Skip to content
Menu
Finabase Learn

Portfolio and risk · Long-run planning

Time diversification: are stocks safer in the long run?

What happens to stock risk as the holding period lengthens: annualized returns narrow while the range of dollar outcomes widens. SPY data from 1993 to 2026.

On this page
  1. The idea
  2. How it works
  3. Using it
  4. Where it fails or misleads
  5. Compared with its neighbours
  6. How Finabase does it
  7. Frequently asked questions
  8. Summary
  9. Check your understanding

Time diversification is the idea that stocks become less risky the longer they are held. Bad years are offset by good ones, the argument runs, so an investor with decades ahead can hold stocks with little danger, and one with only a few years cannot. It is among the most widely repeated ideas in personal finance, and it underlies the advice to hold more in stocks when young.

It is half right. One measure of risk does shrink with time: the spread of the average annual return. Another grows: the spread of the amount of money the investor ends up with. Which of the two is "risk" depends on what the investor cares about, and since people spend dollars and not annualized percentages, the second deserves at least as much weight. This lesson shows both in the same data.

The idea

Suppose stock returns in each year are independent draws with a volatility of 15%. Average many of them and the average becomes more predictable: its standard deviation is 15% divided by the square root of the number of years. Over 25 years that is 3%. This is the law of large numbers, and it is why long-run annualized returns sit in a narrow band.

Now look at wealth. Wealth after 25 years is the product of all 25 returns. An annualized return that is one point lower for 25 years leaves about 21% less money. Small differences in the average, compounded over a long time, are large differences in the result. The standard deviation of the total return grows with the square root of time.

Both statements are true together. The average gets more certain; the destination gets less certain.

Return a year: the range narrows as the holding period lengthens −60% −40% −20% 0% 20% 40% 60% 1 year 5 years 10 years 20 years 56.2% median 14.4% −43.4% 28.6% median 11.4% −6.7% 16.5% median 8.8% −3.4% 11.3% median 8.8% 4.7% Holding period Annualized total return: worst, median and best period SPY, every overlapping period, March 1993 to October 2026. Before inflation.
Worst, median and best annualized total return of SPY over every overlapping period of each length, March 1993 to October 2026. Before inflation.

The first figure is the one usually shown in support of time diversification. The range narrows dramatically.

Dollars: the range widens as the holding period lengthens $0k $20k $40k $60k $80k $100k 1 year 5 years 10 years 20 years $15,622 $5,658 $35,151 median $17,191 $7,082 $46,243 median $23,281 $7,044 $85,815 median $54,462 $25,030 Holding period What $10,000 became: worst, median and best period SPY total return, every overlapping period, March 1993 to October 2026. Before inflation.
Worst, median and best value of $10,000 invested in SPY over every overlapping period of each length, March 1993 to October 2026. Total return, before inflation.

The second figure uses the same periods and reports dollars. The range widens.

How it works

With SPY's volatility of 14.7% over this sample:

Holding period Std. deviation of the annualized return Std. deviation of the total log return
1 year 14.7% 14.7%
5 years 6.6% 32.9%
10 years 4.7% 46.6%
20 years 3.3% 65.9%

A worked example: every holding period since 1993

The data: SPY's monthly total returns, March 1993 to October 2026 (404 months), and every overlapping period of 1, 5, 10 and 20 years within them.

Holding period Periods Annualized return: worst / median / best $10,000 became: worst / median / best Gap, best − worst Periods that lost money
1 year 393 −43.4% / 14.4% / 56.2% $5,658 / $11,445 / $15,622 $9,964 17.8%
5 years 345 −6.7% / 11.4% / 28.6% $7,082 / $17,191 / $35,151 $28,069 14.2%
10 years 285 −3.4% / 8.8% / 16.5% $7,044 / $23,281 / $46,243 $39,199 8.4%
20 years 165 4.7% / 8.8% / 11.3% $25,030 / $54,462 / $85,815 $60,785 0.0%

Three things can be read from the table.

The chance of loss falls. From 17.8% of one-year periods to 8.4% of ten-year periods to none of the twenty-year periods. This is the true part of time diversification.

The worst case does not improve for a long time. The worst ten-year period, beginning March 1999, turned $10,000 into $7,044. That is a larger dollar loss than most single bad years produce, and it took a decade. The worst five-year period lost about as much.

The spread of outcomes grows. Two investors who each held for twenty years ended with $25,030 and $85,815, depending only on when they began (April 2000 or June 2006). Both earned a positive return, and one has more than three times as much as the other. If the goal was a particular sum, the long horizon did not make reaching it more certain.

After inflation

Holding period Annualized real return: worst / median / best $10,000 became, in constant dollars Periods that lost purchasing power
1 year −43.4% / 12.2% / 52.2% $5,658 / $11,215 / $15,216 20.9%
5 years −9.1% / 8.7% / 25.6% $6,216 / $15,171 / $31,257 25.9%
10 years −5.9% / 6.1% / 14.5% $5,454 / $18,151 / $38,829 10.9%
20 years 2.6% / 6.3% / 8.6% $16,584 / $34,207 / $51,723 0.0%

In real terms a quarter of all five-year periods lost purchasing power, more than the share of one-year periods that did. The worst decade cost 45% of the investment's real value. And measured from month-end highs, SPY's total return after inflation spent as long as 152 months below an earlier peak (74 months before inflation).

Using it

  • Separate two questions. "How likely am I to lose money?" has an answer that improves with time. "How uncertain is the amount I will end with?" has an answer that gets worse. A plan should state which one it is solving.
  • Use the right reasons for holding more stock when young. Bodie, Merton and Samuelson pointed to human capital: a young worker's future earnings are a large, relatively safe asset, so the financial portfolio can carry more risk, and a young worker can respond to a bad market by working longer or saving more. These reasons hold. "Time makes stocks safe" is not needed and is not true. Glide paths: how a portfolio’s stock share changes with age builds on the sound version.
  • Match assets to the date of the need. Money needed in three years faces a real chance of loss in stocks (about one period in seven at five years). Money needed in twenty years has, in this record, not lost. Money needed as a specific amount on a specific date is better matched with bonds maturing then.
  • Plan on a range. For a twenty-year goal, the worst and best cases in this sample differed by a factor of more than three. Saving enough for the lower end is the prudent response; expecting the median is not.
  • Remember the investor, not only the investment. A horizon is only as long as the investor's ability to stay invested. Someone who would sell after a 40% fall does not have a twenty-year horizon, whatever their age.

Where it fails or misleads

  • The sample is one country and one fortunate period. These 33 years contain fewer than two non-overlapping twenty-year periods. The 165 "periods" in the table share most of their months and are not independent observations. Other markets have had losing stretches of twenty years and more: Japanese stocks did not regain their 1989 peak for over three decades.
  • Insurance prices contradict the claim. Bodie pointed out that if stocks became safer with time, the cost of insuring a stock portfolio against earning less than the risk-free rate would fall as the horizon lengthened. Option pricing says it rises.
  • Samuelson's argument. An investor whose attitude to risk does not depend on wealth should choose the same stock share for any horizon when returns are independent. The extra years add chances to recover and equal chances to fall further.
  • Mean reversion is uncertain. If poor decades tend to be followed by good ones, long-run risk is lower than the √T rule says, and US data show some evidence of that. Pástor and Stambaugh argued that once uncertainty about the true average return is included, stocks are more volatile over long horizons, not less.
  • "Never lost over twenty years" is a statement about the past. It is the product of a high average return in the country that has been studied most because it did best.
  • Time works against the unlucky. A longer horizon also means more time for a bad outcome to compound.

Compared with its neighbours

Idea Claim Status
Time diversification Stocks are less risky over long holding periods True for annualized returns and odds of loss; false for dollar outcomes
Diversification Spreading across assets lowers risk at a point in time Well established
Human capital argument The young can bear more risk because of future earnings Widely accepted
Dollar-cost averaging Investing gradually spreads the entry price Lowers timing risk, usually lowers return
Sequence of returns risk The order of returns matters when there are cash flows Well established

Diversification across assets works because the assets are held at the same moment and their movements partly cancel. Years are not held at the same moment. Each year's result is applied to the whole of what the previous years left.

How Finabase does it

Finabase does not compute rolling-period returns, the distribution of outcomes for a holding period or the probability of loss over a horizon.

A symbol's chart at the maximum range shows the record directly: for SPY, the long flat stretch from 2000 to 2013 is visible as well as the rise that followed. The History tab lists prices by date, so the return between any two dates can be worked out, and the Dividends tab supplies the distributions to add for a total return. The comparison view reports price returns over fixed windows up to two years, which is suited to the short end of this table and not to the long end.

Frequently asked questions

Are stocks less risky if you hold them longer?

The probability of ending with a loss has fallen with the holding period in US history. The range of possible ending values has widened. Whether that counts as less risky depends on which of the two matters for the goal.

Has the US stock market ever lost money over twenty years?

Not in the period covered here, and rarely if ever in longer US records on a total-return basis. After inflation there have been stretches of well over a decade with no gain, and other countries' markets have lost over twenty years.

Why should young people hold more stocks, then?

Because most of their wealth is future earnings, which are not exposed to the stock market, and because they have time to adjust saving and working plans after a bad outcome. The length of the horizon matters through those channels.

What is the time diversification fallacy?

The mistaken step from "the average return becomes more predictable over time" to "the final amount becomes more predictable over time". The first is true. The second is the reverse of the truth.

How long should I be prepared to wait for stocks to recover?

In this sample SPY took up to 74 months to regain an earlier month-end high in nominal terms and up to 152 months after inflation. Longer waits have occurred elsewhere.

Summary

Lengthening the holding period narrows the range of annualized returns and widens the range of dollar results. For SPY from 1993 to 2026, twenty-year annualized returns ran from 4.7% to 11.3%, a narrow band, and the corresponding values of $10,000 ran from $25,030 to $85,815, a wide one. The odds of a loss fell to zero in this sample at twenty years, on the evidence of fewer than two independent periods. Time reduces the chance of a bad result without limiting its size, and the better reasons for a long-horizon investor to hold stocks are future earnings and the ability to adapt.

Check your understanding

5 questions on this lesson. Each answer comes with the reason, and nothing is scored.

  1. As the holding period lengthens, what happens to the range of annualized stock returns and the range of ending dollar values?

    1. Both narrow
    2. Annualized returns narrow; ending dollar values widen
    3. Both widen
    Show the answer

    B. Annualized returns narrow; ending dollar values widen The average return becomes more predictable, but small differences in it compound over more years into large differences in final wealth.

  2. An asset has a volatility of 16% a year and independent yearly returns. What is the approximate standard deviation of its annualized return over 16 years?

    1. 16%
    2. 4%
    3. 1%
    Show the answer

    B. 4% Volatility ÷ √T = 16% ÷ √16 = 16% ÷ 4 = 4%.

  3. In the SPY sample no twenty-year period lost money. Why is that weak evidence that twenty-year losses cannot happen?

    1. Because the 165 periods overlap heavily; the sample contains fewer than two independent twenty-year periods from one successful market
    2. Because the data exclude dividends
    3. Because SPY is not a stock fund
    Show the answer

    A. Because the 165 periods overlap heavily; the sample contains fewer than two independent twenty-year periods from one successful market Rolling periods share most of their months. A 33-year record from one country cannot establish what is possible over twenty years.

  4. Which is the stronger reason for a 25-year-old to hold a high share in stocks?

    1. Stocks cannot lose over long periods
    2. Most of their wealth is future earnings, and they can adjust saving and working after a bad market
    3. Volatility falls to zero over time
    Show the answer

    B. Most of their wealth is future earnings, and they can adjust saving and working after a bad market The human-capital argument holds whether or not returns are predictable. The claim that time removes stock risk does not.

  5. Two investors each held SPY for twenty years and ended with $25,030 and $85,815 per $10,000. What does this illustrate?

    1. That long holding periods guarantee similar outcomes
    2. That the starting date still mattered greatly to the final amount, even over twenty years
    3. That one of them traded too often
    Show the answer

    B. That the starting date still mattered greatly to the final amount, even over twenty years Both earned positive returns, yet one ended with more than three times as much. A long horizon did not make the dollar result predictable.

Sources and further reading

  1. Paul A. Samuelson, “Risk and Uncertainty: A Fallacy of Large Numbers,” Scientia, vol. 98 (1963)
  2. Zvi Bodie, “On the Risk of Stocks in the Long Run,” Financial Analysts Journal, vol. 51, no. 3 (1995)
  3. Zvi Bodie, Robert C. Merton and William F. Samuelson, “Labor Supply Flexibility and Portfolio Choice in a Life Cycle Model,” Journal of Economic Dynamics and Control, vol. 16, nos. 3–4 (1992)
  4. Jeremy J. Siegel, Stocks for the Long Run, 6th ed. (McGraw-Hill, 2022)
  5. Ľuboš Pástor and Robert F. Stambaugh, “Are Stocks Really Less Volatile in the Long Run?”, Journal of Finance, vol. 67, no. 2 (2012)
  6. Finabase chart data — monthly bars and dividends of SPY, March 1993 to October 2026; CPI-U (CPIAUCSL) via FRED; statistics computed by Finabase