Investment Rating - The report maintains a "Buy" rating for the company [1][7] Core Views - The company has shown continuous high growth in performance, with a focus on expanding its computing power services and robotics sector [1][6] - In 2024, the company achieved a revenue of 7.41 billion yuan, representing a year-on-year growth of 59.08%, and a net profit of 692 million yuan, up 140.80% year-on-year [4][8] - The company is actively expanding its computing power services and has obtained NVIDIA Cloud Partner certification, enhancing its capabilities in AI model services and cloud computing [6] Financial Performance - The company reported a significant increase in revenue across various business segments, with intelligent IoT business revenue reaching 2.26 billion yuan, up 60.82% year-on-year [5] - The overall gross margin improved by 3.81 percentage points to 17.36% in 2024, driven by high-margin businesses [5][10] - The company forecasts revenues of 10.95 billion yuan, 14.52 billion yuan, and 18.55 billion yuan for 2025, 2026, and 2027 respectively, with net profits projected at 1.09 billion yuan, 1.55 billion yuan, and 2.06 billion yuan [7][8] Business Expansion - The company is strategically focusing on the robotics sector, integrating its subsidiary to enhance its capabilities in developing products like robotic dogs and humanoid robots [6] - Collaborations with major enterprises in cloud computing and AI applications are underway, indicating a robust growth trajectory in these areas [6][8]

Quantitative Model and Construction - **Model Name**: Transportation Index Tracking Model - **Model Construction Idea**: The model assumes that the price movement of the target has strong local continuity, always following a certain trend. Reversal periods are shorter than trend continuation periods. In cases of narrow-range consolidation, the model assumes the continuation of the previous trend. When a major trend exists, given a short observation window, the movement will follow the local trend within the window. Reversals are identified when price changes at the start and end of the observation window exceed the range caused by random fluctuations, eliminating the impact of randomness[3] - **Model Construction Process**: 1. Calculate the difference between the closing price on day T and day T-20, denoted as `del` 2. Calculate the volatility (`Vol`) from day T-20 to day T (excluding day T) 3. If the absolute value of `del` exceeds N times `Vol`, the current price is considered to have exited the original oscillation range and formed a trend. The trend direction (long/short) corresponds to the sign of `del`. If the absolute value of `del` is less than or equal to N times `Vol`, the current movement is considered to continue the previous trend direction (same as day T-1) 4. For tracking, N is set to 1, considering the higher volatility and smaller wave opportunities in the stock market compared to the bond market 5. Combine the returns from both long and short directions to evaluate the model's overall performance[3] - **Model Evaluation**: The model is not suitable for direct application to the Transportation Index due to its inability to achieve significant cumulative returns during the tested period. Additionally, the model's drawdown is relatively large compared to its annualized return[4] Model Backtesting Results - **Annualized Return**: 2.27%[3] - **Annualized Volatility**: 16.87%[3] - **Sharpe Ratio**: 0.13[3] - **Maximum Drawdown**: 13.66%[3] - **Total Return of Index During Period**: -9.74%[3]

Quantitative Model and Construction - **Model Name**: Real Estate Index Trend Tracking Model [2] - **Model Construction Idea**: The model assumes that the price movement of the target has strong local continuity and is always in a certain trend. Reversal trends are shorter in duration compared to trend continuations. In cases of narrow-range consolidation, the model assumes the continuation of the previous trend. When in a large-scale trend, given a short observation window, the movement will follow the local trend within the window. If a reversal occurs, the price change at the start and end of the observation window will exceed the range caused by random fluctuations, thus eliminating the impact of randomness. The model also assumes the ability to perform both long and short operations for more rigorous evaluation of relative returns. [3] - **Model Construction Process**: 1. Calculate the difference `del` between the closing price on day T and the closing price on day T-20: $ del = P_T - P_{T-20} $ [3] 2. Calculate the volatility `Vol` during the period from T-20 to T (excluding T): $ Vol = \sqrt{\frac{1}{n-1} \sum_{i=1}^{n} (P_i - \bar{P})^2} $ [3] 3. If the absolute value of `del` exceeds N times `Vol`, it is considered that the current price has broken out of the original oscillation range and formed a trend. The trend direction (long or short) corresponds to the sign of `del`. Otherwise, the trend direction remains the same as on day T-1. [3] 4. For tracking, the parameter N is set to 1, considering the higher volatility and more frequent small-wave opportunities in the stock market compared to the bond market. [3] 5. The combined results of both long and short returns are used as the final evaluation basis. [3] - **Model Evaluation**: The model effectively identifies trends in the real estate index. It performs well even when the index's total return is negative, achieving high annualized returns without significant long-term drawdowns. [4] Model Backtest Results - **Annualized Return**: 26.03% [3] - **Annualized Volatility**: 30.70% [3] - **Sharpe Ratio**: 0.85 [3] - **Maximum Drawdown**: 18.18% [3] - **Total Return of the Index During the Period**: -27.38% [3]

- The report introduces an "Industry Crowdedness Monitoring Model" to track the crowdedness levels of Shenwan primary industry indices on a daily basis. The model identifies industries with high and low crowdedness levels, such as agriculture, banking, and environmental protection being highly crowded, while automotive and media are less crowded. The model also monitors significant daily changes in crowdedness levels, highlighting industries like automotive and home appliances with notable variations[6][11] - A "Premium Rate Z-score Model" is constructed to screen ETF products for potential arbitrage opportunities. The model uses rolling calculations to identify ETFs with significant deviations in premium rates, which may indicate arbitrage potential or risks of price corrections[6][14]

2025 年 04 月 08 日 行业策略 看好/维持 农林牧渔 农林牧渔 沪深300 ◼ 子行业评级 | 种植业 | 看好 | | --- | --- | | 畜牧业 | 看好 | | 林业 | 中性 | | 渔业 | 中性 | | 农 产 品 加 工 | 看好 | | Ⅱ | | 相关研究报告 <<农业周报（第 14 期）：贸易加征关 税利于农产品进口减少和价格上 涨>>--2025-04-06 <<益生股份年报点评：祖代白鸡行业 龙头，种鸡种猪业务盈利稳健>>-- 2025-04-01 <<优然牧业（09858）2024 年年报点 评：科技赋能+精益管理双轮驱动，经 营业绩逆势改善>>--2025-04-01 农林牧渔 行业点评：加征关税影响下，重视农业板块的防御和反制属性 ◼ 走势比较 (30%) (18%) (6%) 6% 18% 30% 24/4/8 24/6/19 24/8/30 24/11/10 25/1/21 25/4/3 证券分析师：程晓东 电话：010-88321761 E-MAIL：chengxd@tpyzq.com 分析师登记编号：S1190511050002 事件：美国总统特朗普 ...

Quantitative Model and Construction - **Model Name**: Utility Index Trend Tracking Model [3] - **Model Construction Idea**: The model assumes that the price movement of the target has strong local continuity, always following a certain trend. Reversal periods are significantly shorter than trend continuation periods. In cases of narrow-range consolidation, the model assumes the continuation of the previous trend. When observing a large-scale trend, a short observation window is used to capture the local trend. Reversals are identified when price changes at the start and end of the observation window exceed the range caused by random fluctuations, eliminating the impact of random noise. [4] - **Model Construction Process**: 1. Calculate the difference between the closing price on day T and day T-20, denoted as `del`. 2. Calculate the volatility (`Vol`) from day T-20 to day T (excluding day T). 3. If the absolute value of `del` exceeds `N` times `Vol`, the current price is considered to have exited the original oscillation range and formed a trend. The trend direction (long/short) corresponds to the sign of `del`. 4. If the absolute value of `del` is less than or equal to `N` times `Vol`, the current trend direction is assumed to continue, matching the direction on day T-1. 5. For stock markets with higher volatility compared to bond markets, `N` is set to 1 for tracking. 6. Combine the returns from both long and short directions to evaluate the final strategy performance. Formula: $ del = P_{T} - P_{T-20} $ $ Vol = \sqrt{\frac{1}{20} \sum_{i=1}^{20} (P_{T-i} - \bar{P})^2} $ where $ P_{T} $ is the closing price on day T, and $ \bar{P} $ is the average price over the observation window. [4] - **Model Evaluation**: The model is not suitable for direct application to the Utility Index due to its inability to achieve significant cumulative returns and its poor adaptability during periods of continuous market fluctuations, leading to sustained drawdowns. [5] Model Backtesting Results - **Annualized Return**: -16.67% [4] - **Annualized Volatility**: 15.99% [4] - **Sharpe Ratio**: -1.04 [4] - **Maximum Drawdown**: 32.10% [4] - **Total Return of Index During Period**: -0.35% [4]

Group 1 - The report indicates that the overall market liquidity has strengthened, with a net inflow of funds amounting to 46.46 billion yuan, despite a decrease in trading activity, as the total trading volume for the week was 4.54 trillion yuan, down from the previous week [8][9][11] - The net outflow of margin financing reached 193.49 billion yuan, while the trading volume of margin financing accounted for 8.75% of the total A-share trading volume [27][34] - The issuance scale of new equity funds was 4.87 billion yuan for IPOs and 69.67 billion yuan for refinancing, indicating a mixed demand for capital [36][39] Group 2 - The report highlights a significant net withdrawal of funds from the open market, totaling 5019 billion yuan, leading to a decrease in the yield spread between 10-year and 1-year government bonds [11][12] - The yield on 10-year government bonds decreased by 8 basis points, while the yield on 1-year government bonds fell by 5 basis points, resulting in a narrowing of the yield curve [11][12] - The market anticipates a 51% probability that the Federal Reserve will not cut interest rates in May, reflecting a shift in market expectations [19] Group 3 - The report notes a decline in turnover rates across major indices, with a corresponding decrease in trading volumes, indicating reduced trading activity among institutional investors [20][22] - The top five sectors where equity funds increased their positions included pharmaceuticals, banking, food and beverage, agriculture, and public utilities, while the sectors with the largest reductions included power equipment, household appliances, electronics, automotive, and non-ferrous metals [23][24] - The total number of ETF shares increased by 269.7 billion, with the broad-based index A500 ETF receiving the most inflow of funds [28][31]

2025 年 04 月 07 日 行业周报 看好/维持 医药 医药 板块持续跑赢大盘，关注对等关税下医药供应链影响（附 OSA 专题研究）(2025.03.31-2025.04.06) 走势比较 OSA 存在未满足需求，GLP-1RA 药物对于 OSA 治疗具有显著效果。阻 塞性睡眠呼吸暂停（obstructive sleep apnea，OSA）是指由于气道变窄 或阻塞而导致睡眠期间的呼吸中断，与高血压的发生发展密切相关，可合 并多种代谢异常、内分泌激素紊乱等，导致心脑血管事件发生风险明显增 加。根据流行病学研究，目前中国成年人中 OSA 患者有 1.76 亿，在糖尿 病及肥胖患者中更为常见，住院 T2DM 患者 OSA 的患病率在 60%以上，肥 胖的 T2DM 患者 OSA 的患病率可达 86%，90%以上 BMI 超过 40 kg/m2 的肥 胖症患者合并 OSA。减重是 OSA 重要的治疗方式，GLP-1RA 药物对于 OSA 治疗有显著效果，礼来的替尔泊肽的 3 期临床 SURMOUNT OSA 研究结果显 示，在肥胖合并中重度 OSA 患者中，与安慰剂相比，替尔泊肽显著降低了 受试者 AHI（呼 ...

2025 年 04 月 07 日 投资策略 估值与盈利周观察——4 月第 1 期：市场分化，红利上成长下 ◼ 上证综合指数走势（近三年） (20%) (12%) (4%) 4% 12% 20% 24/3/18 24/5/29 24/8/9 24/10/20 24/12/31 25/3/13 证券分析师：张冬冬 市场表现：稳定、微盘股、红利表现最好；创业板指、深圳成指、成 长表现最弱。 行业表现：公用事业、医药生物、农林牧渔涨幅前三，计算机、电力 设备、家用电器表现最弱。 相对估值：创业板指/沪深 300 相对 PE 下降，创业板值/沪深 300 相 对 PB 下降。 全 A ERP：ERP 小幅上升，处于一倍标准差附近。 指数 PEG & PB-ROE: 红利 PEG 值最小，科创 50 风格 PB-ROE 值最小。 大类行业估值：原材料、设备制造、工业服务、交通运输、消费、科 技历史估值皆处于 50%以下，金融地产估值高于 50%历史分位。 一级行业估值：各行业估值分化，非银金融、煤炭、公用事业、交通 运输、农林牧渔估值处于近一年低位。 热门概念：科技板块热度较高，东数西算、华为鸿蒙、机器人、中字 头央企等 ...

金 金融工程点评 [Table_Message]2025-04-07 医药生物指数趋势跟踪模型效果点评 [Table_Author] 证券分析师：刘晓锋 电话：13401163428 E-MAIL：liuxf@tpyzq.com 执业资格证书编码：S1190522090001 研究助理：孙弋轩 电话：18910596766 E-MAIL：sunyixuan@tpyzq.com 一般证券业务登记编码：S1190123080008 模型概述 结果评估： 区间年化收益：6.44% 波动率（年化）：24.71% 夏普率：0.26 最大回撤：22.65% 指数期间总回报率：-19.00% 融 工 程 点 评 ◼ 设计原理：模型假定标的价格走势具有很好的局部延续性，标的价格永远处 于某一趋势中，出现反转行情的持续时间明显小于趋势延续的时间，若出现 窄幅盘整的情况，亦假设其延续之前的趋势。当处于大级别的趋势之中时， 给定较短时间的观察窗口，走势将延续观察窗口内的局部趋势。而当趋势发 生反转时，在观察窗口始末位置的价格变动方向会明显超出随机波动造成的 趋势背离范围，从而排除随机波动的影响。虽然指数本身在实际中进行双向 操作有 ...