Invest1 publisher3 min readPublished
Goldman's record negative-beta count measures how heavily the S&P 500 leans on a few AI stocks
Goldman Sachs says a record share of S&P 500 stocks now show negative beta, nearing half the index over a short recent window. The readings track how far a few AI names set the index's direction, so a low-beta portfolio can hold risks its beta hides.
The Investor · Invest desk
What happened
- Goldman Sachs says a record share of S&P 500 companies now show negative beta, having recently tended to fall when the index rises and rise when it falls.
- Over a short recent window that group has approached half the index, while over one year the share is smaller but still historically elevated.
- The gap between the index and the median stock's distance from its own 52-week high is among the widest in decades.
- Goldman compared the pattern with the late-1990s technology bubble, when a few high-flying names pulled the averages away from the typical stock.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Position sizing and hedging built on S&P 500 beta now rest on a benchmark set by a few AI names, so a low reading misstates how volatile a stock itself is.
- exposure Portfolios tilted toward the laggards carry bond-yield, dollar, energy and financing risk that a low or negative index beta does not register.
- decision Holding negative-beta stocks as protection amounts to betting that the next mega-cap selloff spares the broader economy; a rate or dollar shock would test that directly.
Beta measures a stock's moves against a benchmark [4]. Change what the benchmark holds and every constituent's beta changes, even if nothing has happened at the company [17]. Right now the S&P 500 leans on a few names. A small cluster of mega-cap technology and AI stocks holds a large portion of its value [5], and those same names carry higher betas than the rest of the index [9]. When an energy producer or an industrial shows a negative beta, the reading mostly records that it moved against the leaders on the days they set the index's direction [17]. Crowdfund Insider, reporting the Goldman work, puts it the same way: the prevalence reflects concentration more than hundreds of companies turning into defensive hedges [7].
The two measurement windows agree on direction and differ on strength. The short-window share is close to half the index. The one-year share is lower [2], so the effect is heaviest in the most recent data [18]. The report does not give the exact percentages or the length of the short window.
Suppose the leaders stumble. Historically, that is when extreme divergence has reversed, and suddenly [15]. If the negative-beta group then rises, those stocks were hedges after all, and the concentration reading is wrong. The stumble could instead come from the pressures already weighing on the laggards: higher bond yields, a stronger dollar, elevated energy prices and tighter financing [12]. Then both groups can fall together, and the negative beta will have been a product of one window's correlations. If nothing stumbles, the share stays high, and the index can keep making new highs while large parts of corporate America trade to a different rhythm [19].
I'd expect the second outcome more than the first. The pressures the source lists on the laggards are tied to the broader economy [12], and I see nothing in them that would make those stocks rise when AI stocks fall. The counter-case is the one-year figure. A lower share over the longer window means part of the short-window reading could fade without any shock at all [2].
The allocation consequences split between active and passive money. An active manager who stays out of the mega-caps can end up looking more defensive than intended, because the benchmark is defined by a few names and not because of any deliberate hedge [14]. A beta of 0.6 on that book may say more about missing exposure to the AI theme than about low volatility [13]. Index funds move the other way, buying more of the leaders automatically as their weights grow [11]. Neither group is buying the laggards because they lag.
Goldman's comparison is with the late-1990s technology bubble [8]. The source concedes one difference: earnings at today's leaders have been stronger than those of many late-1990s darlings [16].
What to watch
- Goldman's next reading of the negative-beta share, and whether the one-year figure climbs toward the near-half short-window level or the short-window share falls back.
- How the negative-beta group trades on the next sharp fall in mega-cap AI stocks: gains would support the hedge reading, losses the concentration reading.
- Bond yields and the dollar, the pressures named on the laggards, as a possible cause of a selloff that hits both groups at once.