Major US indices such as the S&P 500 (SPX) and Nasdaq Composite are sitting near record highs, but a large portion of non‑AI and non‑mega‑cap stocks are already in significant drawdowns. This divergence reflects a classic narrow‑breadth regime, where a small group of very large AI‑linked and mega‑cap tech leaders carry the headline benchmarks.
ETFs like SPDR S&P 500 (SPY) and Invesco QQQ Trust (QQQ) are highly exposed to this concentration, given their heavy weights in names such as Microsoft (MSFT) and NVIDIA (NVDA). Cap‑weighted strength in these vehicles is therefore masking a de facto bear market in many smaller‑cap and non‑AI sectors.
Historically, similar phases of narrow leadership and weak internal breadth preceded major downturns, including the Nifty Fifty era in the early 1970s, the late‑1990s dot‑com peak, and the 2007 pre‑GFC setup. In those episodes, divergence between strong indices and a weakening median stock typically persisted for 6‑24 months before visible bear markets emerged.
Pattern reliability has been conditional rather than automatic, but past instances shared common features with the current tape: leadership concentrated in a single dominant growth theme, elevated valuations in the leading cohort, and a growing share of constituents already down 20% or more from highs. With AI‑linked mega‑caps now fitting that leadership profile, index‑level resilience is increasingly dependent on a shrinking group of names.
Terminology
- 01Market breadth: Degree to which many stocks, not just a few, participate in a move.
- 02Cap‑weighted: Index methodology where larger companies receive higher percentage weights.
- 03Bear market: Extended period when prices fall at least 20% from recent highs.
References
- https://aol.com/articles/were-witnessing-stock-market-something-105601000.html
- https://www.fool.com/investing/2026/09/27/history-says-a-bear-market-is-coming-heres-what-th/
- https://www.nber.org/system/files/working_papers/w8638/w8638.pdf
- https://www.newyorkfed.org/research/economists/medialibrary/media/research/epr/09v15n1/0907chav.pdf