TALLIVOWhat the Stock Market Returned, by How Long You Held It
Every rolling holding period in 1,867 months of history — inflation-adjusted, dividends reinvested. Not an average: the whole distribution, because the range is the finding.
Annualised real total returns. Every window with a complete period is counted — from 1,855 one-year windows down to 1,507 thirty-year ones.
The median hardly moves. The range collapses.
This is the finding, and it is not the one most people expect. Holding stocks longer does not raise your expected return — the median is 8.8% over one year and 6.8% over thirty, which is essentially the same number.
What changes is the spread. Over one year, outcomes ranged 209 percentage points from worst to best. Over thirty, just 9. Time does not improve the average; it narrows the distance you are likely to land from it. That is the whole practical case for a long horizon, and it is a claim about risk, not return.
A decade is not a safe horizon
Ten years feels long. In this record only 89% of ten-year windows finished positive after inflation, and the worst began in 1999-03 — losing 5.9% a year for a decade. That is recent enough that plenty of people investing today lived through it.
Twenty years gets you to 99.9%. Twenty-five and thirty reach 100% — every window in 156 years of history was positive, with the worst thirty-year stretch still returning 1.9% a year in real terms.
Where 7% comes from, and what it actually means
Projection tools across this site — our own compound interest calculator included — default to 7%. This table is why: it sits near the median at every horizon, and 46.9% of 30-year windows beat it.
Which is the point worth internalising. A 7% assumption is roughly a coin flip, not a floor. Run your own numbers against the real record in the investment backtest.
Cite or republish this study
These figures are free to reuse — in an article, a newsletter, a class, or a report — with attribution and a link. No permission needed and no paywall.
Suggested citation
Tallivo, “Stock Market Rolling Returns by Holding Period (1871–present),” updated July 1, 2026. https://tallivo.com/data/stock-market-rolling-returns
Key findings
- The median real return barely changes with horizon — 8.8% over one year, 6.8% over thirty — but the range collapses from 209 percentage points to 9.
- Only 89% of 10-year windows were positive after inflation; the worst began 1999-03 and lost 5.9% a year.
- All 1,507 thirty-year windows were positive, the worst still returning 1.9% a year in real terms.
- A 7% assumption is close to a coin flip rather than a floor: 46.9% of 30-year windows beat it.
The data
Download the full dataset (CSV)
All 50 states with the federal, FICA and state components broken out. Generated by the same engine that renders the table on this page, so the file and the page cannot disagree.
Chart image
Open the shareable chart (PNG, 1200×630)
Free to republish alongside a credit to Tallivo and a link to this page. Please do not alter the figures in the image.
Dates & method
Published July 25, 2026 · Figures last verified July 1, 2026.
Full methodology · how we check these numbers · report an error
Writing something and need a figure we have not published — a different salary, a specific state, a filing status? Ask and we will run it.
Common questions
What is a rolling return?
How long do I need to hold stocks to not lose money?
Is a 7% return realistic?
Why do the medians barely change but the range collapse?
Are these real or nominal returns?
Do these figures include fees and taxes?
How we calculate this
A monthly total-return index is built from S&P Composite price and dividends, then deflated by CPI. For each holding period we compute the annualised return from every starting month with a complete window, and report the distribution rather than a single average.
Index returns before fees and taxes, US large-cap only. This is also the record of the most successful stock market of the last century — precisely the sample most likely to flatter. The construction is unit-tested against known history; see the backtest for the method and the interactive version.
Historical index returns 1871-01 to 2026-07, before fees and taxes. Past performance does not predict future results. Not financial advice.