What Is a Time Horizon in Investing? 98 Years of Stock, Bond and Cash Data

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-08 · Updated 2026-11-02 · Reviewed 2026-11-01

A time horizon in investing is the length of time until you need to spend the money. A longer one gives a falling market more years to come back before you have to sell, which is why it shapes how much risk a plan can carry.

Most explanations stop at “short, medium and long.” This page measures the difference instead: every 1-, 3-, 5-, 10-, 15- and 20-year stretch of U.S. stock, bond and cash returns from 1928 through 2025, and one $50,000 goal followed through real years.

How it works

The clock that matters is the one ending on the day you withdraw. Money you will spend in eighteen months and money you will not touch for twenty-five years can sit in the same account, but they face completely different risks.

A fall in prices is only a paper loss until you sell. With years left, the portfolio has time to recover before anything is spent. With months left, whatever the account is worth on the withdrawal date is what you get.

So length changes the question you are answering. For short money the question is how much could be gone on the day it is needed. For long money it is whether the money keeps ahead of rising prices over decades, and cash can fail that test quietly.

The usual bands are rough. Most guides call under three years short, three to ten medium and over ten long. The figures further down show why those cutoffs are a starting point, not a rule.

You have several

An emergency reserve is needed within days. A house deposit might be three years away, a child’s college twelve, and retirement thirty, after which the money must last another thirty.

Putting all of them into one mix fits none of them. The reserve ends up too exposed to a fall, and the retirement money sits too cautiously for decades.

The fix is bookkeeping, not forecasting: a separate account or a clearly labeled bucket for each goal, each with a mix set by its own date.

One goal can hold several dates. Four years of college is four separate withdrawals, so the last payment has three more years to run than the first; a mix built for the first payment is too timid for the last one.

A worked example

Take $50,000 saved for a home deposit due in two years, invested on 1 Jan 2007. The returns below are full calendar years from the Damodaran data set, so the arithmetic can be redone line by line.

In stocks, the deposit was $16,536 short on the day it was needed, a loss of 33.07%. In T-bills it had grown by $2,971; in 10-year Treasuries, which rallied in 2008, it had grown by $16,182.

Waiting was the only cure, and it took four years. The stock version was still below $50,000 at the end of 2011 and first passed it at the end of 2012, at $57,255.

Change only the horizon and the outcome flips. Left in the S&P 500 until the end of 2025, a 19-year horizon, the same $50,000 became $342,545. In T-bills it became $66,444, and in 10-year Treasuries $84,941. Same start date, same money; the only thing that differed was when it was needed.

One caution on reading this: a single starting year is one draw from history, chosen here because it includes a crash. The full record follows.

The original data

What was measured: calendar-year total returns for the S&P 500 (dividends reinvested), 3-month Treasury bills and 10-year Treasury bonds, 1928 to 2025, from Aswath Damodaran’s NYU Stern data set, downloaded 1 Nov 2026. Every run of consecutive years was tested at each length, so there are 98 one-year periods, 94 five-year periods and 79 twenty-year periods.

A period lost money if $1 at the start was worth less than $1 at the end. “After inflation” divides each year’s return by that year’s change in consumer prices from the same file.

The share of losing periods falls steadily with length. About one calendar year in four lost money; 11 of 94 five-year stretches did; a single fifteen-year stretch did, 1929 to 1943, by 0.20% a year.

Inflation moves every line up. After inflation, 31.63% of one-year periods, 23.40% of five-year periods and 12.36% of ten-year periods lost purchasing power. The worst real ten years ran from 1999 to 2008, at -3.81% a year, and even fifteen years was not always enough: 1965 to 1979 lost 0.56% a year in real terms.

Bonds and bills behaved differently. 3-month T-bills never had a losing period in dollars at any length. 10-year Treasuries lost money in 20 of 98 single years, including 17.83% in 2022, and in 5 of 94 five-year periods, but in none of the ten-year ones.

The range narrows as the horizon lengthens, and that matters more than the average. The chart below plots the worst, median and best annual rate for stocks at each length.

The middle barely moves; the edges do. The median stays between 10.40% and 15.22% a year at every length, while the spread from worst to best shrinks from over 96 points for a single year to about 15 points over twenty.

Cash pays for its safety over long horizons. Over 20 years the median T-bill return was 3.38% a year against 10.91% for stocks, and stocks finished behind T-bills in none of the 79 twenty-year periods, but in 32 of the 98 single years. After inflation, T-bills fell behind in 43 of 89 ten-year periods.

Check: the stock series was compared with Yahoo’s S&P 500 Total Return index year-end closes for 1989 to 2025. The 37 annual returns differed by a median 0.16 points and at most 0.53 points.

Limits, stated plainly: these are calendar years, so falls inside a year (such as late 2008 into March 2009) are deeper than shown. The periods overlap, so 79 twenty-year periods hold only about four fully separate ones. It is one country’s market, and no tax or fees are included.

The question: The deposit is needed in two years. Invest it for growth?

The answer is that two years is not enough time for a fall to recover. A 20% decline needs 2.90 years at 8%, which is longer than the horizon — so the money would be spent at whatever it happened to be worth. A horizon shorter than the recovery time makes growth an inappropriate objective, regardless of how attractive the long-run figures look.

It shortens every year

A thirty-year horizon becomes a twenty-year one after a decade, and a mix set at the start does not adjust itself. In the range chart above, that is a move from the narrow bottom rows toward the wide top ones.

A glide path automates exactly this step. Someone running their own portfolio has to do it on purpose, because nothing prompts it.

The recovery arithmetic sets the pace. A 20% fall needs a 25% gain to get back, a 40% fall needs 66.67%, and at an assumed 8% a year those take 2.90 and 6.64 years. The drawdown recovery calculator runs other cases.

New money is the cheapest way to follow a shortening horizon. Sending contributions to the safer side does most of the shift without selling, and so without trading costs or, in a taxable account, a tax bill.

Retirement is not the end of it

The date you retire is when withdrawals start, not when the horizon ends. A portfolio may have to pay out for thirty years after it, so only the next few years of spending face a short horizon.

That is why the most cautious point of a lifetime plan is rarely retirement day itself. The later decades of spending still have a long horizon. After inflation, 3-month T-bills lost purchasing power in 28 of the 79 twenty-year periods since 1928, with a median real return of 0.56% a year.

When it fails

Most often, a date moves while the mix stays put. A job loss, an early retirement or a sudden decision to buy a house pulls a withdrawal forward by years, and nothing in the account signals the mismatch until the money is needed.

The second is treating five years as a safe line. It cut the share of losing stock periods to 11.70% of the history, not to zero; 1928 to 1932 lost 49.34% in total.

A third is running one mix across several horizons. It fits none of them.

A fourth is confusing horizon with tolerance. Horizon is arithmetic about dates; risk tolerance is how you behave when the balance drops. A long horizon helps only if you do not sell in the middle of it.

A fifth is forgetting inflation on long money. Cash with a thirty-year horizon never shows a loss on a statement, but it can lose ground to prices year after year.

And a sixth is reading the past as a promise. No twenty-year stock period from 1928 to 2025 lost money, but 79 overlapping periods from one market are a small sample of what can happen.

Questions people ask

What are short, medium and long time horizons?

Short (roughly under three years), medium (three to ten) and long (over ten). The cutoffs are conventions; in the 1928 to 2025 data, stock losses still appeared in 5.62% of ten-year periods.

Is five years a long enough horizon for stocks?

It cut the odds sharply but did not remove them. 11 of 94 five-year S&P 500 periods since 1928 lost money, and 22 of 94 lost ground to inflation.

How does time horizon affect risk?

It changes which risk dominates. Over short horizons the risk is a fall just before the money is needed; over long ones, it is growth too slow to beat prices.

Can my time horizon change?

Yes, every year and sometimes overnight. It shortens by a year annually, and life events can move the date, which leaves a mix built for a plan that no longer exists.

Related

Risk tolerance is the behavioral half of the same decision. Turning each horizon into stock and bond weights is the job of asset allocation, and the cash line in the data above is Treasury bills. And a glide path follows a shrinking horizon down on a fixed schedule.

This page is educational, not financial advice. Test every idea on your own charts before risking money.