard found

Vanguard examined S&P 500 forward returns from January 1950 to September 2025, comparing entries made on days that set an all-time high against entries made on every other day. The figures are cumulative returns over each horizon, not annualized:

Horizon	After an all-time high	After all other days	Difference
1 year	9.5%	9.2%	+0.3
3 years	30.2%	28.5%	+1.7
5 years	55.8%	51.9%	+3.9
10 years	108.8%	121.8%	−13.0
20 years	243.1%	348.8%	−105.7

Cumulative average returns, S&P 500, January 3, 1950 to September 24, 2025. The 20-year figures correspond to roughly 6.4% a year against 7.8%.

Vanguard states the reversal plainly: “while the market tends to outperform in the short- to intermediate-term after hitting all-time highs as compared to non-all-time high entry points, this trend reverses over longer holding periods.” It is equally direct about the size of the effect it does find: “the outperformance has been marginal, and it’s crucial to remember there is significant variance within average returns.”²

The mechanism is not mysterious. All-time highs cluster inside long bull markets, so an entry at a high is disproportionately likely to be followed by more of that bull market over the next few years. Push the window out far enough to contain the end of the cycle and the advantage of having bought at an elevated valuation disappears.

One source carries the reversal

This deserves stating rather than burying. Vanguard appears to be the only major publisher extending this comparison past five years. J.P. Morgan’s widely circulated version of the statistic stops at two years, where it agrees that returns after highs are marginally better. The short-horizon result is corroborated many times over; the long-horizon reversal rests on one analysis, from one market, over one historical period. Treat it as the best available evidence rather than as settled.

The same firm states it more carefully

The interesting comparison is internal. Ben Carlson, writing at Ritholtz Wealth Management, uses the identical fact and attaches every qualifier: “your returns are actually better when you invest at those levels than putting your money to work on all other days over 1, 3 and 5 year windows.”³ He bounds the size of it too, calling it “slightly above average results,” and he names the tail: “one of these all-time highs will be THE peak that occurs before a nasty market crash.”

Elsewhere he supplies the mechanism as well, noting that the numbers “make more sense when you consider the new highs tend to cluster during the big, beautiful bull markets.”⁴

Same firm, same underlying data, and a large difference in how much certainty each attaches to it. That is worth noticing before treating any firm as a single voice, and Ritholtz Wealth Management says as much itself: its published disclosures state that blog, podcast and social commentary “reflects the personal opinions, viewpoints, and analyses of the Ritholtz Wealth Management employees providing such comments, and should not be regarded as the views of Ritholtz Wealth Management LLC.”⁵

Credit where it is due on the timing question

Ritholtz has also written that “Academic research and data overwhelmingly reveal that stock selection and market timing do not work,”⁶ which is the same absolute register applied to a claim that holds in a narrower form: both are difficult enough that most investors should not attempt them. Our treatment of what the evidence supports is in Most Stocks Lose to T-Bills and Missing the Market’s Best Days.

What he does not do is hide the tension. He maintains a public page of his own major market calls and labels it himself: “There is plenty of subjectivity to the process, as well as to this admittedly cherry picked list,” and, of the record as a whole, “some were really good, some were way early, and some were damned lucky.” Of a May 2010 move to cash the day before the flash crash, he writes: “I consider this the dumbest of dumb luck.”⁷ He also explains why the firm stopped: broad pronouncements proved “confusing and counter-productive for clients.” That is more candour about one’s own hit rate than most of the industry offers.

What we recommend

Keep the behaviour and drop the certainty. If you have money to invest and a long horizon, an index sitting at a record is not a reason to wait, because the alternative is holding cash while the market spends years making new records. That conclusion survives the full table above.

What the table does not support is using all-time highs as an argument that you are getting a better deal. Over the horizons that actually match a retirement plan, entering at a high has historically produced lower cumulative returns, and the reason is the same reason valuations matter at all. If you want the decision framed properly, it is a lump-sum question rather than a timing question, and we work it through in Lump Sum vs. Dollar-Cost Averaging and The Problem With Buy the Dip.

Frequently asked questions

Is it bad to invest at an all-time high?

No. Over one to five years the historical average has been slightly better than investing on other days, and over any horizon it has beaten holding cash while waiting. Over ten and twenty years the average has been lower, so a record is not a bonus either.

Why do returns after all-time highs reverse over long horizons?

Because highs cluster in bull markets. A short window after a high tends to capture more of that bull market; a long window eventually contains the end of it, and the entry valuation starts to matter.

Should I wait for a dip before investing a lump sum?

Waiting is itself a market-timing decision, and the money sits in cash while you wait. The evidence favours investing promptly, with dollar-cost averaging as a reasonable concession to regret rather than as the higher-returning strategy.

Does this mean market timing works?

It does not. A long-horizon average difference measured across seventy-five years is not a tradable signal, and nothing here identifies which highs precede poor decades. The finding is about how a statistic should be quoted.

Key takeaways

The horizon decides the sign. In Vanguard’s 1950 to 2025 data, returns after all-time highs are higher at one, three and five years and lower at ten and twenty.
Over twenty years the cumulative figures are 243.1% against 348.8%. That is roughly 6.4% a year against 7.8%, in one market over one historical period.
The long-horizon reversal rests on a single publisher. Other widely cited versions of the statistic stop at one or two years.
The advice is right even where the framing is not. Selling or waiting because of a record has been a losing strategy; calling the data unequivocal is what the evidence does not support.
A firm is not one voice. Ben Carlson bounds the same fact to one, three and five year windows, and Ritholtz Wealth Management’s own disclosures say employee commentary is personal rather than firm strategy.

Related guides

The Problem With Buy the Dip why the money was usually supposed to be invested already.
Lump Sum vs. Dollar-Cost Averaging 98 years of evidence, including the third of the time DCA wins.
Missing the Market’s Best Days the other famous market-timing chart, and the rows it never shows.
Good Heuristics, Bad Theorems the series this belongs to, and the scale we grade claims on.

Author disclosure

Summitward publishes portfolio tools and competes for the same readers as the writers quoted here. We have given the behavioural advice its due, because we think it is correct, and confined the criticism to one word in one sentence. Nothing here is personalized financial advice.

Sources

Ritholtz, B., “5 Things I Am Thinking About,” June 3, 2026. Read September 14, 2026. ritholtz.com
Kinniry, F.M., Jaconetti, C., Dinucci, T. and Walker, D.J., “U.S. equities set all-time highs 1 out of every 3 days this quarter,” Vanguard, September 30, 2025. S&P 500, January 3, 1950 to September 24, 2025; figures are cumulative, not annualized. advisors.vanguard.com
Carlson, B., “Investing a Lump Sum at All-Time Highs,” A Wealth of Common Sense, July 4, 2025. awealthofcommonsense.com
Carlson, B., “New All-Time Highs,” A Wealth of Common Sense, April 19, 2026. awealthofcommonsense.com
Ritholtz Wealth Management, “Disclosures,” blog disclosures section. The page exposes no version date; read September 14, 2026. ritholtzwealth.com
Ritholtz, B., “10 Biggest Ideas in ‘How NOT to Invest’,” March 18, 2025, item 6. The same sentence appears in a May 6, 2026 restatement. ritholtz.com
Ritholtz, B., “Major Market Calls,” standing page, last modified August 2021; latest call listed is April 2020. ritholtz.com

Editor’s note

Educational content, not investment advice. The Vanguard figures are averages of cumulative returns from a single index over a single historical period, and averages conceal wide dispersion, which Vanguard says explicitly. The long-horizon reversal comes from one publisher and we have not found it corroborated elsewhere. All quotations were read from the live pages on September 14, 2026; the Ritholtz Wealth Management disclosures page carries no version date, so that citation records only the date we read it.