What is the overvaluation probability?

It is a single number from 0% to 100% that blends the ten valuation and sentiment gauges — Shiller CAPE, Excess CAPE Yield, Buffett Indicator, Tobin’s Q, S&P 500 ÷ M2, price-to-sales, the interest rate model, VIX, the high-yield credit spread, and household equity allocation — into one reading of how stretched prices are versus their own history. Recession gauges (yield curves, Sahm Rule) are shown for cycle context but excluded from this number. 50% means the headline gauges sit near historical norms; 100% would mean every one of them is at a historic extreme.

How is the overvaluation probability calculated?

For each valuation and sentiment indicator we take its current reading, express it as a z-score (standard deviations from its own trend), flip the sign where a low value is the dangerous one (Excess CAPE Yield, VIX), and pass that through the normal distribution to get a percentile. The headline figure is the equal-weighted average of those ten percentiles. Recession indicators do not enter the average.

Does the composite account for interest rates?

Yes — through the Excess CAPE Yield (earnings yield minus the real 10-year Treasury yield) and the interest rate model (S&P 500 position conditioned on the 10-year yield). Both are valuation gauges in this composite.

Does a high probability mean a crash is coming?

No. A high reading means valuations are historically stretched across many independent measures, which has preceded lower returns over the following decade — but markets can stay expensive for years, and the gauge says nothing about timing. The "What happened next" section on this page shows the actual historical outcomes. It is a descriptive summary of valuation conditions, not financial advice or a market-timing signal.

How often is it updated?

Every US market weekday, about an hour before the opening bell, alongside the underlying indicators. The composite is recomputed from the latest available reading of each valuation and sentiment input.

← All indicators

Overvaluation probability

One composite gauge blending valuation and sentiment indicators into a single percentile of overvaluation.

92%

Very High

As of September 2, 2026

As of September 2, 2026, the composite overvaluation probability for the US stock market is 92% — a Very High reading. 0% means indicators are historically cheap, 50% means fair, and 100% would mean every gauge is at a historic extreme.

No single formula decides this: valuation gauges alone read 96%, sentiment gauges 84%, recession gauges (excluded from the composite) 57%, and the headline indicators range from 66% to 100%. The wider that range, the more the gauges disagree.

Share on

Historical crashes marked on this chart (6)

The dashed red lines flag major US market crashes. Note how the composite probability behaved in the run-up to each.

Kennedy Slide (1962): A rapid spring 1962 sell-off (the 'Flash Crash of 1962') wiped roughly 27% off the S&P 500 amid fears over the economy and a clash with steelmakers. Markets stabilized and recovered within about a year.
1973–74 oil crisis bear market (1973): The OPEC oil embargo, Watergate, and stagflation drove a roughly 48% decline over nearly two years — the worst bear market since the 1930s. Stocks took years to recover in real terms.
Black Monday (1987): On October 19, 1987 the Dow fell 22.6% in a single day — the worst one-day drop in history — driven partly by program trading. The economy avoided recession and markets recovered within two years.
Dot-com crash (2000): The tech bubble burst in March 2000; the Nasdaq lost about 78% by 2002 as internet valuations collapsed. A mild recession followed and tech stocks took roughly 15 years to reclaim their peak.
Global Financial Crisis (2008): The subprime mortgage meltdown and Lehman Brothers' collapse cut the S&P 500 about 57% into March 2009. It triggered the Great Recession, sweeping bailouts, and a decade of near-zero interest rates.
COVID crash (2020): Pandemic lockdowns caused the fastest bear market ever in February–March 2020, a roughly 34% plunge in weeks. Unprecedented Fed and fiscal stimulus sparked a rapid recovery to new highs.

What’s driving it today

Each gauge’s own probability of signalling overvaluation (percentile of its history). The composite above is the equal-weighted average of the valuation and sentiment rows only.

Bar color matches the verdict badge. Gray tick = fair midpoint (50%); red tick ≈ Overvalued threshold (+1σ, ~84%). A Fair Value gauge can still show a high % — the percentile climbs before the verdict flips to Overvalued at ~84%.

FairOvervalued

Recession context (not in the composite)

Shown for cycle risk. These do not vote on the overvaluation probability above.

What happened next

When the stock market is overvalued, you are paying a high price for future earnings. That does not schedule a crash next month — but it has historically meant lower returns over the following decade. Buying expensive leaves less upside; buying cheap leaves more. This section is the evidence for that claim using our composite.

For every month with a full decade of later data, we take the composite level at the time and measure the annualized real (inflation-adjusted) S&P 500 return over the next 10 years. That is a forward decade return: what an investor who bought the index then actually earned after inflation, year by year, until ten years later.

Today’s reading is Very High — historically, months like this were followed by a median -0.7%/yr real return over the next decade, and 53% of those decades finished negative. When the composite was Very High — readings clustered near past market peaks — the following decade usually delivered the weakest real returns in the sample, and negative decades were far more common.

Read across the table: as the composite moves from Low → Very High, median decade-ahead returns fall and the share of losing decades rises. That pattern is why a high overvaluation probability matters for long-horizon investors even when the near-term path is unknowable.

Reading at the time	Median 10y real return	Range	Negative decades	Months
Low	+7.1%/yr	-3.9% to +13.5%	5%	238
Elevated	+6.6%/yr	-8.2% to +11.8%	23%	162
High	+3.6%/yr	-7.7% to +10%	18%	162
Very Hightoday	-0.7%/yr	-7.5% to +9.1%	53%	197
All periods	+4.4%/yr	-8.2% to +13.5%	24%	759

Median 10y real return — typical annualized inflation-adjusted S&P 500 return over the decade after a reading in that bucket. Lower medians at High / Very High are the core finding.
Negative decades — share of starting months where an investor lost purchasing power over the full ten years. That share climbs as valuations get richer.
Overlapping monthly windows are not independent observations, and only dates with a full decade of later data count — the most recent 10 years are excluded. Markets can stay expensive for years (the late 1990s) before those lower returns show up. This is historical description, not a forecast or advice to buy or sell.

How it’s calculated

Collect the latest reading for each valuation and sentiment indicator from public data (FRED and Robert Shiller’s dataset via multpl.com). Recession gauges are tracked separately and not averaged in.
Fit each indicator against its own trend and express the current reading as a z-score: the number of standard deviations above or below that trend. Long-history gauges with structural regime shifts (the Shiller CAPE, Tobin’s Q, and household equity allocation) use a trailing 30-year average, growing ratios (the Buffett Indicator and S&P 500 ÷ M2) use an exponential regression, and the rest use their flat historical mean.
Flip the sign for gauges where a low value is the dangerous one (the Excess CAPE Yield and the VIX), so that a higher adjusted z-score always means “more overvalued” — see inversion in the FAQ.
Convert each adjusted z-score into a probability with the normal CDF — the percentile the reading sits at within its own history.
Average the ten valuation and sentiment probabilities with equal weight and round to a whole percent. Recession gauges are left out. The historical line reconstructs this same calculation for every date on which at least five headline indicators have data.

Reading the levels

Low (below 40%): Indicators sit near or below their historical norms — valuations are not stretched.
Elevated (40–59%): Some indicators are moving into expensive territory; conditions are richer than average.
High (60–74%): A majority of indicators are meaningfully above their trends — valuations are historically rich.
Very High (75% and above): Indicators are clustered near historic extremes, comparable to prior market peaks.

Limitations

The indicators cover different spans — the Shiller CAPE reaches back to the 19th century while the VIX and credit spreads begin in the 1990s. The historical line only starts once at least five headline (valuation or sentiment) indicators have data, so earlier decades are deliberately excluded rather than shown as a thin, misleading composite.
The Shiller CAPE, Tobin’s Q, and household equity allocation use trailing 30-year baselines, so their z-scores rely only on data that existed at each point in time. The exponential-trend and flat-mean gauges are still fit over their full history, so those parts of the historical line remain a descriptive, hindsight view rather than a fully point-in-time signal.
Equal weighting is a deliberately simple choice. It treats a sentiment gauge like the VIX as just as important as a valuation gauge like the Buffett Indicator, which reasonable people can disagree with — the valuation / sentiment sub-scores above show how much the groups differ. Recession gauges are shown separately and do not enter the composite.
The “What happened next” table is based on overlapping monthly windows covering only a handful of independent decades, so its return figures are historical description, not a forecast.
This is a summary of long-run valuation conditions, not a market-timing tool. High readings have historically meant lower returns over the following decade — not an imminent decline. Nothing here is financial advice.

Frequently asked questions

What is the overvaluation probability?: It is a single number from 0% to 100% that blends the ten valuation and sentiment gauges — Shiller CAPE, Excess CAPE Yield, Buffett Indicator, Tobin’s Q, S&P 500 ÷ M2, price-to-sales, the interest rate model, VIX, the high-yield credit spread, and household equity allocation — into one reading of how stretched prices are versus their own history. Recession gauges (yield curves, Sahm Rule) are shown for cycle context but excluded from this number. 50% means the headline gauges sit near historical norms; 100% would mean every one of them is at a historic extreme.
How is the overvaluation probability calculated?: For each valuation and sentiment indicator we take its current reading, express it as a z-score (standard deviations from its own trend), flip the sign where a low value is the dangerous one (Excess CAPE Yield, VIX), and pass that through the normal distribution to get a percentile. The headline figure is the equal-weighted average of those ten percentiles. Recession indicators do not enter the average.
Does the composite account for interest rates?: Yes — through the Excess CAPE Yield (earnings yield minus the real 10-year Treasury yield) and the interest rate model (S&P 500 position conditioned on the 10-year yield). Both are valuation gauges in this composite.
Does a high probability mean a crash is coming?: No. A high reading means valuations are historically stretched across many independent measures, which has preceded lower returns over the following decade — but markets can stay expensive for years, and the gauge says nothing about timing. The "What happened next" section on this page shows the actual historical outcomes. It is a descriptive summary of valuation conditions, not financial advice or a market-timing signal.
How often is it updated?: Every US market weekday, about an hour before the opening bell, alongside the underlying indicators. The composite is recomputed from the latest available reading of each valuation and sentiment input.