Does the CAPE ratio actually predict stock market returns?
Short version: almost not at all over a year, and surprisingly well over a decade. Here is the same data the optimists and the skeptics are both looking at — and why they can both be right.
By Dominic Roe · Data Engineer & Business Intelligence Developer
The honest answer
Over the next year, the starting CAPE explains only about 5% of what the market goes on to do — close enough to noise to ignore. Stretch the horizon to ten years and that rises to roughly 28%; to fifteen, about 40%. So CAPE is a poor market-timing signal and a useful long-range planning input. Those are not contradictions — they are the same fact measured over different horizons.
Predictive power grows with the holding period
This is the heart of a long-running argument. One camp says the CAPE has “so little predictive value over any given timeframe” that watching it is pointless. The other says that “when the timeframe is subsequent 10-year returns, the correlation is actually pretty decent.” Both are reading the same numbers — they are just standing at opposite ends of this chart.
- 1 yr5%
- 2 yr8%
- 3 yr10%
- 5 yr16%
- 7 yr19%
- 10 yr28%
- 15 yr40%
- 20 yr41%
The pattern is consistent and it is the whole story: at one year, valuation is swamped by sentiment, news and momentum. As the horizon lengthens, those wash out and the price you paid starts to dominate what you earn. By ten to fifteen years, starting valuation is doing real explanatory work.
You can see the signal — and you can see the noise
Here is every starting month plotted against the real return that followed over the next decade. A cheap start (left) was usually followed by stronger returns; an expensive one (right) by weaker ones — the line slopes down. But look at the vertical spread: at any given CAPE, history delivered a wide range of outcomes. That spread is why the same chart fuels both sides of the debate.
Want to read a specific number off this relationship — what today’s valuation has historically implied for the next ten years, and what that does to a lump sum? That is the job of the expected-returns calculator.
Why it is useless for timing the market
Even a strong long-run relationship tells you nothing about when a richly valued market will give those returns back, or by how much. A high CAPE is like loading a structure beyond its rating: it raises the odds of a problem without telling you the day it fails. US CAPE first cleared 30 in the late 1990s and has spent much of the time since above its own long-run average while the market climbed regardless. Anyone who treated “expensive” as “sell” spent years — sometimes a decade — watching from the sidelines.
The honest reading of the data is humble: valuation moves the odds, it does not set the schedule. That is enough to plan around. It is nowhere near enough to trade on.
Why it still earns its place
A long-run return assumption sits underneath almost every big financial decision: how much you need to save, when you can stop working, and what withdrawal rate is safe in retirement. The default is to plug in the historical average return. But the historical average came from a mix of cheap and expensive starting points — and, as the chart above shows, the starting point matters over the horizons that retirement planning cares about.
Using valuation to temper that assumption is defensible in a way that using it to time entries and exits is not. It is the difference between “returns may be thinner than average over my next fifteen years, so I will save a little more / withdraw a little less” and “the market is high, so I will sell.” The first is what the data supports; the second is what it warns against. Our expected-returns page turns today’s reading into that planning number, and the US valuation dashboard shows where today sits across several gauges.
The caveats that move the “normal” line
The fit explains only part of long-run returns even at its best (R² ≈ 0.28 at ten years), and the “normal” level of CAPE is not fixed. Accounting standards, profit margins, interest rates, the long shift from dividends to buybacks, and a heavier weighting of asset-light technology firms have all plausibly raised the level investors are willing to pay. Measures built on forward earnings look less extreme than the backward-looking CAPE or a trailing P/E.
None of that rescues an expensive market, though. Persistently higher valuations still mean fewer earnings bought per pound, which still points to thinner long-run returns unless earnings grow unusually fast. The caveats widen the band of plausible outcomes; they do not flip the sign. See the methodology for the full assumptions — and treat everything here as information, not investment advice.
Frequently asked questions
Does a high CAPE mean a crash is coming?
No. The CAPE ratio carries almost no information about the next year — in the data it explains only a few percent of what returns actually do over twelve months. Markets can stay expensive, or grow more expensive, for many years. CAPE is a long-horizon valuation gauge, not a timing signal.
Hasn't CAPE been 'wrong' for over a decade?
In the timing sense, yes: US CAPE has sat well above its historical average for most of the period since 2013 while the market kept rising, so anyone who sold on a high reading missed strong returns. That is exactly why we frame it as a planning input rather than a signal. A high CAPE shifts the odds toward lower long-run returns; it does not promise them, and it says nothing about when.
How is 'predictive power' measured on this page?
We fit a simple regression of the real return the market went on to deliver against the starting CAPE (specifically against its logarithm), separately for each holding period, using Robert Shiller's monthly data back to 1881. The percentage shown is the R²: the share of the variation in returns that starting valuation explains. The whole table is regenerated from the source data — no hand-typed numbers.
If CAPE can't time the market, what is it good for?
Setting expectations. A long-run return assumption feeds real decisions — how much to save, when you can retire, what withdrawal rate is safe — and starting valuation is one of the better guides to that assumption over a 10-to-15-year horizon. You act on it through your plan, not by jumping in and out of the market.
Is CAPE still comparable to its own history?
Imperfectly. Accounting standards, profit margins, interest rates, the shift from dividends to buybacks, and a heavier weighting of asset-light technology firms have all plausibly lifted the 'normal' range over time. That is a real argument for not reading today's level against the 1900s average too literally — but persistently higher valuations still mean fewer earnings bought per pound, which still points to thinner long-run returns unless earnings grow unusually fast.
Is the market cheap or expensive? Get the monthly read.
Once a month: the global CAPE and where it sits in history, the cheapest and most expensive markets, and what 10-year return today’s valuation implies. Free, no spam, unsubscribe anytime.
We use a double opt-in and never share your address. See our privacy notice.