金 融 工 程 权益黄金尽墨，全球资产 BL 模型 2 本周 微录正收益 大类资产配置模型周报第 40 期 本报告导读： 全球资产 BL 模型 2 录得涨幅 0.01%；国内资产 BL 模型 1、基于宏观因子的资产配 置模型、国内资产风险平价模型、全球资产风险平价模型、全球资产 BL 模型 1 和 国内资产 BL 模型 2 分录跌幅 0.32%、0.31%、0.27%、0.2%、0.17%和 0.15%。 投资要点： | [Table_Authors] | 郑雅斌(分析师) | | --- | --- | | | 021-23219395 | | | zhengyabin@gtht.com | | 登记编号 | S0880525040105 | | | 张雪杰(分析师) | | | 0755-23976751 | | | zhangxuejie@gtht.com | | 登记编号 | S0880522040001 | | | 朱惠东(分析师) | | | 0755-23976176 | | | zhuhuidong@gtht.com | | 登记编号 | S0880525070025 | [Table_Rep ...

- The BL model is an improvement of the traditional mean-variance optimization (MVO) model, developed by Fisher Black and Robert Litterman in 1990. It integrates Bayesian theory to combine subjective views with quantitative asset allocation models, optimizing asset weights based on investor forecasts of market returns. This model addresses MVO's sensitivity to expected returns and offers higher tolerance compared to purely subjective investment approaches, providing efficient asset allocation solutions[12][13] - The BL model was implemented for both global and domestic assets. For global assets, it utilized indices such as S&P 500, Hang Seng Index, and Nanhua Commodity Index. For domestic assets, it included indices like CSI 300, CSI 1000, and SHFE Gold. Two versions of BL models were developed for each market, focusing on equities, bonds, commodities, and gold[13][14] - The Risk Parity model, introduced by Bridgewater in 2005, aims to equalize risk contributions across asset classes in a portfolio. It calculates initial asset weights based on expected volatility and correlation, then optimizes deviations between actual and expected risk contributions to determine final weights[17][18] - The Risk Parity model was constructed in three steps: selecting appropriate underlying assets, calculating risk contributions of each asset to the portfolio, and solving optimization problems to determine asset weights. It was applied to both global and domestic assets, using indices like CSI 300, CSI 1000, and COMEX Gold for domestic assets, and S&P 500, Hang Seng Index, and Nanhua Commodity Index for global assets[19][21] - The macro factor-based asset allocation model incorporates six macro risks: growth, inflation, interest rates, credit, exchange rates, and liquidity. Using Factor Mimicking Portfolio methodology, high-frequency macro factors were constructed. The strategy involves calculating asset factor exposures, determining benchmark exposures, setting subjective factor deviations based on macro forecasts, and solving for asset weights to reflect macro risk judgments[23][26] - The macro factor-based model was applied to domestic assets, including indices like CSI 300, CSI 1000, and SHFE Gold. For example, in September 2025, subjective factor deviations were set as 0 for growth, inflation, interest rates, and credit, 1 for exchange rates, and 0 for liquidity, reflecting macroeconomic conditions at the time[25][27] - Domestic BL Model 1 achieved weekly returns of 0.1%, monthly returns of 0.38%, and annual returns of 3.97%, with annualized volatility of 2.23% and maximum drawdown of 1.31%[14][17] - Domestic BL Model 2 recorded weekly returns of -0.01%, monthly returns of 0.48%, and annual returns of 3.68%, with annualized volatility of 2.02% and maximum drawdown of 1.06%[14][17] - Global BL Model 1 delivered weekly returns of 0.54%, monthly returns of 0.03%, and annual returns of 1.02%, with annualized volatility of 2.04% and maximum drawdown of 1.64%[14][17] - Global BL Model 2 achieved weekly returns of 0.37%, monthly returns of 0.35%, and annual returns of 2.43%, with annualized volatility of 1.65% and maximum drawdown of 1.28%[14][17] - Domestic Risk Parity Model recorded weekly returns of 0.14%, monthly returns of 0.34%, and annual returns of 3.47%, with annualized volatility of 1.34% and maximum drawdown of 0.76%[21][22] - Global Risk Parity Model achieved weekly returns of 0.22%, monthly returns of 0.39%, and annual returns of 2.99%, with annualized volatility of 1.46% and maximum drawdown of 1.2%[21][22] - Macro Factor-Based Model delivered weekly returns of -0.25%, monthly returns of 0.73%, and annual returns of 4.29%, with annualized volatility of 1.54% and maximum drawdown of 0.64%[27][28]

- The Black-Litterman (BL) model is an improved version of the mean-variance optimization (MVO) model developed by Fisher Black and Robert Litterman in 1990. It combines Bayesian theory with quantitative asset allocation models, allowing investors to incorporate subjective views into asset return forecasts and optimize portfolio weights. This model addresses MVO's sensitivity to expected returns and provides a more robust framework for efficient asset allocation[12][13][14] - The BL model was implemented for both global and domestic assets. For global assets, it utilized indices such as the S&P 500, Hang Seng Index, and COMEX Gold. For domestic assets, it included indices like CSI 300, CSI 1000, and SHFE Gold. Two variations of the BL model were constructed for each asset category[13][14][18] - The Risk Parity model, introduced by Bridgewater in 2005, aims to equalize risk contributions across asset classes in a portfolio. It calculates initial asset weights based on expected volatility and correlation, then optimizes deviations between actual and expected risk contributions to determine final portfolio weights[17][18][20] - The Risk Parity model was applied to both global and domestic assets. Global assets included indices such as CSI 300, S&P 500, and COMEX Gold, while domestic assets incorporated CSI 300, CSI 1000, and SHFE Gold. The model followed a three-step process: selecting assets, calculating risk contributions, and solving optimization problems for portfolio weights[18][20][21] - The Macro Factor-based Asset Allocation model constructs a framework using six macroeconomic risk factors: growth, inflation, interest rates, credit, exchange rates, and liquidity. It employs Factor Mimicking Portfolio methods to calculate high-frequency macro factors and integrates subjective views on macroeconomic conditions into asset allocation decisions[22][24][25] - The Macro Factor-based model involves four steps: calculating factor exposures for assets, determining benchmark factor exposures using a Risk Parity portfolio, incorporating subjective factor deviations based on macroeconomic forecasts, and solving for asset weights that align with target factor exposures[22][24][25] Model Performance Metrics - Domestic BL Model 1: Weekly return -0.11%, September return -0.14%, 2025 YTD return 3.23%, annualized volatility 2.19%, maximum drawdown 1.31%[14][17] - Domestic BL Model 2: Weekly return -0.11%, September return -0.13%, 2025 YTD return 2.84%, annualized volatility 1.99%, maximum drawdown 1.06%[14][17] - Global BL Model 1: Weekly return 0.04%, September return 0.11%, 2025 YTD return 0.84%, annualized volatility 1.99%, maximum drawdown 1.64%[14][17] - Global BL Model 2: Weekly return 0.00%, September return 0.03%, 2025 YTD return 1.84%, annualized volatility 1.63%, maximum drawdown 1.28%[14][17] - Domestic Risk Parity Model: Weekly return -0.06%, September return 0.05%, 2025 YTD return 2.99%, annualized volatility 1.35%, maximum drawdown 0.76%[20][21] - Global Risk Parity Model: Weekly return -0.07%, September return 0.13%, 2025 YTD return 2.50%, annualized volatility 1.48%, maximum drawdown 1.20%[20][21] - Macro Factor-based Model: Weekly return 0.04%, September return 0.26%, 2025 YTD return 3.29%, annualized volatility 1.32%, maximum drawdown 0.64%[26][27]

Group 1 - The report indicates that the overall asset allocation strategy has experienced fluctuations due to domestic equity asset volatility, with various models recording different degrees of decline [1][4][7] - The performance of major asset classes from September 15 to September 19, 2025, shows that the S&P 500, Hang Seng Index, and other indices recorded gains, while convertible bonds and gold experienced declines [7][10] - The domestic asset BL model 1 and model 2 both reported a weekly return of -0.04%, while the global asset BL models had slightly better performance with a return of -0.01% for model 1 and -0.03% for model 2 [15][17] Group 2 - The Black-Litterman (BL) model is highlighted as an improvement over traditional mean-variance models, integrating subjective views with quantitative models to optimize asset allocation [12][13] - The domestic asset risk parity model achieved a return of -0.02% for the week, while the global asset risk parity model recorded a positive return of 0.05% [21][22] - The macro factor-based asset allocation strategy reported a weekly return of -0.1%, with a year-to-date return of 3.25%, indicating its performance amidst changing economic conditions [27][28]