Myth

Do stocks always go up in the long run?

VerdictIt dependsChecked against the data on 20 November 2026

It depends: SPY, the S&P 500 in one fund, ended higher 10 years later from 92.1% of the trading days we tested as start dates, but not from all of them. Money put in on the worst start day, in early 1999, returned −4.2% a year over the next decade, even with dividends reinvested.

The myth, and why people believe it

"Stocks always go up in the long run" is the first thing most new investors hear, usually as a reason to hold on through a crash. It has a solid base. The S&P 500 has recovered from every fall so far, and a long price chart slopes up from left to right. The weak points are the word "always" and how long "long" has to be.

Two terms first. A start day is the day the money goes in. An annualised return is the steady yearly rate that would turn the starting amount into the ending amount, the same idea as CAGR. A negative annualised return means the money was worth less at the end than at the start.

The test

How we tested this

DataSPY daily closes from EODHD, adjusted for splits and dividends (total return), for every return in the study; the same source for the SPY and SPY 60 / IEF 40 baskets.
Period26 July 2002 to 20 November 2026, the longest period every holding has prices for
BasketsSPY 60 / IEF 40: IEF 40%, SPY 60%, rebalanced yearly; SPY: SPY 100%, held throughout (benchmark)
StudySPY 29 January 1993 to 20 November 2026. A record close is a close at or above every earlier close on dividend-adjusted closes; the first 0 trading days are never records. What $1 bought on each record close was worth 12 months, 3 years, 5 years, 10y later, with dividends reinvested (returns past 12 months are a year), compared with the same windows from every trading day. A fall counts when the fund closed at least 10% below the price paid within the window. Halves split the start dates at the midpoint of the test.
How we testedThis page uses the study's any-day part only: every trading day from the start date is a start day, and its 12-month, 3-year, 5-year and 10-year returns are measured (annualised past 12 months). The record-close part of the study is not used here. The verdict reads the 10-year and 5-year any-day figures against lines set before the test was run. The baskets feed the drawdown chart, the calendar bars and the Clone button.
Not includedTrading costs, taxes and inflation. Other countries' markets, where long losing stretches have lasted longer. Adding money over time instead of investing once.
Verdict rule, set before lookingBusted if the share of 10y windows from any day that gained is below 80.0%; Proven if the worst 10y outcome a year from any day is at least zero and the share of 5-year windows from any day that gained is at least 90.0%; It depends otherwise.

We took SPY from 29 January 1993 to 20 November 2026, with dividends reinvested, and treated every trading day as a possible start day. For each one we measured the annualised return over the next 12 months, 3 years, 5 years and 10 years. Start days too recent to have a full window are left out, so the 10-year test uses start days from 1993 to 2016.

We fixed the verdict before running it. Proven if no 10-year start lost money and at least nine in ten 5-year starts ended higher. Busted if fewer than four in five 10-year starts ended higher. It depends otherwise.

The result

Holding longer helped, but more slowly than the myth suggests, and even 10 years was not always enough.

Every start day of SPY since 1993, by how long it was held

Held for Start days that ended higher Median return a year Low end (one start in ten did worse) Worst start, a year
12 months 81.9% +14.4% −12.0% −47.4%
3 years 83.7% +12.2% −5.8% −17.3%
5 years 84.9% +11.5% −1.1% −8.4%
10 years 92.1% +8.8% +1.7% −4.2%

The surprise is the middle of the table. Waiting 5 years instead of 12 months only moved the share of winning starts from 81.9% to 84.9%, because many 5-year windows began before the 2000 to 2002 crash or ran into 2008. It took 10 years to lift it to 92.1%.

Every 10-year start that lost money came between late 1998 and late 2000, near the top of the dot-com boom, and most start days in that span did lose. Those investors met two crashes back to back: the fall after 2000, then 2008, before ten years were up. Even with dividends, SPY took 6.6 years to climb back above its high of 24 March 2000, which it passed again on 26 October 2006, only to fall below it once more in 2008.

The 2008 crash shows how deep a fall can go on the way. From a start in 2002, SPY fell 55.2% from its peak and spent 4.9 years below it. A drawdown is the fall from the highest value so far to the lowest point after it.

SPY 60 / IEF 40SPY
Drawdown from the previous high, full period. Worst falls: SPY 60 / IEF 40 29.8%; SPY 55.2%-40%-20%0%20032006200920122015201820212024
  • SPY 60 / IEF 40 worst fall 29.8% (9 October 2007 to 9 March 2009), longest time below a previous high 2.5 years
  • SPY worst fall 55.2% (9 October 2007 to 9 March 2009), longest time below a previous high 4.9 years

Full period: Indices Lab backtest, total return, mixes rebalanced yearly, single funds held throughout, 26 July 2002 to 20 November 2026. Hypothetical results: past performance does not predict future returns.

Test your own start date Opens the builder with IEF 40%, SPY 60% against SPY. Free account, no card.

Where the myth has a point

Over long periods the typical result was good. The median 10-year start grew +8.8% a year, and the low-end 10-year start, the one that only one start in ten did worse than, still grew +1.7% a year. Over 12 months the low end was −12.0%. Time did shrink the range of outcomes, a lot.

Single years were mostly up too. SPY fell in only a few full calendar years since 2003, but they included 2008 at −36.8% and 2022 at −18.2%.

SPY 60 / IEF 40SPY
Calendar-year returns, SPY 60 / IEF 40 best +21.9% (2019), worst -17.0% (2022); SPY best +32.3% (2013), worst -36.8% (2008)-40%-20%0%+20%20032006200920122015201820212024
20032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025
SPY 60 / IEF 40+17.7%+8.1%+4.0%+10.5%+7.2%−14.9%+13.2%+12.8%+7.4%+11.1%+16.9%+11.7%+1.3%+7.6%+14.0%−2.3%+21.9%+15.0%+15.9%−17.0%+17.2%+14.7%+13.8%
SPY+28.2%+10.7%+4.8%+15.8%+5.1%−36.8%+26.4%+15.1%+1.9%+16.0%+32.3%+13.5%+1.2%+12.0%+21.7%−4.6%+31.2%+18.3%+28.7%−18.2%+26.2%+24.9%+17.7%

Full calendar years only. Full period: Indices Lab backtest, total return, mixes rebalanced yearly, single funds held throughout, 26 July 2002 to 20 November 2026. Hypothetical results: past performance does not predict future returns.

What it means for your portfolio

  • A long horizon raises the odds; it does not remove the risk. Ten years was usually enough for SPY, but not from the worst starting points, and those came after a long boom.
  • If you need the money on a date, look at the worst rows, not the median. A plan that only works at the median return fails in the bad-luck case.
  • A mix can soften the path. Over the same period a mix of 60% SPY and 40% IEF, a fund of 7 to 10-year US Treasury bonds, rebalanced once a year, fell 29.8% at worst against 55.2% for SPY, and grew 8.3% a year against 11.0%. Less fall, less growth: see stocks vs bonds and max drawdown for the trade-off.
  • Test your own start date. A backtest from one flattering start says little; run yours from several, including late 1999 and late 2007.
Backtest your own portfolio Opens the builder with IEF 40%, SPY 60% against SPY. Free account, no card.

Questions

Do stocks always go up in the long run?

Not always. In our test of SPY since 1993, it ended higher 10 years later from 92.1% of start days. The ones that lost money all started between late 1998 and late 2000.

How long do you need to hold stocks to avoid a loss?

No holding period ended in a gain from every start day in our test. Over 5 years SPY ended higher from 84.9% of start days and over 10 years from 92.1%; the odds rose with time but never reached every start day.

What was the lost decade for stocks?

The 2000s. Most money put into SPY between late 1998 and late 2000 was worth less ten years later, even with dividends, because the dot-com crash and the 2008 crash came within the same decade.

Does the stock market go up every year?

No. SPY ended higher 12 months later from 81.9% of start days since 1993, and it fell in calendar years such as 2008, 2018 and 2022.

Does this mean stocks will go up over the next 10 years?

No. This is one fund's history since 1993, not a forecast. The losing starts came after a long boom, a reminder that the starting point matters.

Build your index Opens the builder with IEF 40%, SPY 60% against SPY. Free account, no card.

About this page. Figures on this page come from Indices Lab’s own data, powered by EODHD, and every number is recalculated after each market close; the text is reviewed by a person before it is published. Indices Lab is a research and educational tool: everything here is hypothetical and informational, not investment advice and not a recommendation to buy or sell any security. Past performance does not predict future results. See our methodology.