Group 1 - The report indicates that the asset allocation for August shows a marginal increase in risk assets, with a corresponding decrease in safe-haven assets. The allocation to U.S. stocks and commodities has been raised, while U.S. Treasuries and dollar allocations have been reduced [3][5][6] - The A-share market is expected to revert to a barbell structure in August, with a shift towards low valuation, weak profitability, and small-cap stocks becoming the dominant market styles [9][14][16] - The report highlights that the high profitability and valuation factors observed in July may face a phase of adjustment, with the market likely to experience a temporary recovery in the barbell style [14][16] Group 2 - The report's analysis based on the improved BL model suggests that the relative strength of risk assets is likely to continue, with commodity asset allocations being increased and safe-haven asset allocations being reduced compared to July [3][5][9] - The report anticipates that the internal dynamics of risk assets will show a slight decrease in stock asset allocations, while commodity asset allocations will see an increase. This aligns with the forecast of a "two up, one down" trend in the A-share market for August [5][9][14] - The report emphasizes that the macroeconomic environment, including monetary and credit conditions, will play a crucial role in shaping market dynamics, with expectations of a stable monetary environment and a gradual recovery in credit conditions [14][16]

策略研究 | 证券研究报告 — 点评报告 2025 年 7 月 3 日 资产配置及A股风格半月报 风险资产有望延续优势 风险资产有望延续优势，盈利因子有望修复。 相关研究报告 《风格制胜 3：风格因子体系的构建及应用》 20250606 中银国际证券股份有限公司 具备证券投资咨询业务资格 策略研究 证券分析师：王君 (8610)66229061 jun.wang@bocichina.com 证券投资咨询业务证书编号：S1300519060003 证券分析师：郭晓希 (8610)66229019 xiaoxi.guo@bocichina.com 证券投资咨询业务证书编号：S1300521110001 ◼ 大类资产配置：风险资产有望延续相对优势。我们的大类资产配置模型是 基于周期嵌套理论改良版 BL 模型。模型基于不同周期定位下的大类资产 的表现，将市场均衡观点进行贝叶斯修正，输出满足既定条件的最优资产 组合，模型输出可显著提升组合夏普比率。我们输入的观点基于库存周期 理论，未来一个季度，我们认为内外弱补库有望延续。基于上述主观假设 及限制条件的 BL 模型输出结果为：国内资产方面，2025 年三季度股票 配置比 ...

Group 1 - The report emphasizes that under the current weak replenishment cycle, A-shares are expected to outperform other asset classes, with a recommendation to increase the allocation to A-shares while reducing commodity assets [2][24][25] - The report predicts a weak recovery in A-share earnings, with a projected growth rate of 0-5% for the second half of 2025, and a valuation contribution of 0-7%, leading to an expected median increase of 7% in A-shares [39][40] - The report identifies small-cap stocks, strong reversals, high valuations, and high profitability as the dominant market styles for the second half of 2025, with a particular focus on TMT (Technology, Media, and Telecommunications) sectors [46][47][48] Group 2 - The technology sector is highlighted as a high-probability choice for index breakout, supported by stable capital market commitments and sufficient policy reserves [54] - The report outlines two scenarios for industry allocation: one under a fluctuating market and another under a potential upward breakout, indicating the need for strategic planning [54] - The report suggests that the AI and humanoid robotics industries are expected to experience significant growth, with a focus on high-growth and consumption styles in the top ten recommended industries for the second half of 2025 [24][39]

固收+智能体：BL 模型+小模型实践 20250116 摘要 • BL 模型通过结合市场情况与投资者观点，生成后验收益率，解决了传统资 产配置模型对输入数据敏感的问题，提高了收益率预测的平稳性和准确性。 • 计算市场隐含资产收益率包括构建市场组合、利用 CAPM 计算预期收益率， 并假设无 Arrow 部分计算市场组合超额收益率，从而得到细分资产的隐含 收益率。 • BL 模型通过计算市场组合反映市场风险偏好，即单位风险敞口下市场要求 的收益率补偿，并将此转化为每个细分资产的预期收益率。 • 投资者观点通过绝对观点（如预期收益率）和相对观点（如超额收益率） 结合到 BL 模型中，并考虑资产相关性和置信区间，调整后验收益率。 • 在中国应用 BL 模型需使用基准组合替代市场组合，考虑合同约束（如风 险窗口），控制换仓频率和换手率，并可通过杠杆调整提高资金效用。 • 固收+智能体通过细分资产模型（如 GBR 模型）替代主观观点，利用量价 数据预测收益率，虽单个资产准确率有限，但在组合中可实现波动衰减。 • 引入置信区间（基于历史准确率）和细分资产指标（如转债溢价率、市场 情绪、利差）可显著提高 BL 模型预测准确 ...