Model Overview - The strategy is based on a deviation correction model with a callback-type stop-loss mechanism, assuming a cyclical deviation-regression pattern in the price movements of the target relative to a benchmark [4][5] - The model uses historical drawdown data to determine buy thresholds and employs a tiered stop-loss system once a buy signal is triggered [4][5] Performance Evaluation - Total return of the strategy during the backtest period (January 4, 2010 - March 18, 2025) was 108.52% [4] - The buy-and-hold return for the target was 152.35%, resulting in an excess return of -43.84% [4] - The maximum drawdown recorded was 51.27%, with the longest drawdown period lasting 2261 trading days [4] Strategy Applicability - The strategy was tested with a parameter of X=7 and is noted to be less effective in the computer industry, which is characterized by long-term volatility and occasional explosive growth due to technological iterations [5]

Group 1 - The report emphasizes that the adjustment in the market presents an opportunity to invest in Chinese assets, highlighting the emergence of significant bottom points in the Chinese stock market after recent adjustments [11][12][14] - The report identifies that China's energy consumption has a low oil and gas proportion of less than 30%, which is below the global average, enhancing resilience against risks [11][12] - The report notes that China's relatively stable security situation, complete supply chain system, and proactive industrial development are unique advantages that can counteract the prevailing narrative of stagflation [11][12] Group 2 - The report suggests that the focus on domestic demand and expansionary fiscal policies in 2026 will support consumption and stabilize investment, which is expected to counterbalance the decline in global demand [12][13] - The report highlights the acceleration of capital expenditure in new economic sectors and the growth of global energy transition demands as key drivers for China's growth logic in 2026 [13][14] - The report recommends sectors such as finance, technology manufacturing, and stable domestic demand as primary investment targets, emphasizing the value of high dividend yield in financial and stable sectors [14] Group 3 - The report discusses the performance of Tencent Holdings, which is expected to see solid growth driven by AI investments, with projected revenues of 830.2 billion CNY in 2026 [21] - The report highlights the launch of Claude Cowork, which is anticipated to accelerate CPU demand due to its role in AI applications, suggesting a significant growth opportunity in the electronic sector [24][29] - The report mentions that the communication sector, particularly optical interconnection, is expected to experience high growth due to increasing demand in AI infrastructure [36][39]

Investment Rating - The report maintains a "Buy" rating for Kingsoft Office (688111) with a target price of 351.36 CNY [6][12]. Core Insights - The company demonstrates steady revenue growth, with a 15.78% year-on-year increase in revenue for 2025, reaching 5.93 billion CNY, and a net profit of 1.84 billion CNY, up 11.63% year-on-year [12][12]. - The daily average token usage has increased by over 12 times year-on-year, indicating strong growth in AI-enabled services [2][12]. - The WPS 365 enterprise business revenue surged by 64.93% year-on-year, reflecting robust demand in vertical industries [12][12]. - The overseas strategy is accelerating, with personal business revenue from overseas reaching 282 million CNY, a 53.67% increase year-on-year [12][12]. Financial Summary - Total revenue projections for 2026, 2027, and 2028 are 7.02 billion CNY, 8.33 billion CNY, and 9.77 billion CNY, respectively, with year-on-year growth rates of 18.4%, 18.7%, and 17.3% [4][12]. - Net profit forecasts for the same years are 2.17 billion CNY, 2.67 billion CNY, and 3.16 billion CNY, with growth rates of 18.2%, 22.8%, and 18.5% [4][12]. - The company’s R&D investment reached 2.09 billion CNY in 2025, a 23.57% increase year-on-year, representing 35.34% of total revenue [12][12]. Market Data - The stock has a market capitalization of 108.23 billion CNY, with a 52-week price range of 233.56 to 379.99 CNY [7][12]. - The price-to-earnings ratio (P/E) is projected to decrease from 65.79 in 2024 to 34.26 in 2028, indicating improving valuation over time [4][12].

Investment Rating - The report maintains a "Buy" rating for the companies mentioned, including 罗博特科 (Robotech), 唯科科技 (Weike Technology), 能科科技 (Nengke Technology), and 合合信息 (Hehe Information) [8][57]. Core Insights - The report highlights the transformative impact of Google's TurboQuant algorithm, which allows for at least a 6x lossless compression of large model key-value caches and up to an 8x speed increase on H100 GPUs, revolutionizing vector retrieval efficiency [24][27]. - Coherent showcased a series of groundbreaking optical innovations at OFC 2026, aimed at redefining performance, scalability, and energy efficiency standards in AI-era data centers and communication networks [33][35]. - Granola, an AI note-taking application, completed a $125 million Series C funding round, achieving a valuation of $1.5 billion, a 7x increase from the previous year, with plans to expand its team and product offerings [42][44]. Summary by Sections 1. Computing Power Dynamics - The report notes stable pricing in computing power leasing, with Google's TurboQuant algorithm marking a significant industry change [10][24]. - Token consumption data indicates a weekly increase, with a total of 22.7 trillion tokens consumed, reflecting a 11.27% week-over-week growth [13]. - Major cloud service providers, including AWS and Google Cloud, have initiated a new wave of price increases for cloud services, driven by rising demand and supply chain pressures [18][20]. 2. AI Application Dynamics - Gemini's weekly visit volume increased by 4.00%, indicating growing user engagement [32]. - Coherent's innovations at OFC 2026 include advancements in optical technologies, such as 400G channel rates and 12.8T emerging architectures, aimed at enhancing data center capabilities [33][35]. 3. AI Financing Trends - Granola's recent funding round highlights the growing investment interest in AI applications, with significant backing from Index Ventures and others [42][44]. 4. Investment Recommendations - The report emphasizes the importance of contracts signed by 罗博特科 (Robotech) and its subsidiaries, amounting to 600 million RMB, which represents over 54.23% of the company's audited revenue for 2024 [54]. - The report suggests monitoring companies like 唯科科技 (Weike Technology), 合合信息 (Hehe Information), 能科科技 (Nengke Technology), and 罗博特科 (Robotech) for their growth potential in the AI and semiconductor sectors [55].

Investment Rating - The report maintains an "Accumulate" rating for the company, with a target price of 32.06 CNY, compared to the current price of 25.53 CNY [6]. Core Insights - The company reported a revenue of 5.783 billion CNY for 2025, a decrease of 12.1% year-on-year, while the net profit attributable to shareholders increased by 18.0% to 1.231 billion CNY. The company focused on core product lines and optimized its business structure, leading to improved operational quality despite a decline in overall revenue [2][13]. - The company’s gross profit margin was 71.06%, down by 0.96 percentage points year-on-year, and the operating cash flow reached 1.067 billion CNY, up by 22.91% [13]. Financial Summary - **Revenue and Profit Forecasts**: The company is projected to achieve net profits of 1.412 billion CNY, 1.628 billion CNY, and 1.984 billion CNY for 2026, 2027, and 2028, respectively, with corresponding EPS estimates of 0.75 CNY, 0.86 CNY, and 1.05 CNY [4][13]. - **Revenue Breakdown**: In 2025, the wealth IT revenue decreased by 7.7%, asset management IT revenue decreased by 19.8%, and operations and institutional IT revenue decreased by 1.8% [13]. - **R&D Investment**: The company maintained a high level of R&D investment, totaling 2.180 billion CNY, which accounted for 37.7% of its revenue, with 6,953 technical personnel making up 67.66% of the total workforce [13]. Market Data - The company has a total market capitalization of 48.345 billion CNY, with a total share capital of 1,894 million shares [7]. - The stock price has fluctuated between 24.25 CNY and 39.04 CNY over the past 52 weeks [7]. Valuation Comparison - The company’s PE ratio is 39.27 for 2025, compared to an average of 51 for comparable companies [15].

Market Overview - As of March 31, the Shanghai Composite Index fell by 0.8%, the CSI 300 decreased by 0.93%, the STAR 50 dropped by 2.59%, the CSI 1000 declined by 1.91%, and the ChiNext Index decreased by 2.7%. In contrast, the Hang Seng Index rose by 0.15% [4][5] - The best-performing sectors on March 31 were household appliances (+1.57%), banking (+0.72%), and food and beverage (+0.23%). The worst-performing sectors included coal (-3.67%), electric equipment (-3.21%), electronics (-2.71%), basic chemicals (-2.55%), and oil and petrochemicals (-2.27%) [4][5] - The total trading volume of the A-share market on March 31 was 20,059 billion yuan, with a net inflow of 705 million HKD from southbound funds [5] Important Recommendations - The report highlights Shanghai Yanpu (605128) as a key player in the embodied intelligence industry, emphasizing its collaboration with authoritative institutions and accelerated layout in the sector. The recommendation logic is based on the high synergy between automotive and robotics technologies, with the company capable of reusing technology in the robotics field [6] - The expected revenue for Shanghai Yanpu from 2025 to 2027 is projected to be 2,758.05 million yuan, 3,285.33 million yuan, and 4,019.38 million yuan, with growth rates of 21.16%, 19.12%, and 22.34% respectively. The net profit attributable to the parent company is forecasted to be 193.99 million yuan, 280.68 million yuan, and 352.49 million yuan, with growth rates of 41.55%, 44.69%, and 25.59% respectively [6] Important Insights - The quarterly industry strategy report on computers emphasizes the explosive demand for reasoning capabilities, recommending upstream cloud service providers and computing infrastructure leaders, as well as downstream industrial AI solution providers. The report notes that the token explosion has led to tight computing power, driving changes in the industry chain [7] - The report identifies that the core beneficiaries are upstream cloud and infrastructure sectors, while industrial software scenarios are less likely to be overtaken by large models. The synergy between computing and electricity policies is expected to drive long-term growth [7][8]

Investment Rating - The report maintains a "Buy" rating for the companies mentioned, indicating a positive outlook for their performance in the market [8][57]. Core Insights - The report highlights the transformative impact of Google's TurboQuant algorithm, which allows for at least a 6x lossless compression of large model key-value caches and up to an 8x speed increase on H100 GPUs, revolutionizing the industry [24][27]. - Coherent showcased a series of groundbreaking optical innovations at OFC 2026, aimed at redefining performance, scalability, and energy efficiency standards in AI-era data centers and communication networks [33][35]. - Granola, an AI note-taking application, completed a $125 million Series C funding round, achieving a valuation of $1.5 billion, a 7x increase from the previous year, with plans to expand its team and product offerings [4][42]. Summary by Sections Computing Power Dynamics - The report notes stable pricing in computing power leasing, with Google's TurboQuant algorithm marking a significant industry change [10][24]. - Token consumption data indicates a weekly increase, with a total of 22.7 trillion tokens consumed, reflecting a growing demand for AI applications [13][15]. - Major cloud service providers, including AWS and Google Cloud, have initiated a price increase for their services, driven by rising demand and supply chain pressures [18][20]. AI Application Dynamics - Gemini's weekly visit volume increased by 4.00%, indicating growing user engagement with AI applications [32]. - Coherent's innovations at OFC 2026 include advancements in optical technologies, such as 400G channel rates and 12.8T emerging architectures, aimed at enhancing data center capabilities [33][35]. AI Financing Trends - Granola's recent funding round highlights the increasing investment in AI applications, with a total of $192 million raised to date [4][42]. Investment Recommendations - The report emphasizes the significance of a recent $600 million contract signed by ficonTEC, which represents a major breakthrough in the optical device business and reflects the high demand in the silicon photonics and high-speed optical module sectors [5][54]. - Companies such as 罗博特科 (Robotech), 唯科科技 (Weike Technology), 能科科技 (Nengke Technology), and 合合信息 (Hehe Information) are recommended for their strong growth potential in their respective fields [6][55].

Core Insights - The report highlights the April calendar effect, indicating that the large-cap style tends to outperform in most years, while sectors such as beauty, food and beverage, home appliances, and banking are expected to perform relatively well [2][7] - The average absolute monthly return for the Tonghuashun All A (weighted) index in April over the past 10 years is -1.6%, suggesting a general decline in the market during this month [7][8] - Small-cap and micro-cap styles have significantly underperformed compared to large-cap styles, indicating a structural characteristic in the market [7][10] Industry Performance - The sectors that are expected to outperform in April include beauty, food and beverage, home appliances, banking, and pharmaceuticals, while sectors such as computer, comprehensive, light industry, military, and textile are anticipated to lag [7][13] - The report provides a detailed analysis of the average monthly excess returns of various industry indices compared to the Tonghuashun All A (weighted) index over the past 10 years, showing that certain sectors consistently yield better returns [13][15] - Specific data points indicate that the beauty sector has an average excess return of 2.8%, while the computer sector shows a negative average excess return of -0.7% in April [13][15]

Quantitative Models and Construction GAN_GRU Model - **Model Name**: GAN_GRU - **Construction Idea**: The GAN_GRU model combines Generative Adversarial Networks (GAN) for feature generation and Gated Recurrent Units (GRU) for time-series feature encoding to create a stock selection factor[4][13][22] - **Construction Process**: 1. **GAN Component**: - **Generator**: Generates realistic data samples from random noise using the loss function: $$L_{G}\,=\,-\mathbb{E}_{z\sim P_{z}(z)}[\log(D(G(z)))]$$ where \(z\) represents random noise, \(G(z)\) is the generated data, and \(D(G(z))\) is the discriminator's output probability that the generated data is real[24][25][26] - **Discriminator**: Distinguishes real data from generated data using the loss function: $$L_{D}=-\mathbb{E}_{x\sim P_{data}(x)}[\log\!D(x)]-\mathbb{E}_{z\sim P_{z}(z)}[\log(1-D(G(z)))]$$ where \(x\) is real data, \(D(x)\) is the discriminator's output probability for real data, and \(D(G(z))\) is the output probability for generated data[27][29][30] - **Training Process**: Alternating training of the generator and discriminator until convergence[30][34] 2. **GRU Component**: - Two GRU layers (GRU(128,128)) followed by an MLP (256,64,64) to encode time-series features and predict future returns[22] - Input features include 18 price-volume metrics (e.g., closing price, turnover rate) sampled over 40 days to predict cumulative returns for the next 20 trading days[14][18][19] - Data preprocessing involves outlier removal, normalization, and cross-sectional standardization[18] - Training uses semi-annual rolling windows with hyperparameters such as batch size equal to the number of stocks, Adam optimizer, learning rate of \(1e-4\), and IC-based loss function[18][22] 3. **Feature Generation**: - GAN's generator processes raw price-volume time-series features (Input_Shape=(40,18)) and outputs transformed features with preserved time-series properties[37] - **Evaluation**: The model effectively combines GAN's feature generation capabilities with GRU's time-series encoding, providing robust predictive power for stock selection[4][22][37] --- Model Backtesting Results GAN_GRU Model Performance Metrics - **IC Mean**: 0.1095*** - **ICIR (Non-Annualized)**: 0.88 - **Turnover Rate**: 0.82X - **Recent IC**: 0.1008*** - **One-Year IC Mean**: 0.0514*** - **Annualized Return**: 36.03% - **Annualized Volatility**: 21.87% - **IR**: 1.55 - **Max Drawdown**: 27.29% - **Annualized Excess Return**: 21.87%[41][42][45] Industry-Level Performance - **Top 5 Industries by Recent IC**: - Media: 0.4279*** - Coal: 0.2355*** - Retail: 0.2003*** - Food & Beverage: 0.1701*** - Chemicals: 0.1395***[41][42][45] - **Top 5 Industries by One-Year IC Mean**: - Media: 0.1304*** - Steel: 0.1212*** - Retail: 0.1191*** - IT: 0.1064*** - Food & Beverage: 0.0988***[41][42][45] - **Top 5 Industries by Recent Excess Return**: - Media: 4.57% - Agriculture: 3.26% - Construction Materials: 3.19% - Light Manufacturing: 2.53% - Coal: 2.22%[45][46][48] - **Top 5 Industries by One-Year Average Excess Return**: - Real Estate: 1.83% - Retail: 1.41% - Consumer Services: 1.39% - Automotive: 1.18% - Utilities: 1.07%[45][46][48] --- Quantitative Factors and Construction GAN_GRU Factor - **Factor Name**: GAN_GRU - **Construction Idea**: Derived from the GAN_GRU model, this factor leverages GAN for feature generation and GRU for time-series encoding to predict stock returns[4][13][22] - **Construction Process**: - Input features include 18 price-volume metrics sampled over 40 days[14][18][19] - GAN generates transformed features while preserving time-series properties[37] - GRU encodes these features and outputs predicted returns as the factor[22][37] - Factor values undergo industry and market-cap neutralization and standardization[22] - **Evaluation**: The factor demonstrates strong predictive power across multiple industries and time periods, with significant IC values and excess returns[4][22][37] --- Factor Backtesting Results GAN_GRU Factor Performance Metrics - **IC Mean**: 0.1095*** - **ICIR (Non-Annualized)**: 0.88 - **Turnover Rate**: 0.82X - **Recent IC**: 0.1008*** - **One-Year IC Mean**: 0.0514*** - **Annualized Return**: 36.03% - **Annualized Volatility**: 21.87% - **IR**: 1.55 - **Max Drawdown**: 27.29% - **Annualized Excess Return**: 21.87%[41][42][45] Industry-Level Performance - **Top 5 Industries by Recent IC**: - Media: 0.4279*** - Coal: 0.2355*** - Retail: 0.2003*** - Food & Beverage: 0.1701*** - Chemicals: 0.1395***[41][42][45] - **Top 5 Industries by One-Year IC Mean**: - Media: 0.1304*** - Steel: 0.1212*** - Retail: 0.1191*** - IT: 0.1064*** - Food & Beverage: 0.0988***[41][42][45] - **Top 5 Industries by Recent Excess Return**: - Media: 4.57% - Agriculture: 3.26% - Construction Materials: 3.19% - Light Manufacturing: 2.53% - Coal: 2.22%[45][46][48] - **Top 5 Industries by One-Year Average Excess Return**: - Real Estate: 1.83% - Retail: 1.41% - Consumer Services: 1.39% - Automotive: 1.18% - Utilities: 1.07%[45][46][48]

Market Overview - The domestic market enters the earnings season in April, with ongoing overseas disturbances potentially balancing market styles[3] - Key focus areas include Middle Eastern geopolitical disturbances affecting oil prices and fluctuating inflation expectations[3] Investment Strategy - The strategy emphasizes three main lines: 1. Energy security, focusing on traditional energy price increases and new energy directions due to potential replenishment demand[3] 2. Technology, particularly AI infrastructure, including power, storage, and computing sectors[3] 3. Rebound of previously oversold sectors such as precious metals and commercial aerospace[3] Recommended Stocks - Key recommended sectors and stocks include: - Metals: Zijin Mining - Chemicals: Yara International - Petrochemicals: Shouhua Gas - Power: Longyuan Power H - Coal: Yancoal Energy - New Energy: Jiayuan Technology - Banking: Hangzhou Bank - Agriculture: Dekang Agriculture - Electronics: Zhaoyi Innovation - Communication: Zhongji Xuchuang[6] Risk Factors - Economic recovery may fall short of expectations, with potential slow job growth and reduced market demand[34] - Significant changes in individual stock fundamentals could impact performance[34] Earnings Forecasts - Forecasted earnings per share (EPS) and price-to-earnings (PE) ratios for key stocks: - Zijin Mining: EPS of 3.10 in 2026, PE of 10.5[28] - Yara International: EPS of 4.24 in 2026, PE of 15.2[28] - Shouhua Gas: EPS of 1.42 in 2026, PE of 16.7[28] - Longyuan Power H: EPS of 0.72 in 2026, PE of 9.5[28] - Yancoal Energy: EPS of 1.23 in 2026, PE of 16.5[28] - Jiayuan Technology: EPS of 1.90 in 2026, PE of 21.9[28] - Hangzhou Bank: EPS of 2.84 in 2026, PE of 5.8[28] - Dekang Agriculture: EPS of 2.89 in 2026, PE of 20.3[28] - Zhaoyi Innovation: EPS of 8.62 in 2026, PE of 30.0[28] - Zhongji Xuchuang: EPS of 17.40 in 2026, PE of 34.4[28]