CITIC SEC: The probability of a general recovery in August is still increasing, and the configuration strategy should gradually shift from trading on overselling to a more balanced approach.

CITIC SEC released a research report stating that the current adjustment of A-shares is more of a correction of crowded trading rather than a deleveraging shock like in South Korea. The bank believes that local liquidity pressures still exist, particularly for some non-core AI stocks, whose adjustments have affected the internal positions of the technology sector, causing a temporary failure in the pricing of core targets; however, this impact is now largely eliminated. The probability of a general recovery in August is increasing, but it is not a simple rebound from overselling. The negative narratives surrounding non-AI sectors have marginally improved, and the funding environment also supports appropriate recovery. In terms of allocation, it is recommended to increase exposure to energy and chemical, non-ferrous, non-bank financials, and innovative pharmaceuticals, while the technology sector itself needs to focus more on holdings during the rebound. CITIC SEC's main points are as follows: The current adjustment is more a correction of crowded trading rather than a deleveraging shock like in South Korea. 1) The overall leverage situation is relatively safe, with the number of rising stocks in July actually exceeding that in June. As of July 30, the average guarantee ratio for margin financing in the entire market was 264.5%. Although it fell from 296.4% at the end of June, it reached a low of 261.6% in July, indicating that financing accounts generally have sufficient safety cushions. In terms of market breadth, 2,546 A-shares rose in July, significantly higher than the 1,419 in June; the proportion of rising stocks increased from 25.7% to 46.0%. Among these, non-technology stocks accounted for 2,291 rising stocks, or 50.6%, indicating that the market is not simply contracting, but rather that the previously highly concentrated technology trading is spreading out into more non-technology sectors. The current market adjustment can be seen as a decongestion within technology and a rebalancing of market structure, rather than a systemic reversal of the bull market logic. Even if the Shanghai Composite Index is down about 10% from its peak in this phase, it is still at a relatively mild level compared to major global equity markets (KOSPI -39%, Nikkei 225 -16%, S&P 500 -4%). Overall, the index still possesses strong endogenous stability, and the adjustment of the technology sector does not mean the end of this bull market. 2) Compared to typical deleveraging events in global history, the current decline in financing is not substantial. During this round of adjustment, the financing balance decreased from a peak of 3.01 trillion yuan on June 25 to 2.59 trillion yuan on July 31, a cumulative decline of about 14%; historically, the most severe drop in financing balance occurred in mid-2015, where it fell by 60% from its peak. In similar past scenarios of short-term sharp adjustments, financing balances typically fell by a range of -60% to -14%, with an average drop of -32%. The current decline in A-share financing is at the lower end of this spectrum. However, the financing balance of the TMT sector declined by 20.2% from its peak, exceeding the 13.2% seen in March-April 2022 and the 13.8% at the beginning of 2024, only behind the 56.1% decline in 2025 and 25.3% in early 2026. 3) The ETF market is experiencing continuous inflows, with inflows into technology ETFs also providing liquidity support. From June 25 to July 30, the ChiNext Index suffered a phase decline of 25.8%, while technology ETFs experienced a cumulative net subscription of about 165.5 billion yuan during the same period; among the 15 trading days with drops, 13 recorded net subscriptions, with a total net subscription on down days of about 115.9 billion yuan. If only counting positive subscription amounts on down days, the buying at the bottom was about 121.3 billion yuan, with the largest single day net subscription reaching 23.6 billion yuan. The ongoing incremental buying indicates that the current adjustment does not mean the market's absorption capacity has disappeared; instead, it is more the result of the fragmentation of prior crowded trading, institutions actively reducing concentration, and ETF funds consistently buying against the trend, which is significantly different from historical leverage shocks that led to a disappearance of buying and liquidity exhaustion. Liquidity pressures on non-core AI stocks have led to a temporary failure in the pricing of technology stocks, and this impact has now mostly cleared. During this round of adjustments, the actual stocks that experienced short-term liquidity pressure mainly featured lagging price increases (a major surge this year, with limited increases in recent years), low institutional holdings, high financing proportions, and high chasing costs from previous surges. According to our calculations, there are 37 sample AI technology stocks with significant liquidity pressures: 1) This set of stocks had an average price increase of 114% in Q2 2026, far exceeding the 49% in Q1 2026, with many stocks only catching up to the AI trend this year, with most of the increases concentrated between April and June of this year, predominantly in upstream price chain firms; 2) From the perspective of participating entities, none of these stocks appeared in the top 30 heavyweights of active public offerings during Q2 2026, with most being non-institutional stocks, heavily invested by retail traders, private equity, and industrial funds, and with a high proportion of margin purchases; 3) Funds that chased purchases in May and June suffered substantial floating losses. Our estimates show that by July 31, the average floating loss for weighted costs of buying in May and June was about 34%. These second- and even third-tier AI stocks initially dropped significantly more than core institutional-held technology stocks during the adjustment phase, leading to a passive increase in the weight of core holdings. During this stage, institutional funds struggled to achieve portfolio drawdown control by "selling margins and buying cores," even risking contagion to core stocks. Only when margins have cleared first and core stocks enter the phase of decline will technology funds have the space to actively adjust their portfolios, and the overall liquidity pressure in the sector will subsequently be released. Once this process is complete, differentiation within the technology sector can resume, with the weight of fundamental pricing increasing, thus ending the previous downward negative feedback. From the quantitative tracking indicators, using the previously mentioned 37 AI technology stocks with relatively high liquidity pressure as the "marginal technology holdings" group, and using 10 core technology holdings, represented by optical modules, wafer fabrication, and semiconductor equipment, as the "core technology holdings" group, the excess return difference between them can be used to judge signals for bottoms in the sector adjustment. According to this indicator, the excess return of marginal technology holdings relative to core technology holdings rapidly declined from 31% at the end of June to -26% by July 21, followed by a recovery to around -14% by July 31; this indicator currently shows signs of bottoming out, indicating that overall liquidity shocks in the technology sector have largely ended and will start to differentiate. The probability of a general recovery in August is still increasing, but it is not a simple oversold rebound. 1) Stocks that have fallen significantly tend to have early rebounds, especially those previously subject to liquidity pressure. Our computations show that after severe declines caused by similar past liquidity shocks, the effects of oversold rebounds tend to be realized within 5 to 10 trading days following a low, with the market rebound structure gradually decoupling from the previous downward structure after 10 trading days. The selling pressure following a technological rebound may primarily come from funds that bottom-fished too early during the downturn and from institutional funds that had overly concentrated holdings reallocating. Taking the semiconductor equipment ETF as a typical example, a significant amount of net subscriptions occurred in the early trading days when the corresponding index just began its downward adjustment, and these holdings might become selling pressure during the rebound process. For actively managed products, there are several funds that significantly increased their exposure to technology in June. Among 1,731 flexible allocation funds, 237 increased their Beta relative to the growth enterprise market index by more than 0.05 in June, corresponding to a scale of 208.8 billion yuan; the median Beta for this segment of funds in July still reached 0.44, indicating that their net asset values continued to show a high technology exposure during the technology adjustment phase. These products may eventually adjust their portfolios to reduce volatility due to excessive deviation during future technology rebounds, which is another factor that can constrain the strength of the technology rebound. Thus, even if the overall recovery of the technology sector occurs in August, internal performance may significantly differentiate, and those with larger previous declines might not rebound as strongly. To fundamentally improve the current holding structure, significant breakthroughs that sufficiently expand the imagination for the sector will still be needed on the industrial side to attract incremental funds to absorb the pressure of existing holdings. 2) The marginal negative narratives surrounding non-AI sectors are improving, and the funding environment also supports recovery. The Federal Reserve's July meeting lacked substantial hawkish measures and noted that rising long-term interest rates have led to a certain tightening of financial conditions, further indicating that the Fed does not need to rush to raise rates. Subsequently released U.S. Q2 GDP and June PCE data also came in below expectations, further weakening the short-term rate hike narrative. Over the past few months, a strong dollar and rate hike expectations have been crucial macro factors suppressing demand expectations in non-AI sectors, heightening K-shaped market differentiation; this narrative's marginal shift facilitates some recovery in non-AI sectors. This weeks Politburo meeting also adopted a more cautious tone on the economy, shifting from emphasizing "a strong start with major indicators exceeding expectations" in April to paying close attention to the difficulties and challenges in economic operations, further stressing timely plans for incremental policies and increasing counter-cyclical adjustment efforts. Unlike the downward revisions of relatively high growth expectations throughout the first half, the second half of the market will start from a relatively cautious outlook and will gradually await preferential fiscal spending, monetary tool adjustments, and the implementation of incremental policies for marginal changes. The expected directional shift implies that the market structure in the second half may no longer just be about the strong getting stronger; policy support and expectation recovery may help expand the trend from a single high prosperity to more sectors. Moreover, the market's capital ecological environment in the second half is also favorable for the recovery of non-AI sectors. In the first half, the cumulative net redemption of A-share ETFs reached 1.63 trillion yuan; entering July, the direction of ETF subscriptions and redemptions has significantly reversed, with a cumulative net subscription of 478.4 billion yuan throughout the month, achieving net inflows on 19 out of 23 trading days, which has replenished around 29% of the cumulative net redemption from the first half. For the non-AI sectors already facing low institutional holdings, the return of broad-based ETFs has formed a more stable and balanced passive absorption; combined with the previous loosening of crowded trades, the pricing constraints on non-AI sectors are expected to ease, creating a capital foundation for style balancing and widespread recovery in August. It is recommended to increase exposure to energy and chemical, non-ferrous, non-bank financials, and innovative pharmaceuticals, while focusing technology holdings during the rebound. The short-term market rebound may show more characteristics of oversold recovery, with stocks that have seen significant prior declines and have sufficiently cleared holdings expected to perform better, particularly the non-core AI varieties that previously experienced deleveraging and liquidity shocks. This type of rebound mainly arises from improved liquidity, restored risk appetite, and short covering, and does not imply that the original industrial logic and valuation system have been re-established. As market liquidity and price discovery mechanisms return to normal, the allocation thinking for August should gradually shift from trading excessive dips towards achieving balance. Use the rebound to optimize the portfolio structure and revert to pricing logic based on fundamentals, industry position, and medium-to-long-term profitability. We maintain a mid-term view of "three convergences": 1) Within the AI industry chain, the excess returns of upstream hardware and price-increasing categories are tending to converge with those of downstream platforms, cloud services, and core application segments; 2) The valuation discount of non-AI industrial sectors relative to comparable overseas companies is expected to stage a recovery; 3) The extreme differentiation between technology and non-technology sectors is tending to converge. Within the technology sector, it is advisable to adjust towards core assets (such as leaders in optical communication, wafer manufacturing platforms, and semiconductor equipment) in a timely manner by leveraging rebounds in marginal varieties. For non-technology sectors, it is crucial to increase exposure to energy and chemical, non-ferrous, non-bank financials, and innovative pharmaceuticals. Risk factors: Increasing frictions in U.S.-China technology, trade, and finance; domestic policy strength, implementation effectiveness, or economic recovery may fall short of expectations; macro liquidity tightening both domestically and internationally may exceed expectations; further escalation of conflicts in regions such as Russia and Ukraine, and the Middle East; slower than expected absorption of real estate inventory in China.