Articles
Notes on valuation, wealth, and the gauges.
Mortgage rates are back to normal. Prices are not.
The 30-year mortgage is 6.66%, below its own 7.68% average. Real home prices are still 27.9% above a trailing 30-year mean.
September 1, 2026
2000 answers one question, not four
CAPE sits just under the dot-com peak. Buffett, Tobin's Q and household equity have all passed it. One column matches, three are worse.
August 27, 2026
What the following ten years looked like
Months in the Very High bucket were followed by a median real S&P return of β0.7% a year for a decade, and 53% of those decades lost purchasing power.
August 26, 2026
A good job market doesn't make stocks cheap
Unemployment is 4.1%. That tells you the expansion is intact. It tells you nothing about a 42Γ CAPE, a 214% Buffett Indicator, or a 93% composite.
August 24, 2026
Households already bought
Equities are 45.8% of household financial assets, a 99th-percentile reading. This measures what people own rather than how they feel, which is why it votes.
August 21, 2026
How the 93% is calculated
Ten percentiles, equally weighted, with the two low-bad gauges inverted and the three recession gauges left out. The whole method fits in a paragraph.
August 18, 2026
Lenders have stopped charging for default risk
High-yield spreads are 2.63 points against a 5-point average. On this gauge the small number is the dangerous one, and it's currently small.
August 17, 2026
The Sahm Rule is quiet. It isn't looking at stocks.
Sahm reads 0 against a 0.50 trigger. It's a real-time check on the labor market, and it has no opinion whatsoever about what you're paying for equities.
August 16, 2026
What a VIX of 14 actually tells you
The VIX closed at 14.51, well under its long-run average of 19. It measures the price of insurance, and cheap insurance has never meant the house won't burn.
August 14, 2026
The spending is coming from the people who own the shares
Net worth rises, saving falls, spending rises. The wealth doing that work sits in the same equities our gauges already call expensive.
August 10, 2026
The market costs more than two years of American output
US stocks are worth 214% of GDP, past both the 2000 and 2021 peaks. Globalization explains part of the climb, and we adjust for it. It doesn't explain this.
August 8, 2026
Extreme readings arrive in clusters
Since 1985, a CAPE above 30 has bunched into 2000, into 2021 and into now. Two of those clusters resolved badly. The third hasn't resolved.
August 7, 2026
CAPE crossed 30 in 1997 and peaked in 2000
Three years of being right about valuation and wrong about everything that paid. High readings have meant weak decade returns. They have never set a date.
August 2, 2026
Record highs are easier when there's more money
The S&P divided by M2 is 0.335 and well above trend. Expand the money supply and nominal records take care of themselves. This ratio asks whether stocks outran the cash.
July 31, 2026
Recession gauges get a seat, not a vote
Yield curves and the Sahm Rule answer a question about the economy. We're asking a question about price. Averaging the two would ruin both answers.
July 30, 2026
Buy the companies, or build new ones?
Tobin's Q is 1.82, so the market wants nearly twice what it would cost to rebuild these firms. Intangibles raised the baseline, and it still clears the higher one.
July 29, 2026
The curve went positive. Now comes the waiting.
10yβ2y is +0.39 and 10yβ3m is +0.83. Both positive, both graded Caution, because the damage has usually arrived after the re-steepening.
July 28, 2026
The rates argument has run out of rates
For a decade, low yields justified a high multiple. The 10-year is back near 4.6% and CAPE is still 42.17Γ. The interest rate model reads 4.15Ο.
July 27, 2026
Two gauges, two answers, and rates are the difference
CAPE is 42.17Γ and looks like 1999. The premium stocks pay over real bonds is 1.07 points, which we grade Fair Value. They're answering different questions.
July 24, 2026
The red lines are an index of the past
We mark 1929, 1937, 1962, 1973, 1987, 2000, 2008 and 2020 so you can see what each gauge was doing when things broke. They predict nothing.
July 21, 2026
Why we don't grade this market against 1871
CAPE's all-history average is about 17. Comparing today to that treats 1871 and 2026 as the same market. We use a trailing 30-year average instead.
July 17, 2026
Revenue is the harder number to flatter
The S&P 500 trades at 3.81 times sales against a post-2000 average of 1.8. Buybacks and margin cycles can flatter earnings. They can't do much to the top line.
July 14, 2026