Everyone repeats the same sentence: over the long run, the market goes up. It happens to be true of the last hundred years in the largest markets, and it's nearly useless as guidance, because it describes a destination and tells you nothing about the road.
The road is what makes people sell. Nobody abandons a plan because the average annual return disappointed them. They abandon it in month fourteen of a fall, when the number on the screen is 40% smaller than the number they put in and there's no visible reason to believe it'll ever come back.
So the history worth knowing isn't the average. It's three other numbers.
Drawdown: the number that ends plans
A drawdown is the fall from a portfolio's highest point to its lowest point before it recovers. The maximum drawdown is the worst one on record, or in plainer terms, the deepest hole an investor in that asset had to sit in.
Here are the rough peak-to-trough falls for a broad US stock index, in nominal terms:
| Episode | Approximate fall | Time to regain the old peak |
|---|---|---|
| 1929–1932 | about −86% | roughly 25 years in nominal terms |
| 1973–1974 | about −48% | about 7 years |
| 2000–2002 | about −49% | about 7 years |
| 2007–2009 | about −57% | about 5 years |
| Early 2020 | about −34% in five weeks | about 5 months |
| 2022 | about −25% | about 2 years |
These are round numbers, and they vary with the index, the currency and whether dividends are counted. It's the pattern that matters: falls of a third are ordinary, falls of a half happen several times a century, and one fall took most of a working life to undo.
The British record is not a copy of the American one, and the differences are instructive. The FTSE All-Share fell roughly −73% in real terms across 1972–74, a deeper real fall than anything the US endured outside 1929, driven by the oil shock, a three-day week and inflation running above 20%. It fell about −45% in 2000–03 and about −45% again in 2007–09. And because the FTSE 100 carries a heavy weight in oil, banks, mining and tobacco rather than technology, the two markets diverge for years at a time: the UK's 2000–03 fall was shallower than the Nasdaq's and its recovery from 2009 far slower than America's. The headline index of the country you happen to live in is not the market. It's one sector bet among many.
The question worth asking: Not "what return do I need?" but "what fall can I sit through without selling?" A portfolio you abandon at the bottom returns far less than a duller one you keep. Your tolerance is a real input, not a weakness to be corrected.
Recovery time is a separate number
Depth and duration are independent, and duration is the one that gets underestimated. 2020 fell hard and was over in months. 2000–2002 fell about the same distance and took the better part of a decade.
If you're still saving, a long recovery is quietly a gift, because every contribution during it buys more. If you're withdrawing, it's the central risk of your retirement, because the same fall now has to be paid for by selling. That asymmetry is why the same historical episode is a footnote in one plan and the whole story in another.
Real returns are the only ones you spend
A 1970s portfolio that gained 6% in a year when inflation ran at 11% lost 5% of its purchasing power. The statement was accurate and the money still bought less. British readers should scale that up rather than down: UK inflation peaked near 25% in 1975, and sterling lost purchasing power far faster through that decade than the dollar did. The same arithmetic, applied to a harsher input, is why UK-based safe-withdrawal research lands on lower numbers than Bengen's American 4%.
That's why the 1970s deserve more attention than they usually get. There was no dramatic single-day crash to remember, and yet inflation quietly took a large share of a decade's savings. So always check whether a chart is real or nominal. Nominal charts flatter every long-term result, and the gap is widest in exactly the decades that hurt.
What diversification actually did
Diversification usually gets explained as a free lunch, which sets people up for disappointment. What a century of history actually shows is narrower and more valuable:
- It reduces the depth of the hole, not the existence of it. A mixed portfolio of stocks and bonds fell meaningfully less than stocks alone in most crises, and it still fell.
- Currency is a second axis nobody chose deliberately. A UK investor holding US equities was cushioned in 2008 because sterling collapsed against the dollar, and punished in the years when it recovered. An American investor holding domestic assets simply has no exposure to this. If you are British, decide whether your unhedged foreign holdings are a diversifier or a second bet you never intended to place.
- It buys behavior. Its real product is a fall shallow enough that you keep contributing, and that's worth more than a slightly higher expected return you never collect because you sold.
- Correlations break at the worst moment. In a genuine panic, things that normally move apart move together for a few weeks. Diversification helps over years, not in the third week of a crisis.
- 2022 was the reminder. Stocks and bonds fell together, because both were repricing to the same thing: interest rates. Anyone who believed bonds were an unconditional hedge learned that they hedge recessions, not inflation.
Backtests are evidence, not prophecy
Running a portfolio through past returns tells you something no forecast can: this actually happened, and here's roughly what it would have felt like. That's genuinely valuable, and it comes with limits worth carrying with you.
- Survivorship. The century of data everyone quotes belongs to the markets that won. Some markets in that same period were closed, nationalized or wiped out. It is worth remembering that in 1900 the London Stock Exchange was the largest in the world and Britain the dominant market; a plausible investor of that era would have been diversified into Russian and Argentine railway bonds.
- One sample. A hundred years sounds enormous. Counted as independent 30-year retirements, it's three, and they overlap.
- The future has never been obliged to rhyme. The starting conditions, meaning valuations, rates and demographics, differ every time.
None of that makes history worthless. It makes it a source of scenarios rather than probabilities: what a −50% fall does to your plan, whether you'd still be contributing in year four, what a decade of flat real returns does to your date of freedom.
How to use it
Take the portfolio you actually hold, put it through the Market Time Machine, and start it in the years you'd rather not think about: 1929, 1973, 2000. Then ignore the final figure for a moment and read the middle of the chart instead. Find the deepest point, see how long the line stayed underwater, and ask yourself honestly whether you'd still have been buying there.
If the answer is no, the change worth making isn't a better forecast. It's a portfolio whose worst year is one you can live through, held long enough for a century of history to do what it did.
Ready to Master Your Money Habits?
Sitting through a fall is a habit long before it's a strategy, and the habits that hold are the ones you set up in a calm month. Take our habit assessment to see how your plan would handle a bad year and what to put in place before one arrives.