HOW EXPENSIVE IS THE MARKET?
Three ways to answer the same question.
No single number captures "market valuation" perfectly. This page shows the three most-cited long-horizon measures side by side: the classic trailing PE, Robert Shiller's cyclically-adjusted CAPE, and the market-cap-to-GDP Buffett Indicator. Each looks at a different denominator; together they triangulate. All three currently read historically expensive.
Current Stock Market Price-to-Earnings Ratio
1. Trailing PE — 1871 through today
Monthly readings, connected. The dashed cyan line is the long-term average. The green band below the average marks the "cheap" zone; the red band above marks the "expensive" zone.
Chart y-axis is capped at 50 for readability. The 2009 spike to ~124 is a well-known artifact of the GFC earnings collapse (denominator near zero), not a real "market at 124x earnings" moment.
What happens next when the PE is extreme?
For every month with 12 months of data ahead of it, we recorded the S&P 500's real total return (dividends reinvested, inflation-adjusted) over the next 1 year and next 5 years. Then we bucketed those months by PE ratio. The result is a plain-English answer to: "when the PE was this, the market went on to do this."
2. Shiller CAPE — cyclically-adjusted PE
Divides today's price by the 10-year average of real (inflation-adjusted) earnings. Smoothing over a full business cycle removes the noise that makes trailing PE spike during earnings collapses (2009) or crash during earnings booms. Historically CAPE tops out near 44 (Dec 1999 dot-com peak) and bottoms near 5 (1921). Robert Shiller's canonical long-run valuation measure.
3. Buffett Indicator — US market cap / GDP
Warren Buffett called this "probably the best single measure of where valuations stand at any given moment" (Fortune, 2001). It's total US corporate equity divided by GDP — a rough answer to "how big are stocks vs. the actual economy?" Modern readings run structurally higher than in Buffett's era (different interest-rate, tax, and globalization regime) but the shape still shows extremes clearly.
Buffett's original bands (2001): <70% cheap · 70–90% attractive · 90–115% reasonable · 115–135% expensive · 135%+ danger zone. Post-2013 readings have lived above 135% almost continuously.
4. Composite valuation percentile — four measures on one scale
The three measures above plus Tobin's Q ratio, each scored as a percentile of its own full history: 0 is the cheapest reading that measure has ever shown, 100 the richest. The line is the average of those scores. It is the same construction as Bloomberg's "US stock valuations have never been higher" chart, built from the four of its eight inputs that have history back past 1950 (price-to-book, price-to-sales, EV/EBITDA and forward PE are not available as long free series). Before 1945 only the PE and CAPE contribute.
What the data actually says
Cheap markets pay you.
When the PE ratio dropped below 10, forward 5-year real returns averaged roughly +13% per year. Those readings happened in 1917–1920, the 1930s recovery, 1948–1954, 1974–1982, and briefly in 2009. All were terrifying times to buy — and all of them rewarded buyers handsomely.
Takeaway: The single most consistent pattern in 155 years of stock-market history is that low starting valuations produce high forward returns.
Expensive markets punish you.
When the PE has been above 30 (excluding the 2009 GFC data artifact), forward 5-year real returns have averaged only +1.5% per year and were positive just 43% of the time. This zone captures 1929, 1998–2001, and 2021–today. Compare that to Very Cheap (PE below 10), which delivered +12.4% real annualized with a 92% hit rate.
Takeaway: A high PE doesn't tell you when stocks will fall. It just says the multi-year return that follows is likely to be well below the long-term average — sometimes zero.
PE is a horrible short-term timing tool.
The 12-month forward return is barely correlated with the current PE. A high PE can go higher for years (1997–2000 saw PE climb from 30 to 40+ while stocks doubled). Valuation matters over years, not months.
Takeaway: Don't sell everything because the PE is high. Use it to size your expectations for the next decade, not to time the next quarter.
Methodology & data sources
PE ratio series: Robert Shiller's monthly dataset (Yale) for the historical baseline, extended through the present via multpl.com, which republishes the same series with a monthly refresh.
The PE ratio here is the classic trailing 12-month P/E — index price divided by the sum of the previous four quarters of reported earnings. This is not the Shiller CAPE (10-year cyclically-adjusted PE), which smooths earnings across a full business cycle and typically reads higher.
Forward-return computation: uses Shiller's "Real Total Return Price" series — S&P 500 price with dividends reinvested, adjusted for CPI inflation. Returns shown are annualized real returns, so they're directly comparable across decades regardless of the inflation regime.
Refresh: the current-month PE reading refreshes daily via our GitHub Actions workflow. Real Total Return Price is a stable historical series that only extends as Shiller publishes new numbers (typically once a year).
The 2009 anomaly: the trailing PE hit ~124 in mid-2009 not because stocks were bubblicious but because reported earnings crashed to nearly zero during the financial crisis. The ratio spiked; the expensiveness did not. This is a well-known artifact and why some analysts prefer CAPE.
Shiller CAPE series: same Robert Shiller dataset, refreshed monthly via multpl.com. CAPE = current real price / 10-year average of real earnings.
Buffett Indicator series: computed from two FRED series — NCBEILQ027S (Fed Z.1 Nonfinancial Corporate Business, Corporate Equities as a Liability, in millions of USD) divided by GDP (nominal, seasonally adjusted, in billions of USD), expressed as a percentage. The classic version used the Wilshire 5000 Total Market Full Cap Index as the numerator; FRED retired that series when Nasdaq took over the Wilshire index, so we use the Fed Z.1 nonfinancial-corporate-equity figure that current Buffett Indicator dashboards (longtermtrends, gurufocus post-2023) also use. Slightly narrower scope than the original (excludes financials), but the shape of the series and its interpretation are the same.
Q ratio series: Tobin's Q for US nonfinancial corporations, computed from two FRED series — NCBEILQ027S (corporate equities as a liability) divided by TNWMVBSNNCB (net worth at market value). Quarterly from 1945, refreshed with the nightly workflow.
Composite percentile: for each of the four measures, every reading is ranked against that measure's entire history (the share of all readings at or below it, 0–100). The composite for a month is the plain average of the percentiles available that month; quarterly series carry their latest value forward within the quarter. Ranking against the full history means a new record pushes earlier percentiles down slightly, so the line is redrawn from scratch on every visit rather than stored.
Reminder: None of this is investment advice. Historical valuation patterns are not guarantees. The market has stayed expensive for many years at a time. Always do your own research or talk to a financial professional.