Buying the S&P 500 at an all-time high: what followed, in four US stock indexes since 1988

How this study happened

Under a post on X, people argued over whether urging others to buy at an all-time high means talking them into buying the top. My quick count of the S&P 500 since 1950 showed the year after an all-time high looking like any other year, while a J.P. Morgan chart for 1988-2020, posted by @polservative, showed it doing better than usual. The difference came mainly from the years they covered: before 1988 the year after a high was weaker than usual, from 1988 it was the same or a little better.

@PGerschmann also pointed out that data from decades ago describes a different market. This study starts in 1988, the first year my data has dividends, and next to the S&P 500 checks the Nasdaq-100, the Russell 1000 and the small-cap Russell 2000. The earlier years are in the method section below, because they give a different answer.

Wojciech, 7 October 2026

The findings at a glance

+11.7% / +10.3%

The S&P 500 on average a year after a record close and after any day, since 1988; the gap is within chance

+2.8% / +9.7%

Same measure for small caps (Russell 2000); with the record dates shifted at random, a gap this large came up in 3 of 100 tries

770

S&P 500 record closes since 1988 with a full year after them

separate runs among those records, counting a new run after at least 3 months without one

Claims and verdicts

Three claims and what the data says

Every claim from every study, in one list →

Claim tested	Finding	Verdict
An all-time high in the S&P 500 is followed by a weaker year	Since 1988, a year after a record close the S&P 500 was up 11.7% on average, against 10.3% after any day, and higher 83% of the time against 81%. The gap is within chance, with dividends too	No: about the same
…in the Nasdaq-100 and the Russell 1000	Nasdaq-100: 22.3% against 17.9%; Russell 1000: 11.9% against 10.5%. Both gaps are within chance	No: about the same
…in small caps (the Russell 2000)	Up 2.8% on average a year after a record, against 9.7% after any day; higher 62% of the time against 71%. With the record dates shifted at random, a gap this large came up in 3 of 100 tries. Counting only the first record of each run: 7.9% against 9.7%, within chance. Check it yourself →	Weaker, on the edge of chance

1 · The S&P 500

The year after an S&P 500 all-time high was an ordinary one.

Since 1988 the S&P 500 has closed at a record 806 times, and 770 of those days already have a full year after them. A year later the index was up 11.7% on average, against 10.3% after any day. With dividends it's 13.8% against 12.7%. Fidelity Investments published a similar result in October 2025, on S&P 500 data from 1950 to 2024: with dividends, 12.7% on average in the year after an all-time high, against 12.6% for other 12-month periods.

Only the first month after a record was softer: 0.2% on average against the usual 0.8%, on the edge of chance. At 3 months the gap is still about 0.6 percentage points but well within chance, and after a year it points the other way.

The S&P 500 after an all-time-high close and after any day, since 1988

Horizon	Average after a record	Average after any day	Higher after a record	Higher after any day	Chance
1 month	+0.2%	+0.8%	61%	64%	5%
3 months	+1.9%	+2.5%	72%	70%	27%
6 months	+4.9%	+5.0%	76%	75%	47%
12 months	+11.7%	+10.3%	83%	81%	23%
3 years	+38.2%	+32.9%	92%	84%	12%
12 months, with dividends	+13.8%	+12.7%	86%	83%	27%

WB Trading Labs · Table 1Price index, no dividends, except the last row. "Chance": how often a randomly shifted calendar of records gives a gap in the averages at least as large, the same way round (see the method).

2 · Big and small caps

Small caps (Russell 2000) did worse than usual after their all-time highs, though it could be luck.

The Nasdaq-100 and the Russell 1000 look like the S&P 500: a year after a record, higher on average than after any day (by 1.4 points in the Russell 1000, 4.3 in the Nasdaq-100), within chance. In the Russell 2000 the year after a record averaged 2.8%, against 9.7% after any day, and after 3 years 15.9% against 28.8%. Counting only the first record of each run, the year after averaged 7.9% against 9.7%, within chance.

Average 12-month change after a record against any day: S&P 500 +11.7% against +10.3%, Nasdaq-100 +22.3% against +17.9%, Russell 1000 +11.9% against +10.5%, Russell 2000 +2.8% against +9.7%. — WB Trading Labs · Fig. 1Average change over 12 months after a record close and after any day. Price indexes, no dividends. The Russell 1000 from 1994.

Four indexes after an all-time-high close

Index	Records	12 months after a record	12 months after any day	Chance	3 years after a record	3 years after any day
S&P 500 (from 1988)	770	+11.7%	+10.3%	23%	+38.2%	+32.9%
Nasdaq-100 (from 1988)	686	+22.3%	+17.9%	15%	+64.5%	+63.0%
Russell 1000 (from 1994)	712	+11.9%	+10.5%	35%	+37.3%	+32.7%
Russell 2000 (from 1988)	575	+2.8%	+9.7%	3%	+15.9%	+28.8%

WB Trading Labs · Table 2"Records": days with a full year after them. Average changes in price indexes. With dividends (the Russell 2000 from 1995) the result is the same: 4.3% against 10.7% after a year.

Check it yourselfChoose the index, the horizon, price alone or with dividends, and see every run of records since 1988 (the Russell 1000 since 1994) →

Method and caveats

What this study measures, and what it doesn't

Data: daily closes from Yahoo Finance: the S&P 500 (price index from 1927, with dividends from 1988), the Nasdaq-100 (from 1985), the Russell 1000 (from 1992) and the Russell 2000 (from 1987, with dividends from 1995).
Record: a close at or above every earlier close in the price index's whole history. I skip each index's first year of history, when almost every day is a record.
Why 1988: dividend data starts that year. Earlier it looked different: from 1950 to 1987, the year after an S&P 500 record averaged 4.6%, against 8.5% after any day (no dividends). The choice of period changes the answer, which is why I show it here.
"Chance": records come in clusters, so I don't compare them with random single days. I shift the whole calendar of records by a random number of sessions, longer than the horizon, 4,000 times, and count how often that gives a gap in the averages at least as large, the same way round. It's a rough gauge, not a formal test.
Many comparisons: four indexes and five horizons make 20 comparisons, so a result on the edge of chance could come from luck alone. For small caps the gap points the same way at every horizon, but the horizons overlap, so that's largely one result. Counting only the first record of each run (23 runs), small caps were up 7.9% on average a year later, against 9.7%, within chance.
Independence: the S&P 500's 770 records are in practice about 20 separate runs, counting a new run after at least 3 months without a record. The averages describe the past and say nothing about what happens this time.
Not tested: markets other than the US, or why the market was at a record. Individual stocks are in a separate study.

What came next

The same test, stock by stock

I then ran the same test for individual stocks: 8,496 US stocks listed since 1998 (the ones whose whole price history is in my data; all stocks in a wider check), including the ones later delisted, for example after a bankruptcy or a takeover. A year after their own all-time high, small stocks did worse on average than after any day (+6.8% against +10.9%), but the typical small stock didn't (median +3.7% against +3.4%). The stock-by-stock results →

New studies go up on X first.

Follow @WojciechBernPL

All studies

After an all-time high, the S&P 500's next year looked like any other - for small caps it didn't