Why have small-caps performed so well this year?

Goldman flagged that the Russell 2000’s first-half 2026 gain (+23%) and 12-month surge (+41%) may not repeat, citing a more subdued second-half backdrop. AI infrastructure contributed ~40% of Russell 2000 YTD return, but last week’s index reconstitution cut AI infrastructure weight from 15% to 7%. Despite modeling 48% EPS growth for 2026 vs 24% for the S&P 500, Goldman said 2026 EPS estimates are down 9% YTD and that elevated valuations plus a hawkish Fed could limit next-12-month returns to low single digits, with ~30% of stocks unprofitable and 29% of debt floating rate.

Analysis

The key change is not the level of the index but the quality of the marginal buyer. Once the reconstitution strips out a chunk of the AI-linked winners, the remaining Russell exposure is more rate-sensitive, more levered, and more dependent on multiple expansion than on self-funded earnings power. That makes the tape fragile: a modest backup in real yields can flip small caps from momentum to de-rating faster than consensus models assume.

The second-order loser set is broader than obvious weak balance-sheet names. Regional lenders, unprofitable software, and lower-quality industrials should underperform if financing costs stay sticky, because the market is likely underpricing refinancing risk over the next 2-4 quarters. By contrast, healthcare M&A creates only a narrow offset: it helps advisors, consolidators, and a handful of takeover targets, but it does little for the index-level beta problem.

Contrarian view: the crowd is extrapolating a catch-up rally even as estimates are still being cut. That combination usually means the easy multiple expansion is behind us, and the next leg requires a real earnings inflection, not just a benign macro backdrop. AI-linked small-cap hardware/storage names like SNDK can still work, but they should be traded as single-name secular stories rather than as a Russell beta substitute.

Key Decisions for Investors

• Short IWM on rallies or via 1-2 month put spreads; target a 5-8% drawdown if the next payrolls/CPI prints keep 10Y real yields elevated. Cover if small-cap breadth improves materially and earnings revisions turn positive for a full reporting cycle.
• Pair long GS / short IWM into the next 4-6 weeks: GS has cleaner exposure to M&A and capital-markets activity, while IWM is the cleaner expression of higher-for-longer rates and refinancing stress. Risk/reward is roughly 2:1 if small caps re-rate lower without a broad market selloff.
• Stay selective in AI-linked small caps such as SNDK; own them only on pullbacks and only as single-name secular longs, not as basket exposure. If the group stops outperforming after the next index/flow window, treat that as confirmation the reconstitution effect is fading.
• Watch healthcare M&A names and advisors for tactical longs only; the catalyst window is 1-3 months, but if deal spreads widen or antitrust scrutiny rises, fade the trade quickly. The tell is whether announced-deal premiums and advisory pipelines keep improving into earnings season.