Historical US stock market valuations

Last updated: 2026-09-25

The short answer

The market-cap-weighted trailing P/E for US large-cap stocks (market cap above $10 billion) is currently 26.4x (September 2026). Over the past 31 years, it has ranged from 12.9x (year-end 2011) to 31.9x (year-end 2020).

The median year-end value across this period is approximately 20x. By this measure, the market is above its historical median but below the 2020 COVID-rebound peak.

Other metrics tell a consistent story. Price-to-book is 5.3x (range: 1.9x to 6.0x). The dividend yield is 0.97% (range: 0.97% to 3.16%). Price-to-sales is 3.6x (range: 1.1x to 3.6x). All four measures are near or at their 31-year extremes.

Market-cap-weighted trailing P/E

Aggregate trailing P/E for US-domiciled common stocks with market cap above $10 billion, sampled at each calendar year-end. See methodology for computation details.

* 2026 is a partial year (sampled September 2026). P/E = total market cap / total TTM net income (profitable companies only). "Stocks" = number above $10B market cap. "Profitable" = with positive TTM net income.

Other valuation measures

Price-to-book, dividend yield, and price-to-sales for the same universe, computed with the same market-cap-weighted methodology.

* 2026 partial year. P/B = total market cap / total book value (positive book only, P/B < 100x filter). Dividend yield = market-cap-weighted average of TTM dividend yields. P/S = total market cap / total TTM revenue.

What the P/E ratio tells you (and what it does not)

A market-wide P/E ratio is a snapshot of how much investors are paying per dollar of trailing earnings. It is useful as a rough gauge of valuation levels over time, but it has real limitations.

High P/E does not always mean overvalued. The 31.9x at year-end 2020 was the highest in this dataset, but it reflected the gap between prices (which had recovered from the COVID crash) and trailing earnings (which still reflected the pandemic downturn). Within a year, earnings caught up and the P/E normalized to 24.1x without prices falling.

Low P/E does not always mean cheap. The 13.1x at year-end 2008 looked like a bargain on paper, but earnings were about to collapse further in Q1 2009. The same stock prices with lower future earnings would have produced a higher P/E. The trough P/E at the bottom of a crisis is often misleading because the denominator has not yet caught up with the damage.

Sector composition changes over time. Technology companies made up roughly 15% of US large-cap market cap in 2000 and roughly 35% in 2026. Higher-growth sectors tend to trade at higher P/E ratios. Part of the upward drift in the aggregate P/E reflects a structural shift toward sectors that the market values more richly, not just a rise in valuations within sectors.

The P/E is backward-looking. It uses trailing twelve-month (TTM) earnings, but the market prices forward expectations. A stock trading at 30x trailing earnings might be at 20x next year's expected earnings if growth is strong. Forward P/E, available for individual stocks via analyst estimates in the database, addresses this partially.

What about the Shiller CAPE ratio?

The Shiller CAPE (Cyclically Adjusted Price-to-Earnings) ratio, developed by Robert Shiller, uses 10 years of inflation-adjusted earnings as the denominator. This smooths out business-cycle swings and produces a more stable long-term measure. The CAPE has been a better predictor of 10-year forward returns than trailing P/E in historical backtests.

Shibui does not compute the CAPE ratio. The database does not include inflation data (CPI), and the 10-year earnings lookback requires a different aggregation methodology than what is shown here. The trailing P/E on this page is a simpler, more reproducible metric that tracks similar directional trends (both were elevated in 1999 and low in 2008-2011) but should not be used interchangeably.

For Shiller CAPE data, see multpl.com/shiller-pe, which maintains the original Shiller dataset with monthly updates.

Methodology

Reproducible in full. Every number above comes from one query against the Shibui Finance database.

Universe. US-domiciled common stocks (`country_iso = 'US'`, `type = 'Common Stock'`) with market capitalization above $10 billion on the measurement date. The threshold is a convention. You can change it to $1B, $50B, or any other cutoff when reproducing.

Sample date. Last trading day of each calendar year. 2026 is a partial year (sampled 2026-09-25).

Trailing P/E. Market-cap-weighted: total market cap divided by total TTM net income, counting only companies with positive TTM net income. This means loss-making companies are excluded from the denominator, which inflates the P/E slightly compared to methods that net losses against gains. The alternative (including losses) produces a P/E that swings wildly during recessions as the denominator approaches zero.

Price-to-book. Total market cap divided by total book value, counting only companies with positive book value and P/B below 100x (to exclude extreme outliers from near-zero book equity).

Dividend yield. Market-cap-weighted average of individual stock TTM dividend yields from the `fundamentals_derived_daily` table.

Price-to-sales. Total market cap divided by total TTM revenue, counting only companies with positive revenue.

This is not the S&P 500. The S&P 500 is a committee-selected index with 500 specific constituents. This page shows a market-cap-weighted aggregate of all US common stocks above $10B, which varies from 44 stocks (1995) to 672 (2026). The two overlap substantially in recent years but are not the same thing, especially in the 1990s when fewer companies exceeded $10B.

The SQL

The P/E query. The P/B, dividend yield, and P/S queries follow the same structure with different ratio columns.

WITH year_end_dates AS (
  SELECT EXTRACT(YEAR FROM date)::INT AS yr,
    MAX(date) AS yr_date
  FROM shibui.stock_quotes
  WHERE ticker = 'SPY'
    AND date >= DATE '1995-01-01'
  GROUP BY 1
),
agg_by_year AS (
  SELECT yd.yr,
    SUM(v.market_cap) AS total_mktcap,
    SUM(CASE
      WHEN dd.earnings_yield > 0
      THEN v.market_cap * dd.earnings_yield
      ELSE 0
    END) AS total_earnings,
    COUNT(*) AS stock_count,
    SUM(CASE
      WHEN dd.trailing_pe IS NOT NULL
        AND dd.trailing_pe > 0
      THEN 1 ELSE 0
    END) AS profitable_count
  FROM year_end_dates yd
  INNER JOIN shibui.fundamentals_derived_daily dd
    ON dd.date = yd.yr_date
  INNER JOIN shibui.valuation v
    ON dd.symbol = v.symbol AND dd.date = v.date
  INNER JOIN shibui.general_info g
    ON dd.symbol = g.symbol
  WHERE g.type = 'Common Stock'
    AND g.country_iso = 'US'
    AND v.market_cap > 10e9
  GROUP BY yd.yr
)
SELECT yr,
  ROUND(total_mktcap
    / NULLIF(total_earnings, 0), 1) AS weighted_pe,
  ROUND(total_mktcap / 1e12, 1) AS mktcap_tln,
  stock_count, profitable_count
FROM agg_by_year
ORDER BY yr;

Limitations

Stated plainly, because a reference table is only useful if you know where it breaks.

1. This is not the S&P 500 P/E. The S&P 500 is a committee-selected index. This page shows a market-cap-weighted aggregate of all US common stocks above $10B. The universe varies from 44 stocks (1995) to 672 (2026), which is much broader than 500 constituents. The two measures track similar trends but are not identical.
2. This is not the Shiller CAPE. The CAPE uses 10 years of inflation-adjusted earnings. This page uses trailing twelve-month earnings with no inflation adjustment. The CAPE is a better predictor of long-term returns; this metric is simpler and reproducible from the database.
3. Survivorship bias. The database contains currently listed companies. Stocks that were large-cap in 2005 but have since been acquired, delisted, or gone bankrupt are not in the sample for historical years. This biases the historical P/E slightly upward (survivors tend to be more profitable).
4. Loss-making companies are excluded from the P/E denominator. Only companies with positive TTM net income contribute to the earnings total. This makes the P/E lower than it would be if losses were netted. The profitable-count column shows how many of the total contributed to the denominator.
5. The $10B threshold is a convention. In 1995, 44 companies exceeded $10B. In 2026, 672 do. The threshold creates a sample that grows with inflation and market growth. A fixed-percentile threshold (top 500 by market cap each year) would produce different results.
6. Nothing here is predictive. Historical valuation ratios describe where the market has been. They do not predict where it will go. The market can stay above its historical average for extended periods.

Frequently asked questions

What is the historical average P/E ratio for the US stock market?

The market-cap-weighted trailing P/E for US large-cap stocks ($10B+ market cap) has ranged from 12.9x (year-end 2011) to 31.9x (year-end 2020) over the past 31 years. The median year-end value is approximately 20x. This is computed from individual stock data in the Shibui Finance database, not from an index provider.

Is the stock market overvalued right now?

The current weighted trailing P/E is 26.4x (September 2026), above the 31-year median of approximately 20x. Price-to-book is 5.3x (range: 1.9x to 6.0x). Dividend yield is 0.97% (near the 31-year low). These metrics suggest above-average valuations, but a P/E ratio does not account for interest rates, expected earnings growth, or sector composition shifts. This is data, not a recommendation.

What was the lowest market P/E ratio?

In this dataset, the lowest year-end market-cap-weighted trailing P/E was 12.9x at the end of 2011, followed by 13.1x at the end of 2008. Both came after bear markets when prices were depressed relative to trailing earnings. The 2008 value was deceptive because earnings were about to fall further in Q1 2009.

Can I run my own valuation analysis?

Yes. Connect the free Shibui Finance MCP server to Claude and ask it to compute any valuation metric. Change the market-cap threshold, filter by sector, use operating P/E instead of trailing P/E, or add EV/EBITDA. The SQL on this page runs in under a second.

Is this the same as the Shiller CAPE ratio?

No. The Shiller CAPE uses 10 years of inflation-adjusted earnings. This page uses trailing twelve-month earnings without inflation adjustment. Both show similar directional trends (elevated in 1999, low in 2008-2011) but should not be used interchangeably. For Shiller CAPE data, see multpl.com.

Cite this

Free to use with attribution.

Shibui Finance, Historical US Stock Market Valuations: P/E, P/B, and Dividend Yield Since 1995, September 2026. https://shibui.finance/historical-market-valuations

Charts and tables may be reproduced with a link to this page. If you need a different market-cap threshold, sector filter, or additional metrics, email chris@shibui.finance and we will run it.

Reproduce it yourself

Every figure on this page came from a natural-language question asked of Claude with the Shibui Finance connector attached. No spreadsheet, no Python, no data download.