CICC: A-Share Pullback Has Already Priced In Overly Pessimistic Expectations
N.R. Finch
The CSI 300 has fallen 7% from its June peak, with small- and mid-caps down 17-18%. CICC's strategy team argues the sell-off has already priced in excessive pessimism, and a rebound could arrive within one to two weeks.
What caused this sell-off?
CICC attributes the pullback to four pressures hitting at once: a souring AI narrative overseas, Korean leverage contagion, rising oil prices from U.S.-Iran tensions, and overheated microstructure inside A-shares.
This means → no single catalyst is responsible. Three external shocks collided with crowded domestic positioning, and the combined force produced this scale of drawdown.
The Shanghai Composite has turned negative for the year, falling 2.1% today and breaking below its 250-day moving average for the first time since April 7.
How did the AI story flip from tailwind to headwind?
Apple raised Mac and iPad prices due to memory cost pressure. Meta is considering opening some AI compute capacity to outside clients — sparking fears that upstream AI prosperity is cannibalizing downstream demand.
In plain terms = the market previously assumed "AI lifts all boats." Now it sees that upstream profits may come at downstream's expense, and the optimistic consensus has cracked.
The ChiNext index is down roughly 15% from its high; the STAR 50 is off about 10% — the hardest-hit segments in this correction.
What happened in Korea, and why did it spread to A-shares?
Global leveraged ETFs linked to the Korean market once exceeded $50 billion in assets, with high daily turnover relative to Korean exchange volume.
As the memory narrative reversed, the KOSPI has fallen a cumulative 27.5% from its peak. Margin calls triggered a negative spiral — forced selling drove prices lower, which triggered more forced selling.
This means → risk sentiment traveled along a clear chain: Korean leveraged ETFs → U.S. tech → A-share tech hardware. The A-share decline was not an isolated event.
Does A-share market structure bear any blame?
Daily turnover hit ¥3.8 trillion in late June, with turnover rates exceeding 6% — trading itself had overheated.
Last Thursday and Friday, the TMT sector's share of A-share turnover surged to a record 52%; semiconductors alone accounted for 20%.
In plain terms = nearly half of all market capital was crowded into a single theme. Non-AI sectors were starved of liquidity, so when sentiment turned, the stampede was inevitable.
Why does CICC think pessimism has gone far enough?
The CSI 300 equity risk premium — a gauge of stocks' attractiveness versus bonds — has climbed back above its historical mean, and the dividend yield has risen to 2.7%. Valuations now embed substantial pessimism.
CICC cites its own research on three criteria for judging an AI bubble — productivity gains, leverage levels, and secondary-market valuations — and concludes all three remain in relatively healthy territory.
Annual recurring revenue at Anthropic and OpenAI continues to grow. This reflects that the AI business-model thesis remains intact; only sentiment has overcorrected.
If positioning now, where does CICC see opportunity?
Growth with momentum: AI infrastructure plays including optical communications, PCBs and upstream materials, plus innovative drugs entering clinical data validation.
Cyclical improvement: power-grid equipment, petrochemicals, and construction machinery where supply-demand dynamics are tightening, along with oil shipping and minor metals that benefit from geopolitical uncertainty.
CICC maintains its medium-term view that A-shares will grind higher, driven by global-order restructuring and China's industrial innovation — both drivers remain intact. In volatile markets, high-dividend, low-volatility assets may offer relative downside protection.
Content is for reference only, not financial advice.