Soochow: Short-term market sees a corrective rebound, focus on domestic computing power and the chip semiconductor industry chain, etc.

Soochow released a research report stating that, overall, combined with the seasonal pattern of the "National Day effect," it maintains its judgment that the current market has entered a sentiment repair phase, and the rebound will continue. If subsequent major positive catalysts emerge at the AI/macro level, the rebound expectations can be revised upward. In terms of allocation direction, the current prosperity of AI hardware is solid, long-term growth has not been disproven, and given that share price adjustments have been relatively sufficient, it remains an important lever for obtaining excess returns. Focus on: domestic computing power and the chip semiconductor industry chain, the PCB chain, servers, liquid cooling, cloud/computing power leasing, optical modules, etc. In addition, with the development of the AI industry and the continuous increase in large model penetration, attention should be paid to opportunities in the midstream and downstream of AI, including supporting tools brought by AI application penetration, AI middle-layer tracks, and Workflow/Agent orchestration platforms with a solid native SaaS foundation. Soochow's main views are as follows: The historical "National Day effect" pattern Historically, A-shares have exhibited a certain "National Day effect." By reviewing market performance around National Day over the past 20 years (2006 to 2025), four patterns can be summarized: First, in terms of the broad market, the two trading days before the holiday are often the best window for positioning, and the market may rise rapidly after the holiday. A review of National Day market conditions since 2006 shows that the index undergoes consolidation in the earlier period, stabilizes two days before National Day, and then enters a rebound phase. After National Day, the market may experience a rapid upward move, with the rally mostly lasting until around T+5. Second, at the volume level, the characteristic is "shrinking volume before the holiday and expanding volume after the holiday." Historical data show that pre-holiday market volume usually begins to decline from T-8 (T being National Day), and the trend of shrinking volume typically continues until the first trading day after the holiday. From T+2 onward, the central level of market volume rises significantly, and on-market liquidity gradually recovers. Third, in terms of style, large-cap and consumer styles are relatively favored before National Day; after the holiday, indices generally rise, small caps show better elasticity, and the financial style is clearly stronger. From the perspective of win rate and average interval returns, large caps are relatively more resilient than small caps during the adjustment phase. In the 10/5 trading days before the holiday, the average excess returns of the CSI 300 relative to the Guozheng 2000 reached 1.23pct/2.34pct, respectively; while the return divergence between growth and value styles is relatively mild. If further breaking down into sub-styles, the consumer style has more advantages in win rate and payoff before the holiday, with average absolute returns in the 10/5 trading day intervals before the holiday reaching 0.94% and 1.17%, respectively. In the 10/5 trading days after the holiday, most major market style indices recorded positive returns, with micro-cap and small-cap elasticity leading; in terms of sub-styles, the financial style performed relatively strongly. Fourth, at the industry level, consumer sectors such as food and beverage and pharmaceuticals and biologics performed relatively well, while after the holiday attention should be paid to the elasticity of the large financial sector. Before National Day, food and beverage, pharmaceuticals and biologics, power equipment, and communications all had relatively good win rates and average interval returns. Overall, the consumer sector ranked relatively high in cost-effectiveness; after the holiday, more sectors saw improvements in win rate and average interval returns, among which banks and non-bank financials stood out, and building materials and building decoration in the real estate chain also had relatively high win rates. How should we view this year's "National Day trading"? The short-term market is welcoming a corrective rebound. The September FOMC meeting raised rates by 25BP as expected, and the dot plot still retained policy guidance for further rate hikes within the year. Warsh's remarks at the press conference were generally hawkish, emphasizing the resilience of the disinflation process and the necessity of maintaining restrictive policy. However, the market did not continue the previous tightening trade, but instead staged a typical "bad news exhausted" move. The core reason is that hawkish rate hike expectations had already been relatively fully priced into the earlier market adjustment, and after the, the suppression of risk appetite from the denominator side eased temporarily. At the same time, the AI industry chain adjustment has already been relatively sufficient, the near-term prosperity at the industry level remains solid, and the market has welcomed a repair opportunity, with the STAR Market and ChiNext boards becoming the leading main line of this rebound. However, the current market condition is still essentially a phased repair on the sentiment side, rather than the start of a new round of trend, and the market still shows strong rotation characteristics, with styles also relatively balanced. Over the medium to long term, the core of market pricing lies in two major dimensions: overseas macro constraints and the evolution of the AI industry. At the overseas macro level, the reshoring of U.S. manufacturing and AI industry construction continue to push up capital demand, and combined with the fiscal deficit problem, the central level of interest rates is unlikely to decline systematically; at the same time, under the disturbance of the U.S.-Iran situation, oil prices remain high, bringing inflationary pressure. This setup means U.S. Treasuries face upside risk, and attention needs to be paid to whether geopolitical conditions can improve and bring about a decline in inflation data. At the AI industry level, it is still in a stage of "narrative divergence." In the middle of a bull market, it is normal for controversial industry narratives to disturb the market and bring about phased adjustments in the course of a growth market: in February 2025, affected by the iterative technological development of the DeepSeek large model, the market once fermented the "computing power oversupply theory"; in the fourth quarter of 2025, slower-than-expected progress in Agent commercialization also hit sector valuations. Each round of controversy is accompanied by the digestion of positions and valuations, but it has not reversed the core logic of the industry's long-term upward trend. Recently, the topic of AI safety has gained traction, and the market worries that large model iteration will slow, suppressing capital expenditure and computing power demand. However, driven by commercial competition and great-power technological rivalry, the development of the AI industry will not stagnate. On the contrary, demand for security compliance, private model deployment, and other needs will also bring structural increments in computing power. Against the backdrop that the medium- to long-term growth logic of AI has not yet been disproven, industry controversies will not reverse the market trend. As the industry continues to iterate and mature, the focus of market pricing will also gradually shift from upstream hardware infrastructure to the midstream and downstream AI application implementation. Risk warnings: The pace of economic recovery may fall short of expectations; geopolitical risks; the Federal Reserve may raise rates more than expected; historical experience does not represent the future; statistical differences in measurement may cause data errors.