CICC: High Crowding Meets Liquidity Headwinds, AI Trade Approaching Critical Validation Point
Alina Collins
CICC flags crowding above the 80th historical percentile in the Philadelphia Semiconductor Index, Korean equities, and STAR 50, with US margin balances at a record $1.3 trillion; July's Q2 earnings season will test whether AI capex logic holds as tight liquidity amplifies volatility risk.
What does "crowding" actually mean here?
Crowding — how much capital is piled into the same trade at once. As of June 13, the Philadelphia Semiconductor Index, Korean equities, and STAR 50 all sit above the 80th percentile historically; the Nasdaq 100, even after a pullback, is near 70%.
US margin balances — money borrowed to buy stocks — have topped $1.3 trillion. Korean margin balances are near ₩38 trillion. Both are all-time highs.
This means → crowding alone does not determine direction, but it sharply amplifies the reaction to any surprise — like a packed cinema during a normal screening: everything is fine until someone shouts "fire" and everyone rushes the exits at once.
Why did May payrolls tank both risk and safe-haven assets simultaneously?
May non-farm payrolls came in far above expectations, triggering a rare pattern: risk assets and safe havens fell together while the dollar surged alone.
In plain terms = everyone scrambled for cash, selling stocks and bonds alike — a textbook liquidity-stress signal.
The Philadelphia Semiconductor Index and Korean equities, the most crowded trades, took the hardest hit; pressure only partly eased after Iran tensions cooled and CPI printed below expectations.
Where are June's liquidity headwinds coming from?
The PBoC has drained a net ¥3.8 trillion since March. DR007 — the overnight interbank rate that signals short-term funding conditions — has climbed from a May low of 1.30% to 1.46%.
The ECB hiked 25 bps on June 11. A Bank of Japan hike in June is nearly 100% priced in; USD/JPY has stayed above 160, and Japan's Ministry of Finance has signaled readiness to intervene in FX markets.
SpaceX's Nasdaq listing is viewed as a potential liquidity drain. This means → each factor alone is "known information," but stacked together they can become the catalyst for a volatility event.
New Fed Chair Waller takes the stage — what changes?
Waller delivers his first public appearance as Chair on June 18 (Beijing time). CICC expects him to favor a trimmed-mean measure — stripping out the most extreme price swings to get a "core mainstream" reading — as his preferred inflation gauge.
The Dallas Fed trimmed-mean PCE runs at 2.35% year-on-year, well below headline PCE at 3.77% and core PCE at 3.29%. In plain terms = a different ruler makes inflation look less alarming, opening more theoretical room for rate cuts.
Waller also advocates shrinking the balance sheet, but stresses the process will be "slow and cautious" and requires first easing the SLR — the supplementary leverage ratio, a regulatory cap on how much banks can lend relative to their capital.
How big is the risk of a yen carry-trade reversal?
The carry trade — borrowing cheap yen to buy higher-yielding assets elsewhere — shows mixed signals: CFTC net yen shorts stand at 137,000 contracts, down from a 2024 peak of 190,000; yet April cross-entity lending by foreign bank branches reached ¥14.7 trillion, above ¥12.0 trillion at end-July 2024.
This reflects a divergence: "visible" positioning has pulled back, but "back-channel" exposure has grown.
CICC's base case: impact remains manageable. But if Iran tensions ease sharply, driving US Treasury yields down fast, or if countries sell Treasuries to restock crude reserves, either could trigger a carry unwind and tighten liquidity.
July: how should we read AI's "midterm exam"?
CICC's prior research found US AI trades follow a rhythm of "two strong quarters, one weak quarter." By that pattern, July is the next inflection — landing right on Q2 earnings season.
The key question: can operating cash flow keep pace with the rapid growth in capital expenditure? This means → if the cash companies earn cannot keep up with what they spend, the AI narrative loses conviction.
On inflation, CICC estimates that if oil holds at its May level, May likely marks the year's CPI peak; if oil falls to $80–90/barrel in Q3–Q4, year-end CPI could drop below 3%, opening a path to cuts. The bar for a hike is equally high — oil would need to stay above $120/barrel, which CICC sees as inconsistent with the interests of both the US and Iran.
Content is for reference only, not financial advice.