红利风格择时周报（0224-0227） [Table_Authors] 郑雅斌(分析师) 本报告导读： 上周（20260224 至 20260227）红利风格择时模型综合因子值为-0.10，在连续两周为 正值后转为轻微负向信号，前一周（20260209 至 20260213）因子值为 0.09。 投资要点： | | 021-23219395 | | --- | --- | | | zhengyabin@gtht.com | | 登记编号 | S0880525040105 | | | 梁誉耀(分析师) | | | 021-38038665 | | | liangyuyao@gtht.com | | 登记编号 | S0880524080003 | [Table_Report] 相关报告 低频选股因子周报（2026.02.13-2026.02.27） 2026.03.01 量化择时和拥挤度预警周报（20260227） 2026.02.28 红利风格择时周报（0209-0213） 2026.02.24 量化择时和拥挤度预警周报（20260220） 2026.02.22 高频选股因子周报（20260209-202602 ...

[Table_Page] 金融工程|专题报告 2026 年 3 月 1 日 证券研究报告 [Table_Title] 金融工程：大类资产及权益风格月 报（2026 年 2 月） 权益资金流边际改善，小盘成长风格有望占优 [Table_Summary] 报告摘要： 图：大类资产最新观点(20260228) 表：权益风格最新观点(20260228) | 权益风格 | 宏观视角 最新得分 | 技术视角 最新得分 | 最新 总得分 | 最新观点 | | --- | --- | --- | --- | --- | | 大盘川盘 | -1 | -2 | -3 | 看好小盘 | | 成长/价值 | 1 | 2 | 3 | 看好成长 | [分析师： Table_Author]李豪 SAC 执证号：S0260518070001 021-38003569 lhao@gf.com.cn 分析师： 安宁宁 SAC 执证号：S0260512020003 SFC CE No. BNW179 0755-23948352 anningning@gf.com.cn 请注意，李豪并非香港证券及期货事务监察委员会的注册 持牌人，不可在香港从事受监 ...

Group 1 - OpenAI's internal turmoil has stabilized, allowing SoftBank's Masayoshi Son to reduce personal financial risk related to a $1 billion personal guarantee for investments in OpenAI [3][4] - SoftBank has invested $34.6 billion in OpenAI, acquiring an 11% stake, with the investment recorded in the Vision Fund 2, which previously faced significant losses [4][5] - The Vision Fund 2's value has improved by $19.8 billion due to OpenAI's equity appreciation, reducing its overall loss to approximately 3% [4][5] Group 2 - SoftBank's stock price has doubled over the past year, reflecting market confidence in OpenAI, although the substantial profits from the Vision Fund 2 primarily benefit Masayoshi Son [5][6] - Despite the potential for OpenAI's turmoil to resurface, Son has minimized downside risk as SoftBank has converted its loans to the Vision Fund 2 into preferred shares [6] - The company is heavily reliant on debt to fund its AI investment commitments, indicating a strategic focus on financial engineering [6] Group 3 - Pinterest's stock fell 18% after reporting a slowdown in revenue growth to 14%, attributed to reduced advertising spending from large retailers due to new furniture tariffs [7] - Pinterest's CEO expressed dissatisfaction with the fourth-quarter performance and emphasized the need to restore growth rates to 15%-20% [7] - Other companies like Airbnb and Instacart reported positive earnings growth, with Airbnb's revenue growth accelerating to 12% and Instacart's revenue reaching $992 million, up 12% year-over-year [7]

Investment Rating - The report indicates a positive shift in the dividend style timing model, with a composite factor value of 0.09, signaling a potential market style switch towards dividends [1][6]. Core Insights - The composite factor value increased from -0.08 in the previous week (January 26 to January 30, 2026) to 0.09, suggesting a recovery and a positive signal for dividend style investments [1][6]. - The momentum factor's contribution turned positive, and the downward trend in U.S. Treasury yields has significantly weakened, reducing pressure on dividend stocks [7]. - Market sentiment has cooled, contributing positively to the dividend style, with multiple factors working together to shift the score from negative to positive [7]. Summary by Sections Model Latest Results - The report updates the results of the dividend style timing model for the week of February 2 to February 6, 2026, showing a composite factor value of 0.09, which is a positive change from the previous week's -0.08 [6][7]. Factor Contributions - The report details various factors influencing the dividend style: - Non-manufacturing PMI for services in China: -0.13 - M2 YoY growth in China: 0.21 - U.S. 10-year Treasury yield: -0.24 - Relative net value of dividends: 0.92 - Dividend yield relative to 10-year government bond yield: 0.22 - Net financing purchases: -1.62 - Industry average prosperity: 0.77 [11].

Core Insights - Elon Musk has merged SpaceX with xAI, creating the highest-valued private company in history, with a valuation of $1.25 trillion [2] - The merger combines SpaceX's rocket manufacturing capabilities, the Starlink satellite network, and xAI's data and AI models, aiming for synergistic benefits [2] - Critics argue that the merger is a financial engineering move, relying heavily on Musk's personal brand rather than rational financial logic [2] Merger Details - SpaceX will acquire xAI for $250 billion, aligning with xAI's recent valuation of $230 billion from a $20 billion funding round [3] - xAI shares will convert to SpaceX stock at a ratio of approximately 7:1, with the post-merger entity's stock priced at $527 per share [3] Financial Projections - SpaceX's private valuation has been raised to $1 trillion, driven by revenue growth from Starlink services, an increase of $200 billion from its previous valuation [5] - The merger is set to complete by March 16, with Musk at the helm of the combined entity [6] IPO Plans - SpaceX plans to go public in June, potentially raising up to $50 billion, which would make it the largest IPO in history, surpassing Saudi Aramco's $29 billion in 2019 [6] - The timing of the IPO is speculated to be influenced by astronomical events, but it may also be a strategic move to outpace competitors like OpenAI and Anthropic [6] Investor Concerns - Long-term investors express concerns that merging with the loss-making xAI could complicate or jeopardize the IPO process [7] - SpaceX will issue $250 billion in new shares to finance the acquisition, diluting existing shareholders [7] - SpaceX's annual revenue has reached $16 billion, while xAI's revenue was only a few hundred million, highlighting the disparity in financial health [7]

Investment Rating - The report indicates a cautious increase in investment rating for the dividend style, suggesting a potential style switch as the comprehensive factor value approaches the critical point [1][6]. Core Insights - The comprehensive factor value for the dividend style timing model was -0.08 for the week of January 26 to January 30, 2026, a significant recovery from -0.57 in the previous week, indicating a need to monitor potential style shifts [1][6]. - A substantial decrease in net financing contributed positively to the dividend, while the downward trend in U.S. Treasury yields has lessened its suppression on dividends. The momentum factor showed notable improvement, leading to better performance in dividends [1][7]. - Despite the positive contributions from market sentiment and the stock-bond valuation factor, the downward trend in U.S. Treasury yields and improvements in industry prosperity remain significant negative contributors, creating a counterbalance that brings the comprehensive factor value close to zero [1][7]. Summary by Sections - **Model Latest Results**: The comprehensive factor value for the dividend style timing model was updated to -0.08 for the week of January 26 to January 30, 2026, showing a notable recovery from the previous week's value of -0.57, indicating a potential style switch [1][6]. - **Factor Contributions**: The report highlights that net financing has decreased significantly, providing a positive contribution to dividends. The ongoing decline in U.S. Treasury yields has reduced its negative impact, while the momentum factor has improved, leading to better dividend performance [1][7].

Quantitative Models and Construction Methods Macro Indicator Trend Model - **Model Name**: Macro Indicator Trend Model - **Construction Idea**: Establish the relationship between macro indicators and asset performance by analyzing the trend of macro indicators and their impact on monthly asset returns[17][18] - **Construction Process**: - Use monthly moving averages of macro indicators to classify them into upward or downward trends - Apply T-test to determine whether the distribution of monthly returns of assets differs significantly under upward and downward trends - Formula: $ t = \frac{\overline{R_1} - \overline{R_2}}{\sqrt{\frac{(n_1-1)S_1^2 + (n_2-1)S_2^2}{n_1+n_2-2}(\frac{1}{n_1} + \frac{1}{n_2})}} \sim t_{n_1+n_2-2} $ - $\overline{R_1}$ and $\overline{R_2}$: Average monthly returns under upward and downward trends - $S_1$ and $S_2$: Standard deviations of monthly returns under upward and downward trends - $n_1$ and $n_2$: Number of months under upward and downward trends[17][18] - **Evaluation**: Effectively identifies macro indicators with significant impacts on asset returns[17][18] Technical Perspective Model - **Model Name**: Technical Perspective Model - **Construction Idea**: Evaluate asset trends, valuation, and fund flows using historical data and specific calculation methods[22][23][25] - **Construction Process**: - **Trend**: Use closing prices or LLT indicators to calculate trend indicators. Assign +1 for upward trends and -1 for downward trends[22] - **Valuation**: Calculate equity risk premium (ERP) as the reciprocal of PE(TTM) minus the 10-year government bond yield. Define historical 5-year percentile as: $ (Current ERP - Historical 5-year ERP Minimum) / (Historical 5-year ERP Maximum - Historical 5-year ERP Minimum) $ Assign scores based on percentile levels: +2 for >90%, +1 for 70%-90%, 0 for 30%-70%, -1 for 10%-30%, -2 for <10%[23][25] - **Fund Flows**: Calculate monthly active net inflows for indices and assess marginal changes. Assign +1 for positive changes and -1 for negative changes[26] - **Evaluation**: Provides a comprehensive view of asset trends, valuation, and fund flows[22][23][25] Fixed Proportion + Macro Indicators + Technical Indicators Combination Model - **Model Name**: Fixed Proportion + Macro Indicators + Technical Indicators Combination Model - **Construction Idea**: Adjust asset weights based on macro and technical indicators while maintaining a fixed proportion baseline[36][40] - **Construction Process**: - Set baseline weights for equity, bonds, commodities, and currency assets - Adjust weights monthly based on macro and technical indicator signals[36][40] - **Evaluation**: Balances fixed proportion allocation with dynamic adjustments for improved performance[36][40] Controlled Volatility + Macro Indicators + Technical Indicators Combination Model - **Model Name**: Controlled Volatility + Macro Indicators + Technical Indicators Combination Model - **Construction Idea**: Limit annualized volatility to 6% while dynamically adjusting weights based on macro and technical indicators[46][50] - **Construction Process**: - Use risk parity as the baseline weight - Adjust weights monthly based on macro and technical indicator signals[46][50] - **Evaluation**: Reduces volatility while maintaining competitive returns[46][50] Equity Style Rotation Models - **Model Name**: Equity Style Rotation Models (Large/Small Cap and Growth/Value) - **Construction Idea**: Adjust weights between equity styles based on macro and technical indicators[57][58] - **Construction Process**: - Set baseline weights for large/small cap and growth/value styles - Adjust weights monthly based on macro and technical indicator signals[57][58] - **Evaluation**: Captures style rotation opportunities for enhanced returns[57][58] --- Model Backtesting Results Macro Indicator Trend Model - **Annualized Return**: Not explicitly provided - **Maximum Drawdown**: Not explicitly provided - **Annualized Volatility**: Not explicitly provided Technical Perspective Model - **Annualized Return**: Not explicitly provided - **Maximum Drawdown**: Not explicitly provided - **Annualized Volatility**: Not explicitly provided Fixed Proportion + Macro Indicators + Technical Indicators Combination Model - **Annualized Return**: 10.20%[40] - **Maximum Drawdown**: 9.27%[40] - **Annualized Volatility**: 6.14%[40] Controlled Volatility + Macro Indicators + Technical Indicators Combination Model - **Annualized Return**: 10.46%[50] - **Maximum Drawdown**: 7.37%[50] - **Annualized Volatility**: 5.54%[50] Large/Small Cap Rotation Model - **Annualized Return**: 14.30%[61] - **Maximum Drawdown**: 49.10%[61] - **Annualized Volatility**: 22.30%[61] Growth/Value Rotation Model - **Annualized Return**: 14.43%[68] - **Maximum Drawdown**: 45.18%[68] - **Annualized Volatility**: 21.57%[68]

Content: --------- <doc id='1'>aa 证券研究报告·金融工程报告 2026 年 2 月 2 日 江海证券研究发展部 金融工程定期报告 分析师：梁俊炜 执业证书编号： S1410524090001 联系人：朱威 执业证书编号： S1410124010022 相关研究报告 1.可转债跟踪周报：转债较权益超额 延 续 ， 可 关 注 顺 周 期 板 块 — 2026.01.26 4.可转债跟踪周报：转债震荡调整， 估值有所回落—2026.01.05 5.可转债跟踪周报：转债反弹趋势不 变，股性提升—2025.12.29 转债随权益走弱，整体或将维持震荡 核心内容： ◆可转债市场表现：</doc> <doc id='2'>◆可转债个券表现：</doc> <doc id='3'>◆可转债条款跟踪：</doc> <doc id='4'>2.可转债跟踪周报：市场延续"春季 躁 动 " 行 情 ， 转 债 跑 出 超 额 — 2026.01.19 3.可转债跟踪周报：权益带动转债开 年大涨，可关注高景气度板块— 2026.01.12</doc> <doc id='5'>➢ 近一周（2026-01-26 至 2026-01-30），上证转债、深证转债、中证转债指数 周涨跌幅分别为-2.565%、-2.606%、-2.608%。权益市场中，上证指数周涨 跌幅为-0.440%，收于 4117.95 点；中证全指周涨跌幅为-1.536%，收于 6259.18 点。对比转债市场与权益市场，中证转债相较于中证全指周绝对收 益为-1.072%。 ➢ 近一周（2026-01-26 至 2026-01-30），可转债市场成交量与成交额分别为 247,604.22 万张与 45,107,382.59 万元，周环比变化分别为 2.33%、-3.32%； 可转债对应正股成交量与成交额分别为 5,708,126.68 万股与 100,193,434.58 万元，周环比变化为 9.15%、5.37%。对比上周，转债与正股成交活跃度小 幅提升。 ➢ 截止至 2026-01-30，可转债存续个券数为 380 只，发行规模约为 5979.58 亿 元，剩余规模约为 5068.18 亿元。转债市场转股溢价率中位数约为 32.24%、算术平均数约为 44.30%，周环比变化分别为 1.03%、2.03%。相比 上周，转股溢价率中位数窄幅波动。 ➢ 近一周（2026-01-26 至 2026-01-30）可转债个券涨幅前五分别为联瑞转 债、耐普转 02、汇车退债、天准转债、百川转 2，周涨跌幅分别为 126.51%、57.30%、43.85%、16.12%、12.75%；可转债个券跌幅前五分别为 新致转债、航宇转债、东时转债、冠中转债、惠城转债，周涨跌幅分别为- 21.57%、-20.59%、-16.38%、-16.31%、-15.82%。 ➢ 截止至 2026-01-30，可转债价格小于 100、100-110、110-120、120-130、 130-140、大于 140 的个券数量分别为 0 只（0.00%）、8 只（2.13%）、11 只 （2.93%）、75 只（19.95%）、93 只（24.73%）、189 只（50.27%），转股溢 价率中位数分别为 0.00%、51.23%、40.42%、66.92%、34.47%、26.04%，周 环比变动分别为 0.00%、-58.04%、2.85%、-17.33%、-11.14%、-2.58%。 ➢ 截止至 2026-01-30，触发下修条款的可转债个券数量 95，本周可能触发有 条件赎回条款的个券共 19 只，为富春转债（14/30）、佳禾转债（14/30）、 姚记转债（14/30）、微芯转债（14/30）、凤 21 转债（14/30）、银邦转债 （13/30）、首华转债（13/30）、华辰转债（12/30）、利柏转债（12/30）、众 和转债（12/30）、泰瑞转债（12/30）、力诺转债（12/30）、嘉元转债 （12/30）、信服转债（12/30）、永吉转债（11/30）、荣 23 转债（10/30）、 煜邦转债（10/30）、岱美转债（10/30）、润达转债（10/30）。 ◆风险提示：本报告可能存在数据缺失、数据错误、数据不及时、模型处理错 误等风险。本报告仅从金融工程角度，对可转债的市场与个券数据进行跟 踪、统计、分析，不构成对市场指数、行业或可转债个券进行预测或推荐。</doc> <doc id='7'>| 1 | 可转债市场表现 | 2 | | --- | --- | --- | | | 1.1 市场行情 | 2 | | 2 | 可转债个券表现 | 5 | | | 2.1 个券行情 | 5 | | | 2.2 估值分析 | 7 | | 3 | 可转债条款跟踪 | 8 | | | 4 风险提示 | 9 |</doc> <doc id='9'>| 图 | 1、转债指数近一年走势对比 | 2 | | --- | --- | --- | | 图 | 2、中证转债与权益市场指数近一年走势对比 | 2 | | 图 | 3、转债与正股区间成交对比（周度） | 3 | | 图 | 4、转债市场发行规模与每日存续个券数量 | 3 | | 图 | 5、转债市场剩余规模和每日成交额 | 4 | | 图 | 6、转债市场转股溢价率中位数走势（%） | 4 | | 图 | 7、不同信用评级转债近一年累计涨跌幅 | 5 | | 图 | 8、不同价格转债指数近一年累计涨跌幅 | 5 | | 图 | 9、不同规模转债指数近一年累计涨跌幅 | 6 | | 图 | 10、不同策略转债指数近一年累计涨跌幅 | 6 | | 图 | 11、涨幅前 5 个券与正股涨跌幅对比 | 6 | | 图 | 12、跌幅前 5 个券与正股涨跌幅对比 | 7 | | 图 | 13、不同价格个券数量构成 | 7 | | 图 | 14、不同价格个券转股溢价率走势（%） | 7 | | 图 | 15、不同价格个券转股溢价率中位数变动（%） | 8 | | 表 | 1、近一周可转债个券涨幅 TOP 5 | 5 | | 表 | 2、近一周可转债个券跌幅 TOP 5 | 5 |</doc> <doc id='10'>1 可转债市场表现 1.1 市场行情 近一周（2026-01-26 至 2026-01-30），上证转债、深证转债、中证转债 指数周涨跌幅分别为-2.565%、-2.606%、-2.608%。权益市场中，上证指数 周涨跌幅为-0.440%，收于 4117.95 点；中证全指周涨跌幅为-1.536%，收于 6259.18 点。对比转债市场与权益市场，中证转债相较于中证全指绝对收益 为-1.072%。 近一周（2026-01-26 至 2026-01

Quantitative Models and Construction Methods 1. Model Name: Convolutional Neural Network (CNN) for Price-Volume Data Modeling - **Model Construction Idea**: The model leverages convolutional neural networks to analyze standardized graphical representations of price-volume data, aiming to predict future price trends and map learned features to industry thematic indices[79][81] - **Model Construction Process**: - Standardize price-volume data into graphical formats for each stock within a specific time window[79] - Apply convolutional neural networks to extract features from these graphical representations[79] - Map the extracted features to thematic industry indices, such as the CSI Petrochemical Industry Index, CSI Subdivision Chemical Industry Theme Index, and others[81] - **Model Evaluation**: The model effectively identifies industry themes based on price-volume data and provides actionable insights for sector allocation[79][81] --- Model Backtesting Results 1. CNN Model - **Thematic Indices Configured**: - CSI Petrochemical Industry Index (h11057.CSI)[81] - CSI Subdivision Chemical Industry Theme Index (000813.CSI)[81] - CNI Oil & Gas Index (399439.SZ)[81] - CSI Oil & Gas Resources Index (931248.CSI)[81] - CNI Nonferrous Metals Index (399395.SZ)[81]

Key Points Summary Company Involved - **NVIDIA and CoreWeave**: The collaboration between NVIDIA and CoreWeave is highlighted as a case study in financial engineering, showcasing a complex web of mutual benefits. Core Insights and Arguments - **Equity Appreciation**: NVIDIA's early investment in CoreWeave increased its valuation from $2 billion to $23 billion before the IPO, although it decreased to $19 billion post-IPO. Despite this, NVIDIA still benefits from its equity holdings [1][2] - **Supply Chain Integration**: CoreWeave utilized financing to purchase NVIDIA GPUs, allowing NVIDIA to earn sales profits while maintaining control over market dynamics through preferential supply agreements [1][2] - **Collateral Financing**: CoreWeave secured loans against GPUs, which were then reinvested into further GPU purchases, creating a "borrow-purchase-reborrow" cycle that indirectly boosted NVIDIA's shipment volumes [2] - **Price Control Mechanism**: Through CoreWeave's computing power leasing business, NVIDIA can indirectly influence GPU supply and demand. For instance, during chip upgrades, NVIDIA can quickly clear out old inventory via CoreWeave, helping to maintain high prices for new products [2] Other Important but Potentially Overlooked Content - The collaboration exemplifies a strategic partnership that maximizes financial returns for both companies, demonstrating innovative approaches to capitalizing on market opportunities [1][2]