Does quantitative strategy boost K-type? CITIC Securities: After the game cools down, attention should be paid to the restoration of some non-AI sectors

The Zhitong Finance App learned that CITIC Securities released a research report saying that the three narrative logics correspond to three levels of K-type differentiation. Some of them stem from rational differentiation due to differences in prosperity, which are quite common in global markets, but there are also additional emotional and financial effects. Domestic AI, which has strengthened independently from overseas since June, is currently the strongest K-type. Among them, there are both industry narratives and quantitative strategy position adjustments. The core is that some quantitative stock selection and index increase strategies require position adjustment under the influence of excessive pressure and capital-side pressure to mitigate negative overruns by increasing exposure exposure similar to “double innovation factors”, while subjective bulls clearly have stronger allocation ratios and pricing power for GEM core stocks than the core target of the ETF-led Science and Technology Innovation Board. This kind of structural difference in quantitative strategies has led to an increase in both quantitative strategies The driving effect of “innovation factors” on the Science and Technology Innovation Board in the process of being exposed Significantly larger.

Since June, it has been observed that more and more public offering products have changed from a state of continuous negative excess amounts to the accumulation of positive excess amounts, proving that this kind of position adjustment is gradually progressing. Overall, K-type differentiation in the domestic market is mixed with too many short-term narratives and financial influences. After the capital game cools down, attention should be paid to the restoration of some non-AI sectors.

CITIC Securities's main views are as follows:

Three narrative logics correspond to three levels of K-type differentiation

It has been repeatedly explained over the past few weeks that the global K-type differentiation has been strengthened by narratives. This round of differentiation can be broken down into three clues. First, there is the differentiation between AI and non-AI on a global scale. A strong dollar and interest rate hike expectations are the macro-amplifiers of this differentiation, and the easiest of the three to loosen. This differentiation stems from differences in AI and non-AI boom (leading to differences in sensitivity to interest rates). It is reasonable and is quite common around the world. Second, it is the “carbon-based” and “silicon-based” antagonistic narrative unique to A-shares, making the A-share non-AI sector seriously outperform overseas, even if they operate the same batch of global open businesses. This differentiation stems from the impact of weak domestic demand data on financial sentiment, and domestic investors will use more stringent standards to examine individual stocks in the non-AI sector. Third, since June, domestic AI has independently surpassed global AI. This is not simply an extension or mapping of the North American AI chain boom. Among them are autonomous and controllable valuation premiums for the domestic chain, restoration of stagnant growth compared to the North American chain since last year, and the financial impact of quantified/indexed strategic position adjustments.

Domestic AI is independent from overseas, and is driven by both industry narratives and quantitative strategies to adjust positions

1) Quantitative stock selection and index growth strategies face pressure from the income side and capital side, and there is a need to adjust positions. According to data from the Private Equity Ranking Network, as of June 30, the 1,236 index-enhancing products that had shown results had an average excess income of only 3.1% in the first half of the year, a significant drop compared to 14.2% in the same period last year. The sample private equity index price index tracked showed that in the first half of this year, the Shanghai and Shenzhen 300 Index, the China Securities 500 Index, and the China Securities 1000 Index had excess earnings of 5.02%, 2.84%, and 2.94% of the respective tracking indices. Compared with 7.60%, 10.86%, and 14.50% in the same period last year, they fell 2.58, 8.02, and 11.56 percentage points respectively. Among them, the increase in the China Securities 1000 Index decreased particularly significantly. The decline in excess earnings will put pressure on redemptions, which may force some managers to adjust their strategies and adjust positions.

2) In order to accurately capture the excess profits of this round of the market, it is necessary to introduce “double innovation factors”. The recent continuous pullback in micro-cap stocks is one of the reasons for the excessive decline in quantitative stock selection and index growth products. The Wande Microcap Equity Index had a cumulative decline of 21.4% from its high in mid-May to July 10, and the Science Innovation and Entrepreneurship 50 Index had a cumulative increase of 10% during the same period. This posed a huge challenge to the strategy of exposing exposure to small and micro market stocks. Currently, the cumulative excess yield of the micro-market index compared to the Science Innovation 50 (since 2025) has fallen from a high of 51.0% to -25.8%. As of July 10, out of CITIC's 30 tier-1 industries this year, only 7 had positive returns (electronics 65.6%, communications 52.3%, building materials 25.2%, machinery 9.7%, coal 7.7%, basic chemicals 5.7%, electricity and utilities 0.4%). The vast majority of industries have turned negative. Sectors and industries are so differentiated that traditional quantitative factors can drag down profits as long as they cover a little wider.

3) Differences in holder structure make the “double innovation factor” position adjustment clearly have a greater effect on the Science and Technology Innovation Board. As of the 2025 annual report (top ten holders), it is estimated that GEM accounts for 67% of institutional holdings, while 50 science and innovation institutions account for only 42%. Furthermore, from an industry+theme perspective, only 25% of technology circuit ETF institutions are more open to science and technology innovation. Institutional holdings of GEM constituent stocks are more concentrated, and the core targets are held by stable institutions such as public funds for a long time, and chips are relatively “sticky”; individual investors account for a higher share of science and technology innovation board constituents and ETFs, stabilizing institutions absorb less of the free circulation market, the price elasticity corresponding to marginal capital inflows is greater, and there are usually fears and transactions based on narrative changes, and the second amplification of sentiment on prices is also stronger. Reflected at the transaction level, the turnover rate and volatility of the Science and Technology Innovation Board is significantly higher than that of the GEM, and among them, the lower the share of institutional holdings, the more significant the component, and the impact of quantitative stock selection and additional capital on the Science and Technology Innovation Board is also relatively more obvious. In the first half of this year, the earnings sensitivity of Science and Technology Innovation 50 and GEM 50 increased at the same time, and the linkage with general small capitalization/quantitative capital increased; however, after entering June, science and innovation were clearly decoupled and the correlation declined rapidly. During this period, Science and Technology Innovation emerged from an independent market represented by semiconductors and more volatile. This decoupling coincides with the weakening performance of the micro market strategy during the year, pointing to the possibility that quantitative stock selection/index growth capital may be shifting from micromarket/broad small market exposure to double innovation. The Science and Technology Innovation Board is superior to the GEM market, and is clearly superior to the Korean and US markets. The expression in the industry narrative is that the domestic AI chain is clearly superior to the North American AI chain.

Positive excess indicates that the share of increase is picking up, and quantitative strategy position adjustment is gradually progressing

Using 221 public broad-based index-enhanced initial funds from all over the market as a sample, the benchmarks cover Shanghai and Shenzhen 300, China Securities 500, China Securities 800, and China Securities 1000. The cumulative daily excess income of each fund compared to its comparison benchmark was calculated separately. The pace of accumulation of excess revenue from the beginning of the year to mid-April was close to that of last year, but it began to drastically outperform the same period last year at the end of April and continued until mid-May. The excess increase in public equity indices coincided with the intensification of K-type differentiation last year. More extreme AI/non-AI differentiation has led to a weakening of the effectiveness of traditional quantitative factors, affecting combined returns, which in turn forces quantitative product positions to be adjusted. However, since mid-May, the average cumulative excess yield of the public equity index has gradually rebounded from a low point of 0.2% to rise to 1.76% on July 8 (3.42% in the same period last year), and even surpassed the previous high during the year (1.43% on April 3), indicating that many products were indeed forced to adjust their positions due to K-type differentiation. If you look at the monthly average excess, these samples were -1.81%, -1.22%, and -1.02% from April to June, respectively, and +0.62% since July; similarly, the proportion of products that received positive surpluses from April to June was 41.18%, 51.58%, and 51.13%, respectively, and 71.49% since July. Overall, the absolute level of negative excess amounts and the proportion of products are decreasing month by month. It also reflects that many publicly traded products are actively adjusting their positions. This position adjustment itself is also increasing K-type differentiation.

K-type in the domestic market contains too many short-term narratives and financial influences. After the game cools down, attention should be paid to the restoration of some non-AI sectors

Since March of this year, three narrative logics have boosted three K-type differentiations: the differentiation between AI and non-AI on a global scale, and the “carbon-based” and “silicon-based” antagonistic narratives unique to A-shares. Since June, domestic AI has outperformed the global AI. Some of these are due to rational differentiation due to differences in prosperity, which are common in global markets, and there are additional effects caused by sentiment and capital. As the passive position adjustment of more and more products comes to an end, the impact of sentiment and capital will also decline. After this financial game cools down, we need to pay attention to the restoration of some non-AI sectors. Specifically, in terms of configuration, the AI side favors strong supply constraints, undervaluation, and mid-downstream varieties, such as cloud vendors, storage, servers, gas turbines, and diesel generators; on the non-AI side, the choices are some basic metals (copper, tin), electronics (wait for negative feelings in the industry chain to be digested and return to objective performance), and chemicals (refrigerants, phosphorous chemicals, spandex, dyes, large-scale refining, etc.). Furthermore, we continue to be optimistic about innovative drugs that continue to have an overseas logic and relatively thorough clean-up of chips. The recently released new version of the basic drug catalogue is the first adjustment since 2018, and profit predictions for some companies may have a significant positive catalyst. Still optimistic about undervalued brokerage firms. Defects such as the suppression of liquidity are expected to gradually subside in the second half of the year. The short logic looks at investment returns and reported performance, while the long logic is the expansion and globalization of the brokerage business driven by the broader wave of overseas investment and financing of Chinese enterprises.

risk factors

Frictions in the fields of technology, trade, and finance between China and the US have intensified; domestic policy strength, implementation effects, or economic recovery have fallen short of expectations; macro-liquidity at home and abroad has tightened beyond expectations; conflicts in regions such as Russia, Ukraine, and the Middle East have further escalated; and China's real estate inventories have fallen short of expectations.