Tracker · What stocks cost
CAPE ratio today
The CAPE ratio is 40.10 as of June 2026, which means the S&P Composite costs 40.1 times its average real earnings over the past 10 years. That is the 99th percentile of a record going back to December 1880, against a long-run mean of 17.7 and a 30-year mean of 28.7. The implied earnings yield is 2.49%.
1747 monthly readings
Where 40.10 sits on 146 years
Shiller's own column
He carries his figure to August 2026 at 41.12 by extrapolating earnings past the last reported quarter. We stop where the reported earnings stop, so our headline is June 2026. Both are on this page because the difference is a method choice rather than a disagreement.
The long record
CAPE ratio chart, 1881 to today
Every monthly reading against the 17.7 long-run mean. The ratio has spent most of its life between 10 and 20, and the three excursions above 30 are 1929, 1999 and the period since 2018.
Real price divided by 10-year average real earnings
Shiller CAPE, 1880 to today
June 2026
40.10The dashed rule is the 17.7 full-record mean. The 30-year mean is 28.7, and which of the two you treat as normal decides most of the argument about whether today is expensive.
Source: Robert Shiller's online data, price, earnings and CPI columns. Retrieved 2026-09-09
Download CSVHow to read it
What counts as a high CAPE ratio
A level means nothing without the distribution behind it. Of the 1747 months on record, 23 closed at or above today's 40.10, and all but a handful of those came after 1997.
Count of monthly readings in each band
CAPE distribution, 1747 monthly readings
What we found in the data
What the next ten years did, from every starting level
We sorted all 1629 months that have a finished ten-year window by the CAPE they started at, then measured the real total return of the following decade. Starting under 10 the median was +10.99% a year and not one of those 226 decades lost to inflation. Starting over 30 the median was -1.27% a year and only 39% of them stayed ahead of it.
The middle of the table is not a ladder. The 25 to 30 band beat the 20 to 25 band, because it is 106 months drawn mostly from the mid 1990s and the decade that followed them was a good one. Anybody quoting a smooth relationship here is quoting something this sample contradicts. Returns are real and include reinvested dividends, with no tax, cost or spread applied.
Decades starting under 10 that beat inflation
226 of 226Decades starting over 30 that beat inflation
22 of 56Median real total return a year over the following ten years, by starting CAPE
| Starting CAPE | Months | Median | Worst | Best | Beat inflation |
|---|---|---|---|---|---|
| Under 10 | 226 | +10.99% | +1.8% | +20.0% | 100% |
| 10 to 15 | 465 | +7.83% | -4.2% | +17.6% | 90% |
| 15 to 20 | 515 | +6.25% | -4.6% | +16.1% | 91% |
| 20 to 25 | 261 | +4.92% | -4.0% | +14.2% | 76% |
| 25 to 30 | 106 | +5.50% | -1.2% | +12.0% | 95% |
| Over 30 We are here | 56 | -1.27% | -5.9% | +4.5% | 39% |
Annualised real total return of the S&P Composite over the ten years after each month. Past performance does not predict future results
Every reading above 25
What happened after each CAPE peak
Four episodes in 145 years. Three of them have a finished decade behind them and all three lost money in real terms, which is the strongest thing this page can say. The fourth is still running.
-
The 1929 peak
32.39September 1929
The highest reading in the record until 1998. The index did not regain its 1929 real level until 1958.
Next ten years, real total return -1.42% a year
-
The 1966 peak
24.01January 1966
No crash followed, which is the part usually left out. Inflation did the work instead, over sixteen years.
Next ten years, real total return -1.77% a year
-
The dot-com peak
44.00December 1999
The highest reading ever recorded, and the only one above 40 until the 2020s.
Next ten years, real total return -3.17% a year
-
The 2021 peak
38.81December 2021
Second only to 1999, reached on near-zero real interest rates rather than on an earnings story.
Next ten years, real total return still running
Long history
Average CAPE ratio by decade
The 1990s broke the range the previous eleven decades had kept to, and nothing since has gone back to it. Whether that is a permanent repricing or a long excursion is the open question on this page, and the data cannot close it.
The question this site exists for
What gold did after each CAPE reading
Gold has only been freely priced since 1968, so this covers 579 months against the 1629 the equity table uses, and the high-CAPE readings inside that window come from a small number of episodes. Read it as a description of those episodes rather than as a rule.
Median real gold return a year over the following ten years, by starting CAPE
What this cannot tell you
A high CAPE and a strong subsequent decade for gold have coincided, and the sample is small enough that one long inflation is doing most of the work. Gold was fixed by law for the first two thirds of this record, so the overlap starts in 1968 and contains a single episode of equities being expensive while inflation ran. That is not enough to support a forecast, and a page that dressed it up as one would be doing the thing this site was built to correct.
If you are moving retirement money because of this number
What a high CAPE ratio means for a gold IRA
A CAPE of 40.10 is a statement about expected equity returns over a decade, and it says nothing at all about what gold will do. The two are separate arguments and a sales call will run them together. What the ratio genuinely supports is a lower expected return from US equities than the last fifteen years delivered, which is a reason to look at what you hold rather than a reason to move it somewhere specific.
If you do move metal into an IRA, the dealer's premium over spot is paid on the first day, whatever the ratio does afterwards. On a $50,000 purchase the difference between a 3% and an 8% premium is $2,500, which never reaches the account.
Related
More market valuation tools and trackers
The calculator that runs this record against an amount, the other valuation measure, and the comparison a high reading gets used to argue for.
Common questions
CAPE ratio questions
- Is a CAPE of 40 a sell signal?
- No. It is a percentile, not a trigger. The ratio has closed at or above today's 40.10 in 23 of the 1747 months on record, and on the last occasion it stayed above 40 for more than two years before it fell. CAPE has explanatory power over ten years and close to none over one, so treating it as timing is using it for the one thing the evidence says it cannot do.
- What does CAPE actually predict?
- The next decade's real total return, loosely. Starting under 10 the median was +11.0% a year and every one of those 226 decades beat inflation. Starting over 30 the median was -1.3% a year and only 39% of them did. The bands in between do not form a clean ladder, which is worth knowing before anyone quotes one at you.
- Why is the CAPE ratio higher than it used to be?
- Three arguments are made and none of them is settled. Accounting rules changed in ways that depress reported earnings against the pre-1990 basis. Index composition shifted toward companies that need less capital to grow. And real interest rates fell for forty years, which raises what any stream of future earnings is worth. All three are real effects and none of them is measurable precisely enough to say how much of the gap they explain, which is why this page publishes the 30-year mean of 28.7 beside the full-record mean of 17.7 rather than picking one.
- Does a high CAPE mean gold will do well?
- The record cannot answer that. Gold has only been freely priced since 1968, so it covers 579 of the 1629 months where a finished decade exists, and almost all of the high-CAPE readings inside that window come from two episodes. A relationship fitted on two episodes is a description of those episodes. This page shows what gold did and declines to call it a pattern.
- Is this the same as the P/E ratio?
- No. A plain P/E divides price by one year of earnings, and one year of earnings collapses in a recession, so a plain P/E reports a market as most expensive at the bottom. In 2009 trailing earnings briefly took the S&P's plain P/E above 120 while CAPE read about 13. CAPE uses 10 years of inflation-adjusted earnings for exactly that reason.
- Where does this number come from?
- Robert Shiller publishes the price, earnings and CPI columns behind it, and we rebuild the ratio from those three rather than reading his CAPE column. His column is committed beside ours as the check. Across the 1629 months where both series price the index the same way the two agree to a median of 0.22%, with a worst case of 0.89%.
How this page is built
CAPE is the real price of the S&P Composite divided by the average of its real earnings over the preceding 120 months. Price, earnings and the CPI deflator all come from Robert Shiller's published dataset, so the deflator that adjusts the price is the same one that adjusts the earnings and the ratio is free of any base year. The series starts in December 1880, ten years after the earnings record does, because a ten-year average cannot exist before ten years of earnings do.
We compute the ratio rather than reading Shiller's own CAPE column, and his column is committed beside ours as the check. Across the 1629 months where both series price the index the same way the two agree to a median of 0.22% and a worst case of 0.89%. From October 2016 our price series switches to month-end closes while his stays on monthly averages, so the two diverge on months where the index moved hard inside the month. February 2020 is 10.2% apart for that reason and both figures are right.
Forward returns are real and include reinvested dividends, with no tax, trading cost or spread applied, so a real outcome is worse than the figures here by whatever those cost. The headline reading stops at June 2026 because that is where reported earnings stop. Every chart is server rendered and readable with JavaScript turned off.
Sources
- Robert Shiller, online data for Irrational Exuberance
- FRED SP500, for month-end closes since 2016
- FRED CPIAUCSL, for the gold comparison
- LBMA precious metal prices
Full dataset available as CSV, including our figure and Shiller's side by side so the reconciliation can be checked independently.
Gold IRA Digest is an independent publisher. This page is information, not investment advice, and nothing on it is a recommendation to buy or sell any asset. Past performance does not predict future results.