CITIC Securities: Quantitative Rebalancing Intensifies K-Shaped Divergence, Recovery in Non-AI Sectors Worth Watching
N.R. Finch
CITIC Securities argues that passive quant rebalancing has amplified A-shares' K-shaped divergence, and as that rebalancing winds down, the pressure on non-AI sectors should ease — opening a recovery window.
What is driving the "AI up, everything else down" split?
CITIC breaks the current K-shaped divergence — where one group of stocks surges while the rest keeps falling — into three narrative threads.
The first is global: AI-linked companies genuinely enjoy far higher earnings momentum than non-AI peers. Every major market shows this.
The second is A-share-specific: a "carbon-based vs silicon-based" framing — human-driven traditional industries vs chip-and-compute-driven tech — that has pushed A-share non-AI names further behind their overseas equivalents.
The third is the most important: since June, China's domestic AI chain has outperformed global AI names, driven not just by industry fundamentals but also by quant funds forcibly rebalancing into these stocks.
Why were quant funds forced to chase the STAR board?
Index-enhancement products — funds that track a benchmark while trying to beat it — earned just 3.1% average excess return in H1, down from 14.2% a year earlier. That is a more-than-70% collapse.
CSI 1000 enhanced-index excess fell from 14.50% to 2.94%, the steepest drop. This means → the old playbook of picking micro-cap stocks for alpha stopped working.
The WIND micro-cap equal-weight index fell 21.4% from its mid-May peak to July 10, while the STAR-ChiNext 50 index rose 10% over the same period. In plain terms = the small-stock pool quant funds relied on for returns collapsed, forcing them into the better-performing STAR board.
Institutional ownership on the STAR board is only about 42%, well below ChiNext's 67%. Retail investors hold a larger share, so marginal inflows move prices more. This reflects why China's domestic AI chain looks stronger than its U.S. counterpart — part of the gap is fund-flow mechanics, not fundamentals.
How far along is the rebalancing?
Using 221 public broad-based enhanced-index funds as a sample, average cumulative excess return recovered from a mid-May low of 0.2% to 1.76% by July 8.
The share of products generating positive excess rose month by month: April 41% → May 52% → June 51% → July so far 71%. This means → a growing majority of funds have already completed their rebalancing and are back to generating alpha.
CITIC concludes that as passive rebalancing nears its end, the extra divergence driven by fund flows and sentiment should fade, and the K-shaped split should narrow.
Where should investors look now?
AI side: CITIC favors names with tight supply, low valuations, and mid-to-downstream positioning — specifically cloud vendors, memory, servers, gas turbines, and diesel generator sets.
Non-AI side: watch base metals (copper, tin), new energy (wait for negative sentiment to clear, then refocus on actual earnings), and chemicals (refrigerants, phosphorus chemicals, spandex, dyes, large-scale refining).
Two standalone themes: first, innovative pharma — a new National Essential Medicines List, the first revision since 2018, could meaningfully lift earnings estimates for select companies. Second, low-valuation brokerages — liquidity headwinds should ease in H2, while the longer-term thesis rests on Chinese corporates' overseas financing wave driving brokerages' international expansion.
How do we test whether this call is right?
The key checkpoint is clear: does the K-shaped divergence narrow as fund-flow pressures cool?
If, after July, the share of positive-excess enhanced-index products keeps rising and the gap between micro-caps and the STAR board shrinks, the rebalancing thesis holds and non-AI recovery logic is confirmed.
If, instead, the split widens further, that would suggest fundamental differences — not fund flows — are the primary driver, and non-AI recovery will require a more tangible earnings inflection point.
Content is for reference only, not financial advice.