The S&P 500's Hidden Weakness

The index is up 14%. More than 40% of its stocks are down. What does that actually mean for my next investment?

I have roughly $325,000 invested, including Micron, Marvell and other businesses benefiting from AI spending. I’m not exactly complaining about this year’s market.

But the breadth numbers caught my attention.

A Reuters analysis published October 12 found that the S&P 500 was up roughly 14% for 2026, while the median constituent had gained just 4%. More than 40% of the companies currently in the index were down.

Technology contributed over 10 percentage points to that gain. That’s more than 71% of the benchmark’s appreciation coming from one sector.

But the equal-weight S&P 500 ETF, RSP, rose from $191.56 at the end of 2025 to $213.19 on October 12.

RSP price return = $213.19 / $191.56 - 1 = 11.3%.

That’s only 2.7 percentage points behind the cap-weighted index. The median stock isn’t an investable portfolio, and Reuters’ sample excludes companies removed from the index earlier this year. The market is concentrated, but the 4% median doesn’t tell the entire story. (RSP closing prices)

What does concentration actually cost?

As of August 31, the ten largest S&P 500 constituents represented 37.8% of the index, according to S&P Dow Jones Indices.

Imagine those ten falling 20% while the rest rise 5%:

(37.8% x -20%) + (62.2% x 5%) = -4.45%.

That’s a sensitivity calculation, not a forecast. Most of the index could rise and the whole thing would still fall.

Micron, SK hynix and Marvell are different businesses, but their earnings depend on overlapping parts of the AI infrastructure cycle. Three tickers don’t necessarily give me three independent sources of return.

Are the stocks left behind actually cheaper?

CNN’s Fear & Greed Index was around 46, or Neutral, on October 12, according to a tracker reproducing CNN’s data. Interesting next to an index near its highs. Not a valuation model.

Here’s a more useful five-year comparison:

The formula is simple: five-year price multiple = (1 + EPS growth)^5 x (terminal P/E / starting P/E).

The AI business grows earnings faster, but the slower compounder has a higher modeled share-price return because it avoids multiple compression. These scenarios are illustrative; dividends and taxes are excluded.

My recent Moody’s and Intuitive Surgical Base cases offered roughly 9.1% and 9.0% annualized price returns at their dated research prices. Both were below my 12% hurdle for starting an individual-stock position.

I like both businesses. I didn’t buy either.

The next dollar

I’m not selling my AI holdings because the rally is concentrated, or buying laggards just because they’re down. Some of the earnings growth behind this rally is real. Some weaker stocks deserve to be weak.

I want to keep researching compounders whose earnings are less dependent on AI spending. But the next $5,000 still has to compete against what I own and an index fund.

The market’s headline return is hiding something useful.

It isn’t telling me what to buy.

Market breadth and S&P 500 figures reference October 12, 2026. The top-ten weight is dated August 31; hypothetical scenarios are not forecasts. RSP figures are unadjusted closing prices, excluding distributions.