From Oct 6 to Nov 2, 2026 the Nasdaq 100 fell 10.4% while the average S&P 500 company rose. Here is what the economy's own markets (rates, credit, the dollar, commodities) say about it, measured against every comparable episode since 1962.
Over 20 trading days the equal-weight S&P 500 beat the Nasdaq 100 by 15.7 percentage points, the widest gap since the equal-weight fund began in 2003. Chip makers led the fall (down 24% from their June high). Financials and consumer companies rose.
The 10-year Treasury yield fell from 5.27% to 4.91%; almost all of that (−0.29 points) was the real yield, while 5-year inflation expectations rose just 0.02. The 2-year fell 0.20, so the curve flattened from the long end. The Fed's rate is unchanged at 3.88%. Note the run-up first: the 10-year had climbed 0.26 points in the 60 days before the high.
The extra yield risky companies pay over Treasuries (the high-yield spread) widened 0.40 points to 3.43%, wider than on 99% of days in the past year. The spread for the safest companies widened 0.11 to 0.94%, its widest of the year. Broad financial conditions are still easy (Chicago Fed index −0.49, looser than average).
The dollar index is flat (−0.1%) and the dollar against the yen barely moved (−0.2%), so this is not a repeat of August 2024, when cheap yen borrowing was unwound. Oil rose 7.8% in the month and 23% over 60 days: the one series pointing to a supply shock. Gold fell 5.2%, to the bottom of its one-year range. There was no flight to gold.
The Nasdaq's fear meter (VXN, what Nasdaq options price for the next 30 days) is 30.9; the S&P's (VIX) is 21.2. A gap of 9.7 points is wider than on 99% of days since 2010. In 34 past Nasdaq corrections the VIX stood at a median 25 at this depth, so broad fear is still below the usual level.
The line is the 60-day correlation between S&P 500 returns and Treasury returns. Above zero, bonds fall with stocks; below zero, bonds rise when stocks fall. It ran positive in the inflation era (1962–99, median +0.41), negative in 2000–21 (−0.38), and is near +0.11 since 2022. In the last 20 days it turned to −0.39: the 10-year Treasury fund is up 2.7% since the high.
We split all 34 Nasdaq corrections since 1985 by three things in the 20 days before the first −10% close: the 10-year yield falling, the corporate credit spread (Moody's Baa minus the 10-year) widening, and the dollar flat or falling. Today has all three. The 8 past cases with all three (1986, 1989, 1992, 1997, 2000, 2019, 2024, 2025) bottomed at a median −18.1%; only 2 stopped above −15%. Oil spikes, a yen surge, gold and bond volatility did not separate deep from shallow corrections.
The three signs are one theme, not three votes: a Treasury rally widens that credit spread almost automatically. Ten splits were tested, so one could look meaningful by chance. Read it as a tilt.
Options dealers hedge by trading the index. At the Oct 6 high their positions on the S&P 500 meant buying dips and selling rallies (+$28 billion of "gamma" per 1% move). By Nov 3 it was −$25 billion: they now sell into drops and buy into rallies, which makes daily swings larger. The largest single hedge on the board is 430,000 S&P puts at 7,000, about 7.5% below the market, expiring Dec 18.
Earnings or valuation? This brief has no earnings data. Nvidia's report on Nov 17 is the first hard read on whether AI profits or AI multiples are being cut.
Oil. Up 23% in 60 days against a growth scare in bonds. Its pass-through to 2026 inflation is not modeled here.
A new pattern? This is the third 10% Nasdaq correction of 2026 (March, July, now). Whether frequent, narrow corrections driven by a few AI spenders are a new regime is a question for an economist, not for a dataset.
Small samples throughout: 34 corrections, 16 in the closest price pattern, 8 in the closest economic mix. History, not a forecast.
Sources: FRED (Treasury yields, real yields, breakevens, ICE BofA spreads, Moody's Baa, NFCI, Fed funds), Yahoo Finance (indexes, ETFs, futures, currencies, VIX, VXN), Dragon Quant options-map captures. Data through the Nov 2, 2026 close. Method and code: Dragon research, drawdown study 2026-11-03.