The Market Valuation Guide: Is the Market in a Bubble?

Guide. Every number on this page is computed point-in-time from primary sources (Fed Z.1 & H.8, BEA, FINRA, Shiller's dataset) — values count only after their publication date, so historical scores contain no hindsight. Maintained and reviewed by Yuriy Matso; see our methodology and how we use AI.

Every cycle someone screams "bubble" and someone answers "this time it’s justified" — and both point at charts. This guide is a framework for settling it honestly: which valuation gauges have actually mattered, how to score them against their own published history with no hindsight, what the identical framework read at the 2000, 2007 and 2021–22 peaks, and — the part valuation can never give you — why a maxed-out score still says nothing about timing. Today’s score is live at the top.

What valuation can — and can't — tell you

Valuation is the most abused word in market commentary because it gets asked to do a job it has never done: call tops. It can't. The record is unambiguous — the gauges on this page were screaming in 1997, three years and a Nasdaq triple before the 2000 peak, and our own framework read maxed-out scores through most of 1998–2000. Anyone who sold on valuation alone left the best part of the bubble on the table.

What valuation has done, reliably, is set the odds for the years that follow. Start-point valuation is among the best predictors of 10-year forward returns anywhere in finance, and the outcome table below shows the same result on our own score history: expensive starting points produced weak multi-year returns and much deeper drawdowns, cheap ones produced the generational entries. Read this entire guide with that frame — conditions, not timing — and every chart on it becomes useful instead of terrifying.

The anchors: price vs the economy, vs earnings, vs bonds

Every serious valuation argument reduces to a choice of denominator. Price relative to the economy: the Buffett Indicator (public equity market value ÷ GDP) reads 250% today against a long-run median near 84% — far past its 2000-peak reading. Price relative to earnings: the cap-to-profits multiple reads 20.2× versus 30× at the 2000 record — much less alarming, because corporate profits have boomed. And price relative to ten years of real earnings: the Shiller CAPE at 41.4×, within a few points of its all-time December-1999 record of 44.2.

The gap between the GDP anchor and the earnings anchors is the modern valuation debate: if today's record profit margins persist, the earnings-based reads are the honest ones; if margins mean-revert, the earnings anchors snap up to meet the GDP anchor. There is no data that settles it in advance — which is why the framework scores all of them rather than letting you pick your favorite.

The fourth anchor answers the objection you've already thought of: don't low rates justify this? Shiller's Excess CAPE Yield — the CAPE earnings yield minus the real 10-year Treasury yield, published in the same dataset since 1881 — measures what stocks offer over bonds. It went negative at the 2000 peak (stocks were priced to return less than TIPS), stayed comfortably positive through the zero-rate 2021 top, and reads 1.39% today (77th bubble percentile). When the rates rebuttal is testable, it stops being a rebuttal and becomes a gauge.

Allocation & cash: how invested is everyone already?

Valuation ratios describe prices; allocation gauges describe people. The single best long-horizon predictor in the basket is household equity allocation — the share of household financial assets in stocks, 45.8% today and at records. Its logic is supply and demand at civilizational scale: when everyone is already fully invested, the marginal buyer is gone.

Its mirror is cash. The sideline cash ratio — money-fund assets plus bank deposits relative to the market value of all US equities — sits near its 2000 record low (96th bubble percentile), and household cash allocation reads 14.5%. Beware the “record cash on the sidelines” headline you'll see arguing the opposite: in dollars the cash pile always grows; relative to the market it is supposed to support, it has rarely been thinner. In aggregate, cash cannot “flow into” stocks at all — every buyer hands their cash to a seller — a myth we take apart in its own essay.

Leverage: the accelerant

Valuation extremes funded with cash deflate; valuation extremes funded with borrowed money deleverage — violently. That is why the framework gives leverage its own pillar. Margin debt is watched two ways: the flow (excess leverage — margin growth minus market growth, 25.4 pp today; sustained readings above +20pp marked the late stages of 2000, 2007 and 2021) and the stock (margin debt as a share of M2, 6.1% — the highest share since March 2000, which printed the identical figure). When both are extreme at once, the rally is being carried on borrowed money.

The score — replayed at every peak, with no hindsight

Any basket of gauges can be tuned to look scary today. The test that matters is what the identical framework would have printed at prior peaks, using only data published by those dates. That discipline — release-aware, expanding percentiles, a 20-year rolling cross-check, five years of history before a gauge may vote — is the Bubble Tracker's core, and it produces this calibration:

The 2007 row is the framework's credential. A valuation basket that lights up at every top is a fear generator, not a measurement. This one read almost nothing in October 2007 — correctly, because equity valuations were mid-range then — while its context gauges (real home prices, the Baa credit spread) show exactly where that cycle's excess actually lived.

What happened next — the outcome table

The claim “valuation predicts the decade, not the top” is testable on our own score history, monthly since 1997, joined to SPY's forward returns:

Read the first return column, then the rest. Over the next year, the bands barely differ — that is the timing honesty, in the data. Over three to five years, the bubble-zone months turn negative while the zero-extreme months compound at their best rates, and the average worst drawdown deepens by a third. Valuation is a position-sizing input and a decade forecast; it is not a sell signal.

The breadth modifier

Two markets with the same valuation score can be very different animals. Extremes with broad participation — the average stock making highs alongside the index — have historically run longer than extremes carried by a narrow leadership. The framework keeps breadth out of the score (it measures confirmation, not valuation) and reads it as a modifier via the Hidden Bear Index and Bear Market Breadth. The historically dangerous combination is a high score with deteriorating breadth underneath — covered in depth in the Market Breadth guide.

False positives & honest limits

The valuation toolkit

Related reading

The Market Breadth guide covers the confirmation side — whether the average stock is participating in what the valuation gauges are pricing. The Volatility & Risk guide covers the fast side of risk that actually moves at turns. For the cash-on-the-sidelines myth in full, read There's $8 Trillion Sitting in Money Market Funds, and for the macro backdrop, the Economy section's valuation category holds every underlying series with full history and CSV downloads.