收评：三大指数回调沪指跌0.45% 银行股逆市走强

Market Overview - On September 2, A-shares experienced a collective pullback, with the ChiNext Index leading the decline. The Shanghai Composite Index closed at 3858.13 points, down 0.45%, with a trading volume of 1.2228 trillion yuan. The Shenzhen Component Index closed at 12553.84 points, down 2.14%, with a trading volume of 1.6522 trillion yuan. The ChiNext Index closed at 2872.22 points, down 2.85%, with a trading volume of 803.8 billion yuan. The total trading volume of the Shanghai and Shenzhen markets reached 2.875 trillion yuan, an increase of 125 billion yuan compared to the previous day [1]. Sector Performance - In terms of sector performance, the banking, precious metals, robotics, and oil sectors saw the largest gains, while sectors such as CPO, cross-border payments, PCB, and semiconductors experienced the most significant declines [1]. - Gold stocks continued their strong performance, with companies like Western Gold hitting the daily limit. The solid-state battery concept also showed strength, with companies like Dexin Technology reaching the daily limit. The robotics sector saw a rebound in the afternoon, with Zhejiang Rongtai and Top Group hitting the daily limit. Conversely, computing hardware stocks collectively retreated, with Cambridge Technology hitting the daily limit down, and the cryptocurrency sector weakened, with China National Petroleum Capital also hitting the daily limit down [2]. Institutional Insights - According to Jifeng Investment Advisory, the market is currently experiencing a volatile pullback, with the precious metals sector performing well. The overall market trend remains upward, but there is increasing divergence. Future allocations should prioritize high-dividend banks and consumer leaders with low valuations, as these assets provide a safety margin in a tightening liquidity environment. Focus should also be on policy-determined sectors such as infrastructure and life services mentioned in the "14th Five-Year Plan," which are directly supported by policy and have relatively reasonable valuations [4]. - Furuong Fund noted that market sentiment remains high, with daily trading volume exceeding 3 trillion yuan. The technology growth sector, represented by AI computing power, continues to lead, while dividend and micro-trading sectors perform poorly. Concerns about the AI computing power sector are centered around trading congestion, but the fundamentals remain strong. The core strategy should focus on identifying quality targets with expected differences and marginal increments within the growth sectors [5]. Fund Holdings - According to the 2025 fund mid-report, the pharmaceutical, banking, and electronics sectors saw significant increases in fund holdings. In contrast, sectors like power equipment, food and beverage, and coal experienced notable declines. As of the report period, electronics, pharmaceuticals, and power equipment remain the three largest sectors by fund holding value. Notably, Shutaishen, Anglikang, and Weichai Heavy Industry saw the largest increases in fund holding ratios, while Guotai Haitong, Shenghong Technology, and Zijin Mining led in fund holding value growth [6]. Software Industry Performance - The Ministry of Industry and Information Technology reported that in the first seven months, China's software business revenue reached 83,246 billion yuan, a year-on-year increase of 12.3%. The total profit of the software industry was 10,890 billion yuan, up 12.4% year-on-year. Software product revenue was 18,011 billion yuan, a 10.6% increase, accounting for 21.6% of the total industry revenue. Information technology service revenue was 57,246 billion yuan, up 13.4%, making up 68.8% of the total industry revenue [7]. Robotics Technology - Yushu Technology announced a patent for an immersive robotic remote operation method that allows operators to feel as if they are executing tasks in real-time. This innovation addresses the limitations of existing remote operation solutions, which often lack effective feedback mechanisms and precision. The new method utilizes various models to provide real-time mapping of the operator's actions and generate immersive 3D images for VR devices, significantly enhancing control capabilities and task execution effectiveness [8].