Morning rebound, afternoon plunge, and a late-session climb back — what exactly unfolded in the A-share market today?
On October 20, the market swung upward in the final stretch of trading, with the Shanghai Composite Index closing in the green, the ChiNext Index narrowing its losses, and the STAR 50 Index gaining 1%.
On the sector front, robotics-related concepts stayed active on and off, edge-side AI climbed rapidly, and the food and beverage sector staged an intraday surge. On the downside, the coal sector saw choppy consolidation while dye-related concepts weakened. The divergence between the yellow and white lines was pronounced, with small and mid-cap stocks showing stronger momentum. More than 3,700 individual stocks across the entire market advanced.
Combined turnover on the Shanghai and Shenzhen exchanges reached 1.46 trillion yuan, up 18.6 billion yuan from the previous trading day.
Approaching the close, several broad-based ETFs saw significant volume spikes, with the SSE 50 ETF (510050) trading over 4 billion yuan within 20 minutes.
Building a bottom is a process of restoring market confidence, and today's price action shows that capital remains skeptical about the strength of any rebound along the technology line. This is, in fact, just human nature during periods that are not part of an uptrend.
According to Wind data, from a capital flow perspective: before 10:48, driven by the continued rise of Hengtong Optic-Electric, which had hit the daily limit the previous day, institutional and main-force capital flowed in rapidly, lifting the dual-innovation indices from a low open to a higher close, and it seemed a hint of the "sustainability" the market had been hoping for was emerging.
But the good times did not last. From that point until the morning close, those two streams of capital began to slow their inflows; when afternoon trading opened, capital across the entire market turned to selling in unison, dragging the indices lower. It was not until after 14:12 that returning or bottom-fishing capital once again propped up the indices. The result amounted to some capital "doing a T" — a round-trip trade.
In this process, heavyweight stocks with closer ties to institutional capital showed even more pronounced spikes and pullbacks. Wind data shows that CATL, Zhongji Innolight, and Eoptolink were the top three contributors to the ChiNext's decline, while Cambricon, Piotech, Skyverse, and Changxin Technology led the drag on STAR Market performance.
In terms of price action, both dual-innovation indices pulled back after touching their 5-day moving averages, signaling overhead pressure, such as trapped positions. On the other side, index gauges with weaker ties to institutional capital and a tilt toward small and micro-cap styles remained in the green.
According to Tonghuashun statistics, at one point more than 4,000 individual stocks were advancing, and for most of the session gainers outnumbered decliners. This is actually a continuation of the market style seen since October — namely, under the premise that "technology is bound to the indices," "the index is the index, and sentiment is sentiment." Of course, when the indices fall substantially, individual stock sentiment is still affected.
But as mentioned at the outset, when the market is falling or oscillating, investors across the board will inevitably prioritize "locking in floating profits" and "getting out at breakeven." Regardless of the method, much of the stock trapped above needs to be genuinely "transported" to lower levels before it can be considered "building energy for a rebound."
The strategy team at Huaxi Securities believes that since July, the time and magnitude of A-share adjustments have been fairly substantial. Affected by overseas factors, this round of A-share adjustment began in early July, and by October 16 had lasted about 15 weeks — roughly 75 trading days. During this period, it went through three step-downs: the technology sector's correction in July, a weak rebound and consolidation in August–September, and another bottom-test in the second week of October. The time span has exceeded the norm for a typical intermediate adjustment (usually 4–8 weeks), making it a complete intermediate adjustment.
The institution stated that multiple ETF products have continued to see volume spikes recently, indicating that some capital recognizes A-shares' value for renewed allocation at this level. Therefore, it judges that a repair-style rally in A-shares will gradually unfold, with structural divergence under the dual main lines of "technology plus dividend."
(Note: In late trading, some STAR 50 ETFs and SSE 50 ETFs saw notable volume spikes, with the most typical products shown in the chart below.)
Specifically on allocation, the institution recommends that the technology track focus on AI hardware and computing power infrastructure, prioritizing names with orders and cash flow that can be realized, while avoiding pure-theme and overvalued varieties; on the AI application side, attention should go to directions where commercialization is accelerating.
Dividend assets should serve as the portfolio's ballast, with preference for sectors offering high dividends and stable cash flow to hedge against overseas interest rates and market volatility.
The institution also cautioned: operationally, investors should avoid chasing highs, use earnings verification as the yardstick, screen stocks from the bottom up, and seize A-shares' structural opportunities in the golden autumn.
Finally, back to the sectors. Thanks to fairly decent individual stock sentiment, multiple concepts posted large gains today, for example:
(1) The robotics concept, which extended its strength. The director of the Ministry of Industry and Information Technology's Science and Technology Department recently said at a State Council Information Office press conference that in the first half of this year, China's humanoid robot complete-machine products surpassed 400 models, with output exceeding 40,000 units, and full-year output is expected to exceed 100,000 units. Using humanoid robots as a small entry point, this will drive the development of the broader embodied intelligence industry. Another view noted that the humanoid robot industry still has plenty of catalysts this week that have drawn market attention: on October 22, Unitree's ecosystem partner day, where a world model may be shown for the first time; on October 22, Tesla holds its 2026 Q3 earnings call, with collected questions including new progress on the V3 robot; and from October 20–22, Nvidia's GTC conference, focused on physical AI.
(2) Edge-side chips. Counterpoint estimates that the penetration rate of generative AI phones in the global market will reach 52% by 2027, with edge-side AI shifting from a premium selling point to a factory-standard feature. Gartner data shows that the share of AI PC shipments will jump from 15.6% in 2024 to about 54.7% in 2026, and AI PCs are expected to become the norm by 2029.
(3) The PCB concept. An electronics industry research report from Shenwan Hongyuan Securities Research Institute argues that AI servers, switches, and optical modules impose higher requirements for high-frequency, high-speed, low-loss, and heat-dissipation performance, driving upgrades in PCB and copper-clad laminate products; some non-AI PCB companies are achieving profit repair by passing costs downstream.
(4) Innovative drugs. On the news front, the 2026 European Society for Medical Oncology (ESMO) Congress will be held from October 23 to 27, and with a window for the concentrated release of innovative drug clinical data approaching, market attention is heating up.
Comments