Frequently asked questions

The headline YES/MAYBE is a majority vote of the eight valuation and sentiment gauges only — Shiller CAPE, Excess CAPE Yield, Buffett Indicator, Tobin’s Q, S&P 500 ÷ M2, price-to-sales, VIX, and the high-yield credit spread. Recession indicators (yield curves and the Sahm Rule) stay on the page for cycle context but do not vote on whether prices are rich.

Eleven gauges in total: six valuation (Shiller CAPE, Excess CAPE Yield, Buffett Indicator, Tobin's Q, S&P 500 ÷ M2, price-to-sales), two sentiment (VIX, high-yield credit spread), and three recession (10y–2y and 10y–3m yield curves, Sahm Rule). Only the eight valuation + sentiment gauges feed the headline verdict and composite probability.

Each valuation and sentiment indicator is compared against its own trend — a trailing 30-year average, an exponential regression, or a flat historical mean — and expressed as a z-score. Readings above +1σ count as overvalued; above +2σ as strongly overvalued. Gauges where a low value is the dangerous one (Excess CAPE Yield, VIX) are inverted. If a majority of those eight read overvalued, the headline is YES; otherwise MAYBE. Recession gauges are excluded from that vote.

Yes. The Excess CAPE Yield measures the premium stocks offer over the real 10-year Treasury yield — Robert Shiller’s own answer to the criticism that valuation ratios ignore interest rates. When rates are low, a high CAPE can still leave stocks attractive relative to bonds, and this gauge captures that.

For each valuation and sentiment indicator we take its direction-adjusted z-score, pass it through the normal distribution to get a percentile, and average those percentiles with equal weight. Recession gauges are left out of that average. 50% means the headline gauges sit near their historical norms; 100% would mean every one of them is at a historic extreme.

Every US market weekday, about an hour before the opening bell. The indicators are rebuilt automatically from public data, so the readings and the verdict stay current.

No. High readings have historically meant below-average returns over the following decade, not an imminent decline — markets can stay expensive for years. Nothing here is financial advice or a recommendation to buy or sell any security. Draw your own conclusions.

The Shiller CAPE ratio — also known as CAPE, PE 10, or the cyclically adjusted price-to-earnings ratio — compares the price of the S&P 500 to the average of its inflation-adjusted earnings over the previous ten years. Developed by Nobel laureate Robert Shiller, it smooths earnings across a full business cycle to strip out boom-and-bust swings that distort a simple P/E. Its long-run average is about 17.

The Excess CAPE Yield (ECY) is Robert Shiller’s answer to the criticism that the CAPE ratio ignores interest rates. It flips CAPE into an earnings yield (100 ÷ CAPE) and subtracts the real 10-year Treasury yield, measuring the extra return stocks offer over the bond alternative. It has averaged roughly 3–4 percentage points since the 1960s.

The Buffett Indicator — also called the market-cap-to-GDP ratio — is the total value of the US stock market divided by US gross domestic product (GDP). Warren Buffett once called it “probably the best single measure of where valuations stand at any given moment.” The long-run average is roughly 100%.

Tobin's Q — named after Nobel economist James Tobin — compares the market value of US corporate equities to the replacement cost (net worth) of the underlying companies. A value above 1 means the stock market prices companies above what it would cost to rebuild them from scratch; the long-run mean is around 0.75.

The S&P 500 ÷ M2 ratio divides the S&P 500 index by the M2 money supply to adjust stock valuations for monetary inflation. When central banks expand the money supply, nominal record highs can be misleading — this ratio shows whether US stocks are actually rising faster than the amount of money circulating in the economy.

The S&P 500 price-to-sales ratio compares the market value of the index to the trailing twelve-month sales of its constituents. Since 2000 it has averaged about 1.8; readings near 3× sit far above that modern-era mean and have historically looked stretched.

The VIX — the CBOE Volatility Index, often called Wall Street’s “fear index” — measures the market’s expectation of 30-day volatility, implied from S&P 500 options prices. A high VIX means fear; a very low VIX means complacency. The VIX averages around 19–20.

The high-yield credit spread — the ICE BofA US High Yield Option-Adjusted Spread — is the extra yield that below-investment-grade (“junk”) bonds pay over US Treasuries. It measures how much compensation lenders demand for taking credit risk. The spread averages around 5 percentage points.

The 10y–2y yield curve measures the gap between long-term and short-term US government borrowing costs. Normally long-term yields are higher, giving a positive spread; when the curve inverts and the spread goes negative, a recession has historically followed — usually 6 to 24 months later.

The 10y–3m yield curve is the classic yield-curve recession gauge — the tenor pair used by the New York Fed’s recession-probability model. A negative spread (inversion) has preceded every US recession since the late 1960s, typically by about 6 to 18 months.

The Sahm Rule equals the three-month moving average of the US unemployment rate minus the lowest three-month average over the previous 12 months. A reading of 0.50 percentage points or higher is the classic trigger that a US recession has historically already begun.

A z-score measures how far a current reading sits from its historical trend, expressed in number of standard deviations. The formula is (current value − trend) ÷ standard deviation. A z-score of 0 means the reading sits right on its trend; +1 means one standard deviation above; +2 means two above. On this site, readings above +1 count as overvalued and above +2 as strongly overvalued. For stress gauges where a low value is the warning (the VIX, yield curves, and Excess CAPE Yield), the sign is flipped so that higher always means more overvalued.

Standard deviation measures how spread out a set of historical readings is around its average. If most values cluster tightly around the mean, the standard deviation is small; if they vary widely, it is large. We use it to set the ±1σ and ±2σ bands on each indicator chart: readings within one standard deviation of the trend are near normal, between one and two are elevated, and beyond two are historically extreme. Roughly two-thirds of a bell-shaped distribution falls within ±1 standard deviation, and about 95% within ±2.

Variance is the square of the standard deviation. It measures the average squared distance of historical readings from their mean. We compute standard deviation as the square root of variance, which brings the units back to the same scale as the indicator itself (percentage points, ratios, and so on) so the bands on the charts are easy to read.

A percentile tells you what fraction of historical readings a current value exceeds. A reading at the 90th percentile means it is higher than 90% of all past values and lower than only 10%. We convert each valuation and sentiment indicator’s direction-adjusted z-score into a percentile using the normal distribution, then average those percentiles to build the composite overvaluation probability. A composite of 75% means the average headline indicator sits around the 75th percentile of its own history.

The trend baseline is the expected level of an indicator given its historical pattern — the line running through the middle of each chart. Depending on the indicator, that baseline is a flat historical mean, a trailing 30-year average, or an exponential regression curve. The gap between the current reading and this baseline, divided by the standard deviation of past deviations from the baseline, becomes the z-score.

A trailing 30-year average is the mean of an indicator’s readings over the most recent 30 years, recalculated at each point in time using only data available up to that date. We use it as the baseline for the Shiller CAPE and Tobin’s Q so the verdict adapts to structural regime shifts — such as lower interest rates or more intangible assets — instead of assuming a quick return to century-old norms. The standard deviation is also computed from residuals within that same rolling window.

A flat historical mean is the simple average of all past readings for an indicator, used as a fixed baseline. We use it for gauges whose long-run level has not shifted structurally over time: the Excess CAPE Yield, S&P 500 price-to-sales, the VIX, the high-yield credit spread, both yield curves, and the Sahm Rule. The trend line on these charts is a horizontal line at the all-time average.

An exponential regression fits a curve that grows at a constant percentage rate over time, using log-linear least squares on the full history. We use it for ratios that have grown structurally for decades — the Buffett Indicator and the S&P 500 to M2 ratio — so the baseline rises with the economy rather than staying flat. The standard deviation of residuals around this curve sets the ±1σ and ±2σ bands.

The normal distribution — the bell curve — describes how many natural phenomena cluster around a central value with symmetric tails. Its cumulative distribution function (normal CDF) converts a z-score into a probability between 0 and 1: a z-score of 0 maps to 50%, +1 to about 84%, and +2 to about 98%. We use the normal CDF to turn each indicator’s standardized reading into a percentile before averaging them into the composite overvaluation probability.

Standard deviation bands are the shaded zones on each indicator chart showing ±1σ and ±2σ around the trend line. The inner band (between −1σ and +1σ) is the normal range; the outer band (between ±1σ and ±2σ) is elevated; readings beyond ±2σ are historically extreme. These bands are how we translate a raw indicator value into an overvalued, fair value, or undervalued verdict.

Inversion (or direction adjustment) flips the sign of an indicator’s z-score when a low value is the dangerous signal rather than a high one. The Excess CAPE Yield, the VIX, and both yield curves are inverted: a very low VIX or a deeply inverted yield curve registers as overvalued even though the raw number is small. This ensures every indicator contributes to the composite on the same scale, where higher always means more overvalued.

An equal-weighted average gives each of the eight valuation and sentiment indicators the same influence on the composite overvaluation probability. Recession gauges are excluded. If seven headline indicators read elevated and one reads cheap, the composite reflects the average of all eight percentiles rather than a simple majority vote.

A time series is a sequence of historical readings plotted over time — the line charts on each indicator page. Every point shows where that gauge stood on a given date relative to its trend baseline and standard deviation bands. The composite overvaluation probability chart is itself a time series, reconstructing the equal-weighted average percentile for every date on which at least five indicators have data.

The composite overvaluation probability is a single number from 0% to 100% that blends the eight valuation and sentiment gauges into one reading of how stretched prices are relative to their own history. Recession gauges are excluded. 50% means the headline gauges sit near historical norms; 100% would mean every one of them is at a historic extreme.

Valuation sub-scores aggregate the six valuation indicators; sentiment sub-scores aggregate the VIX and high-yield credit spread. Those two groups are averaged into the headline composite. Recession sub-scores (yield curves and Sahm Rule) are shown for cycle context but excluded from the composite. Each sub-score is the equal-weighted average percentile of its group.

Crash reference lines are vertical markers on each indicator’s historical chart showing major US market downturns — such as 1929, 1987, 2000, 2008, and 2020 — so you can see where the indicator stood before each crash. They are reference points, not predictions: the goal is to compare today’s reading to how the gauge looked at prior turning points.

Stock market valuation measures whether prices are high or low relative to underlying fundamentals — earnings, economic output, replacement cost, interest rates, and the money supply. Valuation gauges describe long-run conditions; they say a lot about the next decade and very little about the next few months.

Earnings yield is the inverse of a price-to-earnings ratio, expressed as a percentage: 100 ÷ P/E. A CAPE of 25 implies an earnings yield of 4%. It represents the annual return shareholders would earn if all profits were paid out and the company never grew.

A real yield is a bond yield adjusted for inflation — the nominal Treasury yield minus expected or trailing inflation. We use the real 10-year Treasury yield in the Excess CAPE Yield to compare what stocks earn against what bonds earn after inflation.

Market cap to GDP is the ratio of total US stock market value to one year of US gross domestic product. It is the same measure as the Buffett Indicator: when market value races far ahead of the underlying economy, it has historically signaled stretched valuations.

M2 is a broad measure of the US money supply: cash, checking and savings deposits, and other near-money. The Federal Reserve publishes it in the H.6 release. Dividing the S&P 500 by M2 adjusts stock prices for monetary inflation.

Treasury yields are the interest rates the US government pays to borrow money. We use the 10-year constant-maturity Treasury yield to compute the real bond yield in the Excess CAPE Yield, and the gaps between the 10-year and the 2-year / 3-month yields form our yield-curve spreads.

Yield curve inversion occurs when short-term Treasury yields rise above long-term yields — for example when the 10y–2y or 10y–3m spread turns negative. It has preceded every US recession since the 1970s. Recessions often begin only after the curve un-inverts (turns positive again), so the months right after a long inversion can still be high-risk even if the live spread looks normal. Our pages report a z-score of today’s level and explain that cycle separately.

A credit spread is the extra yield investors demand to hold risky debt over safe government bonds. The high-yield credit spread measures this for junk bonds. Unusually tight spreads signal complacency; widening spreads signal building financial stress.

Forward decade returns are the actual inflation-adjusted returns the S&P 500 delivered over the ten years after a given valuation reading. When the market is overvalued — you pay a high price for future earnings — those decade-ahead returns have historically been below average, and negative decades become more common. Cheap starting valuations have meant the opposite. That is a statement about the long run, not a crash forecast: markets can stay expensive for years before the lower returns show up.

Overvalued means an indicator’s current reading sits more than one standard deviation above its historical trend; strongly overvalued means more than two. Undervalued and strongly undervalued are the mirror cases below trend. Within ±1 standard deviation of normal, valuation gauges read Fair Value; VIX, credit spread, and Sahm Rule read Neutral; and the yield curves read Caution — a near-average spread is not treated as an all-clear, given how often recessions arrive after the curve un-inverts.