The market in early August is in a phase of oversold rebound, where sectors with larger prior declines show greater elasticity. Currently, the year-to-date returns of the non-ferrous metals, chemical, non-bank financial, and electrical equipment/new energy industries remain below their theoretical midpoints. We have conducted a quantitative assessment of the repair progress for popular sector directions from three perspectives: holding costs, margin unwinding, and congestion levels.
From the perspective of holding costs, floating loss pressure is concentrated in tech-growth and small/micro-cap stocks, which still need further digestion. From the margin perspective, the unwinding process for the leading sectors in this round has already passed the halfway mark. From the congestion perspective, trading activity in the tech sector has not yet cooled down. Overall, sectors like electronics, non-ferrous metals, innovative drugs, and non-bank financials are progressing faster in their repair, while chemicals, electrical equipment/new energy, and communications are relatively slower.
In terms of asset allocation, tech holdings should continue to concentrate towards core assets. For the non-tech portion, increased allocation to energy/chemicals, non-ferrous metals, innovative drugs, and leading brokerages is recommended.
In early August, the market is in an oversold rebound phase. The correlation coefficient between July's gains/losses and this week's (the first week of August) gains/losses is -0.85. Sub-sectors with deeper prior declines saw stronger rebounds this week. Leading gainers like PCB, semiconductor materials, and cables rose 33.0%, 27.6%, and 23.4%, respectively, after falling 35.4%, 43.8%, and 45.4% in July. This characteristic aligns with the judgment that after severe declines caused by liquidity shocks, the oversold rebound effect is typically concentrated in the 5-10 trading days following the low point. Notably, the recovery magnitude in early August correlates with the prior decline magnitude for both the tech and non-tech sectors, except that for tech, the relevant decline is from July, while for non-tech, it is from the second quarter. Therefore, the current market repair is still largely in the oversold rebound stage and has not yet transitioned to a phase driven by fundamental trend pricing.
Currently, the year-to-date returns of the non-ferrous metals, chemical, non-bank financial, and electrical equipment/new energy industries remain below their theoretical midpoints. We estimated the difference between the theoretical expected return and actual return for various style, industry ETFs, and broad-based indices. The theoretical expected return is estimated using the return of the CSI Dividend Index plus the upward/downward revision of the year's earnings expectations. The actual return is calculated as the year-to-date return as of August 7. In terms of style, among the ten style factors analyzed, the year-to-date returns for momentum, technology, liquidity, and size factors are still significantly higher than their theoretical expected returns, while growth and leverage styles are significantly below. Regarding industry ETFs, after the July adjustment, the year-to-date returns of some popular industry or theme ETFs have fallen below their theoretical expected returns, typically including communications, batteries, and power grid equipment. The year-to-date returns for non-ferrous metals, chemicals, and non-bank financials have consistently remained below their expected returns, suggesting market investors are concerned about the sustainability of earnings in these industries and are unwilling to assign higher valuations. At the broad-based index level, the year-to-date return of large-cap blue-chip indices is relatively lower compared to their theoretical expected return position. The ChiNext 50 has returned to below its theoretical value after the July adjustment, while the STAR 50 is one of the few indices with significant gains despite no obvious upward revision of earnings forecasts for the year.
From the perspective of holding costs, floating loss pressure is concentrated in tech-growth and small/micro-cap stocks, which still need further digestion. We constructed a surviving weighted cost model, focusing on positions built since the second quarter of this year, to calculate their weighted average cost. As of August 7, the overall A-share cost deviation rate was -3.9%, with a profitable position ratio of 44%. The clear differentiation is evident across industries, broad-based indices, and themes. The pressure from holding costs is highly concentrated in tech-growth and small/mid-cap stocks, while average holdings in dividend and resource sectors face essentially no floating loss pressure. At the industry level, the deepest negative cost deviation rates were in defense, communications, electrical equipment/new energy, automotive, and light manufacturing. In contrast, the deviation rates for pharmaceuticals, non-ferrous metals, oil/petrochemicals, banks, and coal were positive, meaning their latest prices are still above the average cost of positions built since the second quarter. In terms of style, the deepest negative deviation rates were in growth and volatility, while the momentum portfolio maintained a positive deviation rate after this adjustment. At the broad-based index level, the trapped depth for the CSI 2000, ChiNext 50, and CSI 1000 was significantly greater than for the CSI 300. The deeper the market capitalization, the heavier the cost pressure. For popular themes, the deepest negative cost deviation rates were in power grid equipment, robotics, and batteries. Looking at the structure of above-cost positions, the average floating loss in tech-growth directions is relatively deep, and the proportion of deeply floating loss positions is high, making the selling pressure from breaking even during a rebound relatively stubborn. Segregating positions with costs above the current price into those 0-15% above and those more than 15% above, the proportion of deeply floating loss positions was highest in defense, communications, basic chemicals, and machinery. For the broad-based indices, the CSI 2000 had a 40% proportion of such positions, and for themes, power grid equipment and robotics were in the same tier. We also calculated the unwinding turnover sufficiency for each industry. Non-AI sectors that adjusted earlier in the second quarter have already fully turned over, while tech directions that led the decline since late June still need further turnover for digestion.
From the margin perspective, the unwinding process for the leading sectors in this round has already passed the halfway mark. We measure the unwinding progress of this round's net margin increase by the ratio of the peak year-to-date margin balance minus the latest value over the peak minus the low point before the peak. As of August 7, the unwinding progress for directions that only peaked in margin balances from late June to early July is generally low. Industries like electronics, non-bank financials, building materials, and non-ferrous metals had ratios below 0.6. For broad-based indices, the STAR 50 had a ratio of only 0.51, and the CSI 300 was 0.67. Among styles, liquidity and momentum were at the bottom. For the 21 industries with an unwinding progress greater than 1 (meaning the entire net increase has been fully unwound), the peak margin balance almost all occurred in January-March. At the theme level, the unwinding progress for semiconductor equipment, STAR Chip, AI price chain, and power grid equipment was also around the halfway mark. Securities insurance was the lowest among all groups, while batteries was the only popular theme that achieved excess unwinding.
From the congestion perspective, trading activity in the tech sector has not yet cooled down. The trading concentration in A-shares remains high. The proportion of turnover from the top 5% of stocks stands at 51.4%, only slightly down from the high of 51.9% on July 15, and is still at a high percentile of 98.3% over the past five years. For major broad-based indices, the congestion in the STAR and ChiNext boards remains very high, while the trading congestion of small/mid-cap indices has significantly dropped. At the industry level, the changes in congestion for recently hot sectors have roughly fallen into three types. The first type is where trading congestion remains at high levels, with very strong herd mentality, typical of electronics, communications, and building materials. The second type is where trading congestion has rapidly declined, including basic chemicals, electrical equipment/new energy, non-ferrous metals, power/utilities, and oil/petrochemicals. The third type is where trading congestion has never been high, typical of pharmaceuticals, non-bank financials, food & beverages, and banks. At the style and theme level, congestion for the dividend style is currently low, while AI applications and non-AI popular themes have experienced varying degrees of congestion decline.
Comprehensively analyzing the three aspects of holding costs, current margin situation, and trading congestion, the unwinding progress of current popular industry themes is as follows. The electronics sector, despite high congestion, has seen its margin unwinding process pass the halfway mark and faces very little floating loss pressure. Non-ferrous metals, innovative drugs, and non-bank financials have low floating loss pressure, fast margin unwinding, and low current trading congestion. Electrical equipment/new energy and chemicals, although their margin unwinding process is acceptable and trading congestion has declined, still face significant floating loss pressure. Communications faces deep floating loss pressure, relatively slow margin unwinding, and very high trading congestion. After the comprehensive repair in early August, the popular industries with year-to-date returns still significantly below their theoretical expected returns are non-ferrous metals, chemicals, non-bank financials, and electrical equipment/new energy. Communications and innovative drugs are near their theoretical expected returns. The year-to-date gains for the electronics sector far exceed its theoretical expected return, likely reflecting long-term earnings narratives rather than this year's earnings forecast adjustments.
In terms of asset allocation, the market has begun a repair process. As market liquidity and price discovery mechanisms normalize, the allocation strategy for August should gradually shift from trading oversold bounces to a more balanced approach. Use the rebound to reduce exposure to marginal assets and rebalance towards tech leaders with core competitiveness, as well as non-tech cyclical directions with solid fundamentals, low valuations, and attractive position levels. We maintain the medium-term judgment of "three convergences." First, within the AI industry chain, the excess returns of upstream hardware and price-sensitive products relative to downstream platforms and cloud services tend to converge. Second, the valuation discount of non-AI industrial sectors relative to their overseas comparable companies is expected to be partially repaired. Third, the extreme divergence between tech and non-tech sectors tends to converge. Within the tech sector, it is advised to use the rebound in AI price-sensitive products to promptly adjust positions towards core assets, placing more emphasis on "volume certainty" and being cautious about "price explosiveness." For the non-tech sector, focus on increasing allocation to energy/chemicals, non-ferrous metals, innovative drugs, and leading brokerages with overseas expansion potential.
Risk factors include intensified friction in China-US technology, trade, and finance; domestic policy strength, implementation effects, or economic recovery falling short of expectations; unforeseen tightening of domestic and international macro liquidity; further escalation of regional conflicts in Russia-Ukraine and the Middle East; and slower-than-expected inventory digestion in China's real estate sector.
Comments