What happens after a big year for stocks?
After a strong year many investors worry that a payback is due. A century and a half of returns shows whether big years tend to be followed by weak ones.
Updated Sep 1, 2026
After a big year: +11.2% on average
The S&P 500 returned 15% with dividends in 2025. Since 1872 there have been 53 years of 20% or more; the year after averaged 11.2% and was positive 74% of the time, against 10.9% and 73% for all years. A big year has not, on its own, signalled a bad one to follow.
What came next
First, the average following-year return for each kind of year. Select a tile for every year’s return since 1928.
The year after, by the year before
Every calendar year from 1872 to 2024, grouped by its total return, with the return of the year that followed. Select a column to sort.
| The year before | Latest | Next year, average | Median | Up | Worst | Best |
|---|---|---|---|---|---|---|
Down more than 10% 19 years | – | +13.5% | +21.7% | 74% | −42% | +54% |
Down up to 10% 22 years | – | +12.8% | +15.4% | 77% | −24% | +47% |
Up to 10% 29 years | – | +3.7% | +5.0% | 62% | −39% | +48% |
Up 10% to 20% 30 years; 2025 was one | 2025 | +14.3% | +15.8% | 80% | −12% | +49% |
Up 20% to 30% 31 years | – | +10.7% | +13.2% | 77% | −19% | +38% |
Up more than 30% 22 years | – | +12.0% | +14.7% | 68% | −32% | +38% |
How to read this page
How it’s measured. Calendar-year total returns, with dividends, for the S&P 500 and its predecessors from Robert Shiller’s monthly data since 1871, using December average prices. Each year is grouped by its own return, and the table shows what the following year returned.
Sources. Robert Shiller’s long-run market data.
Calendar years are arbitrary cut-offs, and a few dozen cases per group leave wide error bands. Valuations at the start of a year have mattered more for the long run than last year’s return.
Big years since 1990: 1991, 1995, 1996, 1997, 1998, 1999, 2003, 2009, 2013, 2017, 2019, 2021, 2023, 2024.
This page is information, not investment advice. About our data and methods ›