Quantitative Models and Construction Methods 1. Model Name: Global Asset Allocation Model Based on Artificial Intelligence - **Model Construction Idea**: This model applies machine learning techniques to global asset allocation, leveraging factor investment principles to rank assets and construct a monthly quantitative equal-weight allocation strategy[38] - **Model Construction Process**: - The model evaluates various global assets using machine learning algorithms to assign scores to each asset - Based on these scores, the model determines the optimal allocation weights for each asset class - For March, the recommended weights were: Government Bond Index 75.99%, SHFE Gold 23.50%, and Nasdaq 0.51%[38][42] - **Model Evaluation**: The model demonstrates superior performance in terms of return, risk-adjusted return, and drawdown control compared to the benchmark[40] 2. Model Name: Stock-Bond Allocation Model Based on Dynamic Macro Event Factors - **Model Construction Idea**: This model integrates a macro timing module and a risk budgeting framework to allocate between stocks and bonds for different risk preferences (Conservative, Balanced, and Aggressive)[45] - **Model Construction Process**: - The macro timing module evaluates signals from economic growth and monetary liquidity dimensions - The risk budgeting framework adjusts stock and bond weights based on the strength of these signals - For February, the stock weights were 30.00% (Aggressive), 13.81% (Balanced), and 0.00% (Conservative), with the remaining allocated to bonds[45][46] - **Model Evaluation**: The model has shown strong historical performance, with high annualized returns and Sharpe ratios across all risk profiles[46][51] 3. Model Name: Dividend Style Timing Model - **Model Construction Idea**: This model uses a dynamic macro event factor system based on economic growth and monetary liquidity indicators to time allocations to the CSI Dividend Index[52] - **Model Construction Process**: - The model evaluates 10 indicators across economic growth and monetary liquidity dimensions - For March, the final composite signal was 1, with economic growth indicators (e.g., power generation, PPI YoY, PPI-CPI spread) signaling bullish, while no monetary liquidity indicators signaled bullish[52][54] - **Model Evaluation**: The strategy outperforms the CSI Dividend Total Return Index in terms of annualized return, volatility, and drawdown, with a smoother net value curve[52][53] --- Model Backtesting Results 1. Global Asset Allocation Model Based on Artificial Intelligence - **Annualized Return**: 16.48% - **Annualized Volatility**: 6.78% - **Maximum Drawdown**: -6.66% - **Sharpe Ratio**: 1.15 - **Year-to-Date Return**: 5.72%[40][43] 2. Stock-Bond Allocation Model Based on Dynamic Macro Event Factors - **Aggressive Profile**: - Annualized Return: 20.07% - Annualized Volatility: 14.00% - Maximum Drawdown: -13.72% - Sharpe Ratio: 1.31 - Year-to-Date Return: 5.42% - **Balanced Profile**: - Annualized Return: 10.79% - Annualized Volatility: 8.09% - Maximum Drawdown: -6.77% - Sharpe Ratio: 1.19 - Year-to-Date Return: 1.81% - **Conservative Profile**: - Annualized Return: 5.85% - Annualized Volatility: 3.19% - Maximum Drawdown: -3.55% - Sharpe Ratio: 1.49 - Year-to-Date Return: 0.83%[46][51] 3. Dividend Style Timing Model - **Annualized Return**: 15.85% - **Annualized Volatility**: 16.17% - **Maximum Drawdown**: -21.22% - **Sharpe Ratio**: 0.93 - **Recent 1-Month Return**: 1.35%[52][53]

- The AI-based global asset allocation model suggests a weight of 74.76% for the government bond index, 24.97% for SHFE gold, and 0.27% for the Hang Seng Index for February[5][44] - The model's performance in January showed a monthly return of -0.25%, compared to the benchmark strategy's return of 0.14%[5][44] - Historical performance from January 2021 to January 2026 indicates an annualized return of 7.22%, a Sharpe ratio of 1.07, and a maximum drawdown of 6.66%[46] - The stock-bond allocation model, based on macro timing and risk budgeting, suggests stock weights of 10.19%, 16.91%, and 70.00% for conservative, balanced, and aggressive profiles, respectively, for February[6][50] - The model's performance in January showed monthly returns of 3.65%, 1.22%, and 0.39% for aggressive, balanced, and conservative profiles, respectively[6][50] - Historical performance from January 2005 to January 2026 indicates annualized returns of 20.15%, 10.85%, and 5.87% for aggressive, balanced, and conservative profiles, respectively[51][57] - The dividend timing model recommends a 100% position in the CSI Dividend Index for February[7][58] - The model's performance shows an annualized return of 15.85%, an annualized volatility of 17.26%, a maximum drawdown of -21.22%, and a Sharpe ratio of 0.90[7][59] - The model's recent one-month return is 0.00%, compared to the CSI Dividend Total Return Index's return of 3.76%[7][59]

- The **Artificial Intelligence Global Asset Allocation Model** applies machine learning to asset allocation problems, utilizing factor investment principles to score and rank assets, ultimately constructing a monthly quantitative equal-weighted strategy for global asset allocation[38][39][41] - The model's suggested weights for January include: government bond index (68.27%), SHFE gold (28.55%), Nasdaq (1.02%), ICE Brent oil (1.24%), and CSI 500 (0.92%)[38][41] - Historical performance from January 2021 to December 2025 shows annualized return of 6.78%, Sharpe ratio of 1.04, maximum drawdown of 6.66%, and excess annualized return of -0.38% compared to the benchmark[39][42] - Year-to-date return for the strategy is 7.18%, while the benchmark return is 18.14%[40][42] - The **Dynamic Macro Event Factor-Based Stock-Bond Rotation Strategy** incorporates macro timing modules and risk budgeting frameworks to generate stock-bond allocation weights for three risk profiles: aggressive, balanced, and conservative[43][44][45] - January stock weights are: aggressive (55.00%), balanced (14.60%), and conservative (0.00%)[43][45] - December macro signals include 60% strength for both economic growth and monetary liquidity dimensions[43][45] - Historical performance from January 2005 to December 2025 shows annualized returns of 20.03% (aggressive), 10.84% (balanced), and 5.88% (conservative), outperforming the benchmark's 8.97%[44][49] - Year-to-date returns are 15.77% (aggressive), 4.23% (balanced), and 0.70% (conservative), compared to the benchmark's 15.95%[44][49] - The **Dividend Style Timing Strategy** leverages 10 indicators from economic growth and monetary liquidity dimensions to construct a timing strategy for dividend indices, showing enhanced stability compared to the CSI Dividend Total Return Index[50][51][53] - January recommended allocation for CSI Dividend is 0%, as most signals did not indicate a bullish outlook[50][54] - Historical performance includes annualized return of 16.18%, maximum drawdown of -21.22%, and Sharpe ratio of 0.93, outperforming the CSI Dividend Total Return Index's annualized return of 11.28% and Sharpe ratio of 0.57[50][53]

- The artificial intelligence global asset allocation model applies machine learning to asset allocation problems, using factor investment ideas to score and rank assets, ultimately constructing a monthly quantitative equal-weighted strategy for global asset allocation[38][39][40] - The dynamic macroeconomic event factor-based stock-bond rotation strategy includes three risk preference models (conservative, balanced, and aggressive), utilizing macro timing modules and risk budgeting frameworks to determine stock and bond weights[43][44][45] - The dividend style timing model uses 10 indicators from economic growth and monetary liquidity dimensions, constructing a timing strategy for the dividend index, which shows significant stability improvement compared to the CSI Dividend Total Return Index[51][54][55] Model Backtesting Results - Artificial intelligence global asset allocation model: annualized return 38.76%, Sharpe ratio 1.07, maximum drawdown -6.56%, year-to-date return 6.81%[39][40][42] - Dynamic macroeconomic event factor-based stock-bond rotation strategy: aggressive model annualized return 20.14%, Sharpe ratio 1.30, maximum drawdown -13.72%, year-to-date return 14.42%; balanced model annualized return 10.92%, Sharpe ratio 1.19, maximum drawdown -6.77%, year-to-date return 4.13%; conservative model annualized return 5.94%, Sharpe ratio 1.50, maximum drawdown -3.55%, year-to-date return 0.97%[43][49][50] - Dividend style timing model: annualized return 16.52%, Sharpe ratio 1.07, maximum drawdown -13.77%, year-to-date return 0%[51][54][55]

- Model Name: Global Asset Allocation Model; Model Construction Idea: The model applies machine learning to asset allocation, using factor investment to score and rank assets, and ultimately constructs a monthly frequency equal-weighted quantitative strategy for global asset allocation[34] - Model Construction Process: The model suggests weights for different assets based on their scores. For August, the suggested weights are: Treasury Index (67.01%), ICE Brent Oil (15.25%), German DAX (9.06%), and Hang Seng Index (8.68%). The weights for German DAX and Hang Seng Index were increased, while the weights for ICE Brent Oil and Treasury Index were decreased[34] - Model Evaluation: The model has shown superior performance compared to the benchmark, with a higher Sharpe ratio and lower maximum drawdown[34][35] - Model Test Results: Annualized return: 6.77%, Sharpe ratio: 1.02, maximum drawdown: -6.66%, excess annualized return: 0.68%, excess Sharpe ratio: 0.15, excess maximum drawdown: -10.95%[34][35] - Model Name: Stock-Bond Allocation Model; Model Construction Idea: The model uses a macro timing module and a risk budget model framework to output the weights for three different risk profiles (aggressive, stable, and conservative)[6][40] - Model Construction Process: For August, the stock weights for aggressive, stable, and conservative profiles are 50%, 14.39%, and 0%, respectively. The model signals for economic growth are at 100%, while liquidity signals are at 0%[6][40] - Model Evaluation: The model has shown good performance, with the aggressive and stable profiles outperforming the benchmark in various dimensions[40] - Model Test Results: Annualized return for aggressive: 20.00%, stable: 10.97%, conservative: 6.01%, benchmark: 8.73%. Sharpe ratio for aggressive: 1.29, stable: 1.18, conservative: 1.51, benchmark: 0.53[40][46] - Model Name: Dividend Timing Model; Model Construction Idea: The model uses economic growth and liquidity indicators to construct a timing strategy for the dividend index[7][48] - Model Construction Process: For August, the recommended position for the CSI Dividend Index is 0%. The model combines signals from economic growth and liquidity indicators, which mostly show bearish signals[7][48] - Model Evaluation: The model has shown stable performance, with a significant improvement in stability compared to the CSI Dividend Total Return Index[7][48] - Model Test Results: Annualized return: 16.62%, Sharpe ratio: 0.94, maximum drawdown: -21.22%, compared to the CSI Dividend Total Return Index with an annualized return of 11.34%, Sharpe ratio: 0.57, and maximum drawdown: -36.80%[7][48][51]