From June to August 2026, trading activity on the STAR Market heated up in phases, while trading interest in the dividend index remained relatively subdued.

The STAR 50 Index, centered on hard technology such as semiconductors and communications, saw both prices and turnover rise in tandem from June to August 2026; the CSI Dividend Index, centered on high-dividend sectors such as banks, coal, and transportation, saw turnover contract and prices weaken over the same period. The divergence in turnover between the two shows that incremental funds in this round did not flow evenly across the entire market, but instead concentrated heavily on a handful of growth directions such as semiconductors and communications.

Looking at quantitative indicators, market trading concentration reached a historic high in July 2026. The share of turnover accounted for by the top 5% of stocks by turnover in the whole market is used to measure trading concentration. The higher this ratio, the more trading is concentrated in a few leading stocks. Since 2003, the indicator has fluctuated around 35% most of the time, and days above 45% have not been many; yet from June to August 2026 it stayed above 46% continuously, and briefly broke through 50% in late July. During the June to August 2026 phase, the stocks being heavily traded were mainly concentrated in the AI hardware sector, contributing significantly to the index weights of the STAR Market and ChiNext. Based on individual stock times trading day statistics, the electronics industry contributed about 36% of high-turnover records, clearly higher than other industries; non-ferrous metals, communications, basic chemicals, machinery, and computers followed in order. In the STAR 50, 28 constituent stocks were among the heavily traded names, with a combined weight of about 85%; in the ChiNext Growth Index, 41 stocks carried a weight of about 96%; the ChiNext Index and CSI 300 also had weight shares of about 88% and 77%, respectively. By comparison, the weight of heavily traded stocks in the CSI 1000 and CSI Dividend was only about 31% and 21%.

After the phase of highly concentrated trading, small-cap style may have a relative advantage. After the highly concentrated trading phase, the portfolios of heavily traded stocks continued to underperform the CSI 1000 Index; the relative advantage of the dividend index lacked persistence. Using trading days in the highly concentrated phase as samples, the excess returns of heavily traded stock portfolios relative to the CSI 1000 and CSI Dividend after the concentrated trading ebbed were calculated. It was found that the win rate of heavily traded stock portfolios relative to the CSI 1000 Index was low during this period, and the longer the holding period, the lower the win rate; although the win rate relative to the CSI Dividend Index was also low, it began to improve as the holding period lengthened. From the perspective of index return performance, after the highly concentrated trading phase, the win rate of CSI 1000 relative to CSI 300 exceeded 90%. Using trading days with concentration above 46% as samples, the excess win rate of CSI 1000 relative to CSI 300 rose from about 64% in a 10-day window to about 96% in a 100-day window, with windows above 60 days all near or above 90%. Testing showed that the excess return came from both market-cap style contribution and industry deviation contribution. In the short term (within 60 days), the win rate from market-cap style contribution was higher; in the medium to long term (above 60 days), the win rate from industry deviation contribution was greater. Risk: The conclusions of this article are based on backtesting statistics of historical data, and historical patterns may fail; the threshold setting of the trading concentration indicator is somewhat subjective, and data and models may also contain errors.

June-August 2026: Market Trading Concentration Rises in Phases

From June to August 2026, trading activity in the STAR Market sector was relatively high, while interest in the dividend index remained persistently weak. Since June 2026, a clear phenomenon has appeared in the A-share market: funds concentrated on a few high-prosperity sectors, and trading concentration rose markedly. On one hand, technology growth directions such as semiconductors and communications equipment continued to attract incremental funds, with turnover expanding and trading active; on the other hand, defensive high-dividend sectors such as dividend stocks cooled in phases, with turnover contracting and indices pulling back. Market funds were clearly biased toward a few directions in the short term, while other sectors were relatively quiet.

This phenomenon is even clearer at the index level. The STAR 50 Index, centered on hard technology such as semiconductors and communications, saw both prices and turnover rise in tandem from June to August 2026 (Chart 1); the CSI Dividend Index, centered on high-dividend sectors such as banks, coal, and transportation, saw turnover contract and prices weaken over the same period (Chart 2). The divergence in turnover between the two shows that incremental funds in this round did not flow evenly across the entire market, but instead concentrated heavily on a handful of growth directions such as semiconductors and communications. Chart 1: STAR 50 Index Price and Turnover Trend. Note: As of August 31, 2026. Source: Wind, CICC Research Department. Chart 2: CSI Dividend Index Price and Turnover Trend. Note: As of August 31, 2026. Source: Wind, CICC Research Department.

This round of rising trading concentration is rare in history. In order to objectively measure this phenomenon and study its subsequent impact, a trading concentration indicator was constructed, which is the basis for the analysis below.

Highly Concentrated Trading Phases Often Correspond to Market Phase Highs or Lows

The share of turnover accounted for by the top 5% of stocks by turnover in the whole market is used to measure trading concentration, that is, each day the top 5% of stocks by turnover are selected to see how much of the whole market's turnover they collectively contribute. The higher this ratio, the more trading is concentrated in a few leading stocks. Chart 3: Whole-Market Trading Concentration Indicator and CSI 300 Index Trend. Note: Trading concentration indicator = sum of turnover of the top 5% of companies by whole-market turnover / total whole-market turnover; CSI 300 Index is on the right axis; as of September 18, 2026. Source: Wind, CICC Research Department.

Historically, this indicator has been at normal levels most of the time, but from June to August 2026 it rose to a historically rare high position. Since 2003, the indicator has fluctuated around 35% most of the time, and days reaching above 45% have not been many; yet from June to August 2026 it continuously stood above 46%, and briefly exceeded 50% in late July. Therefore, 46% is used as the dividing line between a highly concentrated trading phase and normal conditions: since 2003, the proportion of trading days with the indicator above 46% was only about 3.5%, which is rare; and June to August 2026 happened to run continuously above 46%, matching the aforementioned concentration of funds in technology sectors in timing.

Putting the CSI 300 and the trading concentration indicator together shows that highly concentrated trading phases mostly correspond to phase highs and lows of the CSI 300 Index: for example, at the end of 2007, July 2015, February 2018, and February 2021, the trading concentration indicator all reached above 46%, corresponding to phase highs in the CSI 300; while in November 2008, trading concentration also reached above 46%, corresponding to a phase low in the CSI 300.

During June-August 2026, Heavily Traded Stocks Were Concentrated in the AI Hardware Sector

By industry, the heavily traded stocks in this round were highly concentrated in the electronics supply chain. Based on individual stock times trading day statistics, the electronics industry contributed about 36% of high-turnover records, clearly higher than other industries; non-ferrous metals, communications, basic chemicals, machinery, and computers followed in order. Chart 4: Primary Industry Distribution of Heavily Traded Stocks During June to August 2026. Note: Statistical interval is June to August 2026 (trading days with concentration above 46%), counted by individual stock times trading day. Source: Wind, CICC Research Department.

Heavily traded stocks contributed significantly to the index weights of the STAR Market and ChiNext. Sectors previously chased intensively by funds accounted for very high weights in major indices. As of August 31, 2026, 28 of the 50 constituent stocks in the STAR 50 were heavily traded stocks, with a combined weight of about 85%; 41 of 50 in the ChiNext Growth Index, with a weight of about 96%; the ChiNext Index and CSI 300 also had weight shares of about 88% and 77%, respectively. In contrast, the weight of heavily traded stocks in the CSI 1000 and CSI Dividend was only about 31% and 21%. In other words, if heavily traded stocks weaken later, previously popular indices such as the STAR 50 and ChiNext will be dragged down relatively more. Chart 5: Weight Share of Heavily Traded Stocks in Major Indices During June to August 2026. Note: As of August 31, 2026. Source: Wind, CICC Research Department.

After Highly Concentrated Trading Phases, Small-Cap Style May Have a Relative Advantage

After highly concentrated trading phases, heavily traded stocks often underperform the CSI 1000. After the highly concentrated trading phase, attention is paid to whether heavily traded stocks have obvious relative advantages or disadvantages versus other indices. To this end, using trading days with concentration above 46% as samples, the excess return performance of heavily traded stock portfolios (stocks ranked in the top 5% by turnover during the highly concentrated trading phase) relative to indices such as CSI 500, CSI 1000, and CSI Dividend over different windows after the highly concentrated phase was calculated.

After the highly concentrated trading phase, the return performance of heavily traded stock portfolios was overall weaker than the CSI 1000. As shown below, the excess win rate of heavily traded stocks relative to CSI 1000 was about 24% after 10 days, about 16% after 20 days, about 4% after 60 days, and about 2% after 100 days; the average excess return also expanded from about -2% after 10 days to about -12% after 60 days and about -24% after 100 days. Both the excess return win rate and average excess return declined as the window lengthened, indicating that after the highly concentrated trading phase, heavily traded stock portfolios continued to underperform the CSI 1000 Index.

The persistent relative weakness was mainly reflected versus the small- and mid-cap style (CSI 1000), while persistence versus the dividend style was not strong. As mentioned earlier, the excess win rate of heavily traded stock portfolios relative to the CSI 1000 Index continued to decline within 100 days after the highly concentrated trading phase; while the excess win rate relative to CSI Dividend rose from 33% at 10 days to 45% at the 120-day window, approaching 50%, indicating that within half a year after concentrated trading ebbed, dividend-style stocks failed to stably and persistently outperform heavily traded stock portfolios. Chart 6: After Highly Concentrated Trading Phases, Excess Win Rate of Heavily Traded Stock Portfolios Relative to CSI 1000, CSI 500, and CSI Dividend. Note: Statistical interval is January 1, 2005 to September 24, 2026, sample is trading days with concentration above 46%; horizontal axis is future return window (days). Source: Wind, CICC Research Department. Chart 7: Average Excess Returns of Heavily Traded Stocks Relative to CSI 1000, CSI 500, and CSI Dividend. Note: Statistical interval is January 1, 2005 to September 24, 2026, sample is trading days with concentration above 46%; horizontal axis is future return window (days). Source: Wind, CICC Research Department.

After Highly Concentrated Trading Phases, CSI 1000 Had a Higher Win Rate Relative to CSI 300

From the perspective of index return performance, using trading days in the highly concentrated trading phase as samples, the excess return win rate of CSI 1000 relative to CSI 300 over different subsequent windows was calculated, and attribution analysis of this excess return was attempted from the dimensions of industry deviation and market-cap style. After the highly concentrated trading phase, CSI 1000 performed better relative to CSI 300, and the longer the holding period, the more obvious the advantage. Using trading days with concentration above 46% as samples, the excess win rate of CSI 1000 relative to CSI 300 rose from about 64% in a 10-day window to about 96% in a 100-day window, with windows above 60 days all near or above 90%; the average excess return also expanded from about 1.4% at 10 days to about 31% at 100 days. Over the same period, CSI 1000 rose about 43% on average in the 100-day window, while CSI 300 rose only about 11%. Chart 8: After Highly Concentrated Trading Phases, Excess Win Rate of CSI 1000 Relative to CSI 300. Note: Statistical interval is January 1, 2005 to September 24, 2026, sample is trading days with concentration above 46%. Source: Wind, CICC Research Department.

The difference between the industry distribution of the CSI 1000 Index and that of the CSI 300 Index was used as weights to calculate a weighted average of industry index returns as the contribution of industry deviation; the difference between the weighted average returns of constituent stocks in each industry within the CSI 1000 Index and the weighted average returns of constituent stocks in each industry within the CSI 300 Index was used as the contribution of market-cap style. Using trading days in the highly concentrated phase as samples, the average values and win rates of industry deviation contribution and market-cap style contribution over different subsequent windows were calculated. When the holding period was within 60 days, the contribution of market-cap style was relatively higher; when the holding period exceeded 60 days, the win rate of industry deviation contribution was greater. As shown below, after the highly concentrated trading phase, the excess return of the CSI 1000 Index relative to CSI 300 had fairly obvious contributions from both industry deviation and market-cap style. However, from the win-rate dimension, differences in their contributions across different windows can be observed. When the holding period was within 60 trading days, the win rate of market-cap style contribution was relatively higher, indicating that in the short term the spillover of funds from leading stocks (high-turnover stocks) mainly flowed to second-tier stocks in mainstream tracks; when the holding period was above 60 trading days, the win rate of industry deviation contribution was relatively higher, indicating that over the medium to long term, spillover funds would also seek investment opportunities in other industries. Chart 9: After Highly Concentrated Trading Phases, Industry Deviation and Market-Cap Contribution to CSI 1000 Excess Return Relative to CSI 300 (Mean). Note: Statistics by mean; statistical interval is October 17, 2014 to September 24, 2026 (the starting date of CSI 1000 constituent stock data is October 17, 2014); sample is trading days with concentration above 46%. Source: Wind, CICC Research Department. Chart 10: After Highly Concentrated Trading Phases, Industry Deviation and Market-Cap Contribution to CSI 1000 Excess Return Relative to CSI 300 (Win Rate). Note: Statistics by mean; statistical interval is October 17, 2014 to September 24, 2026 (the starting date of CSI 1000 constituent stock data is October 17, 2014); sample is trading days with concentration above 46%. Source: Wind, CICC Research Department.

Overall, after a highly concentrated trading phase, small- and mid-cap style is relatively favored, while the persistence of the dividend style advantage is slightly weaker: heavily traded stocks (mainly in growth directions such as electronics and semiconductors) continued to underperform the CSI 1000 Index subsequently, while their excess return relative to the CSI Dividend Index increased as the holding window lengthened. In terms of index performance, the win rate of CSI 1000 relative to CSI 300 after the highly concentrated trading phase was also clearly higher. Therefore, after trading concentration peaks and falls back, small- and mid-cap style indices such as CSI 1000 are worth watching.

Risk Warning

The conclusions of this article are based on backtesting statistics of historical data, and historical patterns may fail; the threshold setting of the trading concentration indicator is somewhat subjective, and data and models may also contain errors; the judgment that small- and mid-cap style will dominate depends on a specific market environment, and if style switching occurs faster than expected, the conclusion may no longer hold.