One year is a short test
In one-year windows, the market lost purchasing power 30% of the time. The worst year started in Jun 1931; the best started in Jun 1932.
What did an investment in the U.S. stock market actually earn between two dates, with dividends, without them, and after inflation?
Pick any two months from Jan 1871 to Oct 2026. The calculator shows the ending value, annualized return, the part that came from dividends, the inflation-adjusted result and the worst decline along the way.
Results describe the past. They are not a forecast or a recommendation.
CALCULATOR
Start and end months are both included. The investment is made at the start month's index level and valued at the end month's level.
RESULTS
The most recent months use the last published dividend rate, because dividend data is reported with a lag.
GROWTH OF THE INVESTMENT
Logarithmic scale: equal vertical distances represent equal percentage changes, so early decades stay readable.
WHAT THE RECORD SHOWS
Over the full 156-year record, the index returned 9.4% a year with dividends reinvested, or 7.1% a year after inflation. Price change alone contributed 4.9% a year. The table shows how much individual periods varied around that average.
| Holding period | Periods tested | Worst | Median | Best | Lost purchasing power |
|---|---|---|---|---|---|
| 1 year | 1,858 | -58.1%from Jun 1931 | 8.9% | 151.3%from Jun 1932 | 30.4% |
| 5 years | 1,810 | -13.2%from Sep 1929 | 7.4% | 33.3%from Sep 1924 | 18.6% |
| 10 years | 1,750 | -5.9%from Mar 1999 | 7.1% | 20.0%from May 1920 | 11.0% |
| 20 years | 1,630 | -0.2%from Jun 1901 | 6.9% | 13.6%from Apr 1980 | 0.1% |
| 30 years | 1,510 | 1.9%from Jun 1902 | 6.8% | 11.2%from Jun 1932 | 0.0% |
In one-year windows, the market lost purchasing power 30% of the time. The worst year started in Jun 1931; the best started in Jun 1932.
Across 20-year windows, real annual returns ran from -0.2% to 13.6%. That spread still makes a large difference to an ending balance.
Neighboring rolling periods share most of their months, so 1,630 twenty-year windows contain far fewer than 1,630 independent observations. Treat these as a description of the record, not as probabilities.
CALCULATION METHOD
Every result on this page comes from the same few rules, applied to the same monthly dataset.
Each month earns the change in the index level plus one twelfth of the annual dividend rate, and that income is reinvested at the month's level.
The ending index level divided by the starting level. This is the number most charts show, and it leaves out dividends entirely.
The dividends-reinvested value divided by the change in the Consumer Price Index, stated in start-month dollars.
The constant yearly rate that turns the starting value into the ending value. It is shown only for periods of at least twelve months.
The largest fall from a previous high to a later low in the monthly dividends-reinvested series. Daily declines inside a month can be deeper.
Taxes, fund expenses, trading costs and cash drag are not included, so a real investor's result would usually be somewhat lower.
DATA SOURCE
Index levels, dividends and CPI come from Professor Robert J. Shiller's U.S. stock market dataset (shillerdata.com), the series behind the CAPE ratio and Irrational Exuberance. This copy was retrieved on 2026-10-14.
Three features matter when reading results. Index levels are monthly averages of daily closes, not month-end closes. Before 1957 the series is the S&P Composite and its reconstructions, not the 500-stock index launched that year. Dividends are published with a lag, so the latest months reuse the last reported dividend rate, and the most recent month can be preliminary until the dataset is next updated. Results can therefore differ slightly from calculators built on month-end prices.
IMPORTANT CONTEXT
This calculator is for research and education. Past returns do not guarantee future results, and nothing here says whether an investment is suitable for you.
Read: What Is Backtesting?