Half the S&P 500 Is 15% Below Its High, and Most of It Is Still Falling (October 2026)

On 2 October 2026, 271 of 502 S&P 500 members closed at least 15% below their 52-week high, while the index, measured by SPY, sat 1.1% below its own. Most of those 271 are still falling: 193 remain below a declining 200-day trend. The index is also calmer than its members. The typical member's daily swings were more than twice the index's.

If you sell puts on single stocks, those two facts matter more than the headline. Below is one day's closes, what they show, and what an AI-assisted process would check next. None of it is a forecast.

The headline numbers

Distance from 52-week closing high	Members	Share of members	Share of index value
15% or more below	271	54.0%	30.2%
10% to 15% below	118	23.5%	25.0%
5% to 10% below	68	13.5%	15.9%
Within 5%	45	9.0%	28.8%

Further down: 192 members (38%) are 20% or more below their high, 100 (20%) are 30% or more below, and 19 are at least 50% below.

On the same close, the funds:

Fund	Below its 52-week closing high
SPY (S&P 500, cap-weighted)	1.1%
RSP (S&P 500, equal-weighted)	5.9%
QQQ (Nasdaq-100)	0.0%, a closing high on 2 October

Most of the "sale" is still falling

A stock 15% below its high is cheaper. It is only a discount if the selling has stopped. So we split the 271 by trend, using the test the levelbox screener applies to every name: below the 200-day average, with the 50-day average still declining, counts as still falling.

Of the 271 members 15%+ below their high	Members	Share
Still falling (below the 200-day, 50-day declining)	193	71%
Below the 200-day, but the 50-day turning up	34	13%
Back above the 200-day average	42	15%

The largest names still falling include Broadcom, Tesla, Walmart, Costco, Bank of America, GE Aerospace, Morgan Stanley, Home Depot, Goldman Sachs and American Express. The largest that have stopped falling include Intel, Lam Research, Applied Materials, Caterpillar, KLA, GE Vernova, Qualcomm, Seagate and Corning, and several of those are tied to the same AI build-out. These are examples of how the split falls, not a list of picks.

The index is calm. Its stocks are not.

2 October 2026	30-day realised volatility	63-day realised volatility
SPY	10.7%	11.0%
Median S&P 500 member	25.0%	30.0%
Middle half of members	19.5% to 33.9%

Over the last 63 sessions, 328 of 502 members had at least one day that moved 5% or more. SPY had none. The members move a lot, in different directions, so their moves cancel out in the index. A rough measure of how closely members move together, computed from these volatilities, was 0.09 over the last three months, against 0.17 a year earlier.

That low correlation is why the index looks calm. It also describes a calm period. Stocks that move independently on a quiet tape can still fall together in a sell-off. For a put seller, the figure that matters is the stock's own volatility, and for the median member it is more than twice the index's.

Why the index hides it

Two reasons.

Weight. The 45 members within 5% of their highs carry 28.8% of the index's value. Nvidia (0.8% below its high), Apple (2.2%) and Microsoft (4.5%) alone are about 20% of it. The 271 members 15% or more down are more than half the names, yet only 30% of the value.

Timing. A member's 52-week high and the index's high did not fall on the same day. Seventy-three members peaked back in October 2025. The index peaked on 13 August 2026. Since then, 391 of 502 members have fallen, with a median move of −7.4%. Over the same period RSP fell 5.8% and QQQ rose 2.4%. The cap-weighted index held up because its largest members kept rising while most of the rest slid.

By sector

Sector	Members	15%+ below high	Median distance from high	Within 5% of high
Consumer Discretionary	47	70%	−23.6%	0
Utilities	31	65%	−16.5%	0
Communication Services	23	61%	−17.5%	1
Consumer Staples	33	61%	−18.8%	0
Industrials	83	61%	−17.3%	8
Materials	24	58%	−17.3%	0
Real Estate	30	57%	−15.9%	0
Information Technology	74	54%	−18.1%	18
Financials	76	47%	−14.4%	0
Health Care	60	33%	−9.5%	14
Energy	21	29%	−9.9%	4

Six sectors have no member within 5% of a high. Information Technology is split: 18 members near their highs, 40 members 15% or more down.

Which stocks are not down

The 45 members within 5% of their 52-week closing high on 2 October, largest first, fall into six groups:

• Mega-cap tech: Nvidia, Apple, Microsoft.
• AI hardware and networking: AMD, Dell, Arista Networks, Amphenol, Lumentum, Hewlett Packard Enterprise, Keysight, Everpure, NetApp, Zebra Technologies, F5.
• Cybersecurity and monitoring software: Palo Alto Networks, CrowdStrike, Fortinet, Datadog.
• Oil and refining: ExxonMobil, Marathon Petroleum, Valero, Phillips 66.
• Healthcare, mostly lab tools and pharma: AbbVie, Thermo Fisher, Pfizer, Agilent, Waters, Illumina, Biogen, Mettler Toledo, West Pharmaceutical, Viatris, Revvity, Solventum, Charles River, Bio-Techne.
• Industrial equipment and logistics: Deere, Johnson Controls, Emerson, Fastenal, Ametek, Expeditors, Nordson, IDEX.

Warner Bros. Discovery is the only Communication Services member on the list.

The deepest falls

The deepest falls among members include Fiserv (65.2%), Fair Isaac (64.8%), AppLovin (63.4%), CoStar (67.7%), Insulet (62.0%), Charter (61.3%), Boston Scientific (59.4%) and Intuit (58.9%). We looked at AppLovin in detail in our October 2026 SWOT.

Has this happened before?

Yes, several times. The outcomes split two ways: either the leaders fell to meet the rest, or the rest caught up.

The documented narrow tops

When	What breadth looked like	S&P 500 afterwards
1972 to 1973	The "Nifty Fifty" era. Only 18% of industries were above their 50-day average in September 1972 (Matso, using Ken French industry data)	From the January 1973 peak: −22% after 12 months, −48% at the October 1974 low
1999 to 2000	"In 1999, the prices of more than one-half of the 500 stocks fell, even as the index rose 20 percent" (Boston Fed, Kopcke, 2000)	From the March 2000 peak: −24.5% after 12 months, −49% at the 2002 low
2018	Ten stocks produced half of the index's first-half return (Goldman Sachs, July 2018)	From the September 2018 high: −15.8% after 3 months, +2.1% after 12
2021	Five stocks produced 51% of the index's gain since April (Goldman Sachs, December 2021). A record 38% of Nasdaq stocks were down 50% from their highs (SentimenTrader, January 2022)	From mid-December 2021: −19.7% after 6 months, −25% at the October 2022 low

S&P 500 returns in this table are price-only, calculated from index closes.

Our own count since 2003. Equal-weight RSP lets us measure the same gap every day back to 2003. We counted every day on which SPY was within 2% of its 52-week high while RSP was 5% or more below its own, using dividend-adjusted closes. Grouped into episodes:

Episode began	SPY 12 months later	Worst fall within those 12 months
September 2007	−20.2%	−21.9%
January 2012	+15.7%	−2.5%
December 2015	+8.7%	−11.7%
August 2020	+36.2%	−1.5%
April 2023	+22.4%	−2.6%
November 2023	+34.0%	0.0%
December 2024	+17.4%	−16.3%

SPY was higher a year later in six of seven, and fell 10% or more at some point in three. For comparison, on any day since 2003 the chance of a 10% fall within the next year has been 31%. With only seven cases, 3 of 7 cannot be told apart from that. The gap on 2 October 2026 sat in the top 1.3% of all days on which SPY was near its high. The widest was in June 2023, and the following year was strong.

Wider studies. LPL Research counted 21 cases since 1991 of the S&P 500 near a high with weak breadth, and found average returns of +1.7% after 3 months and +6.3% after 12: positive, but below normal, with fewer positive outcomes than usual. A separate count of 13 cases since 1998 (Johnson, Backtest) found that in seven the rest of the market caught up without a large index fall, and in five the index fell 5% or more. And looking at bear-market peaks since 1929, Matso found 17 of 19 had broad participation; only 1973 and 2000 peaked on narrow breadth.

Three things follow from the record. Narrow breadth at a high has meant below-average, more uneven returns, though the averages stayed positive. It came before two of the worst bear markets, but most bear markets did not start that way. And we found no published study of exactly today's measure (more than half of members 15% or more below their highs, with the index about 1% from its own), so the history is a range of outcomes rather than a forecast. In 2000 and 2021, the turn came when the leaders themselves started to fall.

When the leaders fell, who fell more?

A natural follow-up: once the leaders gave way, did the stocks that had already lagged fall even harder? Mostly not. In the classic narrow tops, the leaders fell the most. The exceptions were declines driven by the economy, credit or interest rates, where the laggards fell harder.

Declines led by the leaders

Episode	Leaders	The rest
2000 to 2002	Nasdaq-100 −82.9%, S&P 500 −49.1%	Equal-weighted S&P 500 about −11% over 2000 to 2002; cheapest fifth of stocks roughly flat while the most expensive fifth lost 45%
2022	Nasdaq-100 (QQQ) −34.3%	Equal-weighted S&P 500 (RSP) −20.3%, Russell 2000 −25.7%
2018, fourth quarter	QQQ −22.0%	RSP −19.7%
2025, tariff sell-off	QQQ −22.8%	RSP −16.0%

Declines driven by the economy, credit or rates

Episode	Leaders	The rest	Backdrop
1973 to 1974	Largest tenth of stocks −43%, though single "Nifty Fifty" names fell far further (Polaroid −91%, Avon −86%, Xerox −71%)	Smallest tenth −55%, much of it in 1973 while the leaders still held up	Oil shock and inflation
2007 to 2009	QQQ −51.6%	RSP −59.2%, Russell 2000 −59.4%	Credit crisis, led by the banks
August to October 2023	QQQ −9.9%	RSP −13.3%, Russell 2000 −18.3%	Rising bond yields

Figures run from the index's peak to its trough in each episode. Two points sit underneath them. First, "fell less" is measured from the index's peak, and the laggards had usually fallen a long way before it; on 2 October 2026 the median S&P 500 member was already 16.3% below its own high. Second, cheap stocks fell less than expensive ones in both 1973 and 2000, even in 1973 when small stocks did not hold up. Valuation gave some protection. Having lagged did not, on its own.

Today's backdrop carries both kinds of risk. A reversal in the AI leaders would resemble 2000 or 2022. A shock from bond yields, with the 10-year Treasury yield above 5%, would resemble 1973 or 2023, and the rate-sensitive groups that are already lagging would be the ones exposed.

What would an AI do next?

Hand this data to an AI-assisted process built for selling puts on single stocks, and these are the checks it would run next, in order. Each one uses a number above.

1. Separate the falling from the stopped. Set aside the 193 members still below a declining 200-day trend. Being assigned a stock that keeps falling is the outcome a put seller most wants to avoid. That leaves 76 of the 271 to look at, and the value trap question still applies to each of them: is the business holding up, or just the price?
2. Price each put on the stock's own movement. A 0.10-delta put a month out sits about 3.9% below the price at SPY's 10.7% volatility. At the median member's 25% it sits about 8.8% below, and at a 34% member about 11.7%. A put on a single stock carries that stock's risk, and a calm index says little about it.
3. Count correlated names as one position. Several of the names that have stopped falling (Intel, Lam Research, Applied Materials, KLA, Qualcomm, Seagate) ride the same AI capital spending. With correlation at 0.09 they look independent today. In past sell-offs, names that shared a theme have tended to fall together. Sizing them as one bet assumes the worse case.
4. Watch the three stocks holding the index up. Nvidia, Apple and Microsoft are about 20% of the S&P 500 and all within 5% of their highs. In 2000 and 2021 the turn came when the leaders started to fall. If they break their trend while the rest are still falling, the index has nothing left holding it up; if the rest start rising instead, the gap closes from below, as it did in 2023. Who falls harder if the gap closes from above depends on the cause, as the section before shows.
5. Re-run it every week. This is one day's closes. The split between falling and stopped, the volatility gap and the leaders' trend all change, and the levelbox screener re-reads the trend and a quality score on every name daily.

None of these steps says what the market will do. They say what to check before selling a put into it.

Method

• Members: the S&P 500 constituent list as published on 3 October 2026, 503 companies.
• Prices: daily closes, split-adjusted, not dividend-adjusted, from 3 October 2025 to 2 October 2026 (251 sessions).
• 52-week high: the highest close in that window. Using intraday highs instead, 297 members are 15% or more below.
• Index weights: market capitalisation on 2 October. Share classes of the same company (Alphabet, Fox, News Corp) are counted once.
• Trend test: "still falling" means the 2 October close was below the 200-day simple average and the 50-day simple average was lower than five sessions earlier.
• Volatility: annualised standard deviation of daily log returns over the last 21 and 63 sessions. The 0.10-delta distances use 1.28 standard deviations over 30 days at those volatilities.
• Correlation estimate: SPY's 63-day variance divided by the square of the cap-weighted average member volatility. It is a rough average-correlation proxy, not a measured pairwise figure.
• Excluded: Corteva. Its 1 October spin-off of Vylor (one Vylor share per Corteva share) appears in unadjusted closes as an 84% one-day drop.
• Shorter windows: FedEx Freight, Honeywell Aerospace and Qnity Electronics listed within the past year, so their "52-week high" covers only their trading history.
• Who fell more: post-2003 peak-to-trough moves are our own calculation from SPY, RSP, QQQ and Russell 2000 closes. 1973 to 1974 and 2000 to 2002 size and value figures are our calculation from the Ken French data library (month-end, value-weighted, so intra-month falls were deeper); Nifty Fifty names from Fesenmaier and Smith, The Nifty-Fifty Re-Revisited (2002); the 2000 to 2002 equal-weight figure from Institutional Investor (January 2006) and StockCharts (July 2004); Nasdaq-100 from Ben Carlson, A Wealth of Common Sense (December 2025).
• History sources: Boston Fed, Kopcke, "Has the Stock Market Become Too Narrow?" (2000); Goldman Sachs via FA Magazine (July 2018) and Taipei Times (December 2021); SentimenTrader (6 January 2022); LPL Research via Investing.com (September 2026); Backtest, Johnson (September 2026); Matso, thetrading.tools (September 2026). Past index returns are our own calculation from daily closes.

This article is analytical and educational and is not investment advice. Figures are a point-in-time snapshot and can contain errors. This article was generated with AI assistance. If you spot a mistake, tell us.