One Number, Two Very Different Meanings
Most discussions of stock market returns settle on a comforting figure somewhere around 7% a year, adjusted for inflation and with dividends reinvested. That number is real, but it is also easy to misuse, because "7%" can mean two different things that happen to converge only over very long stretches of time.
One version is the simple average of individual yearly returns — add up all the one-year results and divide by the number of years. Over the past century, that arithmetic average comes out closer to 9%, not 7%. The other version is the compounded annual growth rate — the single steady rate that, applied every year, would take your starting dollar to your actual ending balance. That figure is the lower one, closer to 7%.
The gap between them is not a rounding error; it is a mathematical consequence of volatility. A 50% gain followed by a 50% loss averages to 0% arithmetically, but it actually leaves you with 25% less money than you started with. Stitch enough up-and-down years together and the compounded result always drifts below the simple average. Separately, a data set stretching back to the 1870s puts the simple average near 8.6% a year and the compounded average near 7.1% — close enough to the century-long figures to suggest this isn’t an artifact of one particular data window, but the underlying arithmetic of volatile returns.
This matters because the "expected return" for any single year is closer to 9%, while the return you should expect to have actually earned after compounding over many years lands closer to 7%. Neither number is wrong; they’re just answering different questions.
How the Range of Outcomes Narrows — But Never Disappears
The more useful way to think about stock returns isn’t a single number at all, but a distribution: a full range of what has historically happened, and how often. Widening the time horizon changes the shape of that range substantially.
| Holding period | Historical average return (annualized) | Chance of a negative annualized outcome | Observed range of outcomes |
|---|---|---|---|
| 1 month | about 0.7% expected | not isolated in this data, but swings are large | single months have ranged from roughly -26.5% to +50.3% |
| 1 year | about 9.2% expected | roughly 17% chance of losing 10% or more | individual years have ranged from about -37% to +53% |
| 5 years | shifts toward mid-to-high single digits annualized | about 10% of periods finished below -4% annualized | about 26% of periods exceeded 12% annualized |
| 10 years | about 7.0% annualized | about 13% chance of a negative annualized result | includes genuinely weak decades, such as the 2000s |
| 20 years | about 6.9% annualized | no 20-year period in the last century ended with a negative U.S. real return | 20-year annualized outcomes have ranged roughly from +0.5% to +13.2% since the 1870s |
The pattern is clear enough: as the holding period lengthens, the odds of ending up with less money than you started with shrink steadily. But the range of plausible outcomes never collapses to a point. Even at 20 years, the difference between a slow-growth outcome and a fast-growth one has historically been the difference between barely outpacing inflation and multiplying your money several times over.
The Mechanics Behind the Pattern
It helps to see why short and long horizons behave so differently, rather than just accepting the numbers. A rough chain of cause and effect connects daily noise to decade-long trends.
flowchart TD A[Daily and annual price swings] --> B[Gains and losses compound unevenly] B --> C[Arithmetic average drifts above compounded growth] C --> D[Longer horizons: fewer periods end negative] D --> E[But dispersion of outcomes stays wide]
Short-term returns are dominated by sentiment, news, and the unpredictable arithmetic of compounding losses — a single bad year can erase several good ones. Over long stretches, the underlying driver of stock returns shifts toward something steadier: corporate earnings growth, plus reinvested dividends, plus whatever inflation does to the purchasing power of the result. That steadier engine is why the odds of a loss shrink with time. It is not, however, why the surprises disappear — the engine’s output still varies enormously depending on where you start and stop measuring.
It’s worth remembering, too, that these figures already assume dividends are reinvested and returns are adjusted for inflation. Strip out either of those adjustments and both the average and the distribution look different — a reminder that "the stock market’s return" is itself a construct built from several separate ingredients: price appreciation, cash distributions, and the eroding effect of inflation.
When the Starting Point Is the Worst Possible One
The distribution tables above describe a random entry point. But some entry points are far from random in hindsight — they’re the peaks right before a crash. Looking at four of the worst moments to have invested over the past century — September 1929, January 1973, March 2000, and October 2007 — makes the short-horizon risk concrete rather than abstract.
At each of these peaks, the one-year result was deeply negative, with losses ranging roughly from 19% to 38%. Five-year results were negative in every case, and ten-year results were negative in most of them; the 1973 and 2000 investors waited more than a decade just to break even. This is sequence risk in its starkest form: the order in which gains and losses arrive matters just as much as their average size, especially for anyone who needs to draw down money during a downturn rather than ride it out.
The encouraging part of this same data is that, given enough time, each of these entry points eventually turned positive — by the 20-year mark, every period with sufficient history had recovered into positive real territory. That is a genuinely striking historical fact about U.S. markets specifically. It is not, however, a guarantee, and it does not travel automatically to other countries or other eras. Japanese equities took 35 years to reclaim their 1989 peak, and some European markets remain well below levels reached decades ago — a useful check against assuming that "stocks always come back" is a law of nature rather than a pattern observed, so far, mostly in one market’s history.
What This History Does Not Tell You
It’s tempting to read a chart like this as a probability calculator for the future, but that overstates what historical distributions can do. They describe what happened across one long, unusually successful century for one country’s stock market. They do not establish that future U.S. returns will resemble the past, nor do they isolate how much of that past result came from valuation expansion, earnings growth, or dividend reinvestment — three drivers that don’t always pull in the same direction going forward. They also say nothing about whether current valuations are high, low, or irrelevant to what comes next; that is a separate question requiring separate evidence, not something this distribution can settle on its own.
The Practical Lesson
None of this is an argument for pessimism or for market timing — the data doesn’t support precise forecasts in either direction. The more durable lesson is behavioral: a headline average return, whether 7% or 9%, describes a center of gravity, not a promise. Short horizons genuinely carry wide, sometimes brutal, dispersion. Long horizons have historically reduced the odds of loss without eliminating the size of the surprises — on either side of the average. Planning around volatility, rather than around a smooth compounding line that rarely exists in reality, is the more honest way to use a century of evidence.
