# What the Shiller CAPE tells you about the next ten years

_At 42 times earnings, the market is priced for a decade in which very little is allowed to go wrong_

Jon Adair · 8 July 2026 · 3 min read

![CAPE: Cyclically Adjusted Price to Earnings Ratio](https://cdn.sanity.io/images/v3acfbvo/production/2aac33c5e18bb84ad55e68c687aaf7b0dca5dd79-1536x1024.png?w=1600&fit=max&auto=format)

---

The S&P 500 has been more expensive than it is now on only one occasion in almost a century and a half. On a cyclically adjusted basis, it trades at around 42 times earnings. The single clear precedent for that figure is 1999 to 2000; the peak before it was 1929. Both are on the chart below, along with the decade each led into. _(Neil in the margin: The stock-market peak before the Wall Street Crash. The Dow lost roughly 90% of its value from 1929 to its 1932 bottom, and did not reclaim the peak in nominal terms until 1954.)_

![S&P 500 Cyclically Adjusted Price Earnings Ratio](https://r4at4qm6kmohrtvq.public.blob.vercel-storage.com/charts/chart-cape-shiller-line-e520b2a02092-light.png)

## What CAPE measures

_[Embedded media](https://en.wikipedia.org/wiki/Robert_J._Shiller)_

A conventional price-to-earnings ratio divides today’s price by a single year’s earnings, which makes it almost useless when things change. Earnings collapse in a recession, and the ratio looks frighteningly high; earnings peak in a boom, and it looks reassuringly low, both at exactly the wrong moment. Robert Shiller's adjustment is simple. Divide the price by the average of the past 10 years’ earnings, adjusted for inflation. Ten years covers a broader economic cycle, so no single year can flatter or distort the reading.

The result is the [cyclically adjusted price/earnings ratio](https://en.wikipedia.org/wiki/Cyclically_adjusted_price-to-earnings_ratio), also written P/E10 or the Shiller CAPE. Its long-run average sits in the high teens. Today's reading of roughly 42 is more than double that. _(Neil in the margin: Since 1881 the CAPE has averaged around 17. Note the average itself has drifted up over the decades, which is partly what the "unfair to the present" caveat later is about.)_

## Cheap starts pay. Expensive starts do not.

Read the crosses on the chart from the bottom up. Out of the troughs, the following decade was extraordinary. From the single-digit CAPE of 1921, the market returned 269% in nominal terms and 345% in real terms over the next ten years. From 1932, 216% and 163%. From the 1982 low, 475% and 299%. From 2009, 338% and 267%. _(Neil in the margin: Real means after stripping out inflation. It exceeds the nominal figure here because the early 1930s saw deflation, so a pound bought more each year — the mirror image of the 1970s effect he flags later.)_

From the peaks, it ran the other way. The decade after 1929 returned −29% in nominal terms and −13% in real terms. The decade after 2000, −9% nominal and −29% real. Same market, same measure. The only thing that changed was the price paid at the start.

> The same ten years can show a gain on the screen and a loss in your pocket.

One cross is worth stopping on. From 1966, with a CAPE near 24, the market returned 53% in nominal terms over the following decade. In real terms, it lost 12%. The inflation of the 1970s did that: the quoted number kept climbing while purchasing power fell behind it. The starting valuation is what most separates the two figures. _(Neil in the margin: UK inflation peaked above 20% in 1975; US CPI ran into double digits twice that decade. Nominal share prices can rise while your purchasing power quietly erodes — the whole reason CAPE is inflation-adjusted in the first place.)_

## The same signal, every month since 1881

The previous time series chart picks out eight moments. The chart below plots every single one. Each dot represents a single month, with its valuation on the horizontal axis and the annualised real 10 year return that actually followed on the vertical axis.

![Buy expensive, and the next ten years have skewed thin.](https://r4at4qm6kmohrtvq.public.blob.vercel-storage.com/charts/chart-cape-fwd-return-scatter-a20e3f3b992e-light.png)

The cloud slopes down and to the right. Buy at a low multiple, and the outcomes fan out high; buy at a high one, and they compress toward zero, and the strong decades simply stop appearing. At the far right, where the market sits today, the historical record holds no example of a good ten years to follow.

## Four things it cannot tell you

Valuation is a powerful lens and a poor clock. Four limits are worth holding in view.

1. It says nothing about **timing**. CAPE was already high in 1997, and the market climbed for three more years before it broke. Expensive can become more expensive and stay there. _(Neil in the margin: The classic timing problem: valuation told you the 1990s market was dear well before it topped in 2000. Being early is, in practice, indistinguishable from being wrong.)_

2. The comparison may be unfair to the present. Accounting rules have changed, companies now return cash through buybacks as much as through dividends, and the index is dominated by asset-light technology businesses earning far higher returns on capital than the railroads and steelmakers in the old data. A market like that can reasonably carry some premium to its own history. This time really might be different. _(Neil in the margin: Companies buy back their own shares instead of paying dividends. Because buybacks lift per-share earnings and were rare in the older data, some argue today's CAPE isn't strictly comparable with a century ago.)_

3. **Interest rates** change the sum. When the safe return is low, a higher multiple on equities is easier to justify. Part of the re-rating since 2009 is rational rather than manic. _(Neil in the margin: A re-rating means investors agreeing to pay a higher multiple for the same earnings. With bond yields floored near zero after 2009, equities looked relatively more attractive — so part of the rise is arithmetic, not mania.)_

4. It is a **probability**, not a promise. The scatter is a cloud, not a rule. High starts have led to weak decades far more often than strong ones. The word “often” is the important one.

## How Neil reads it

Neil does not use CAPE to call a top. Nobody can, and the history of those who tried is not flattering. He uses it to answer a single question: what am I starting from? A reading of 42 does not say sell, nor does it say when. It says the market is priced for a decade in which very little is allowed to go wrong, and that the margin for error is thin.

That is a reason to know exactly what you own and why you own it, not a reason to watch the number each morning. The price paid is the one variable an investor controls completely, and across ten years, it is the one that counts the most.

_Past returns are worked examples. They illustrate a relationship in the historical record; they are not a forecast, and nothing here is a recommendation to buy or sell any security._