Quantitative Models and Construction Methods 1. Model Name: Brinson Attribution Model - **Model Construction Idea**: The model is used to decompose the excess return of a portfolio into allocation and selection effects, helping to identify the sources of performance[44] - **Model Construction Process**: The Brinson model calculates the contribution of each sector to the portfolio's excess return. The formula is as follows: $ R_{excess} = \sum_{i=1}^{n} (w_{i,p} - w_{i,b}) \cdot R_{i,b} + \sum_{i=1}^{n} w_{i,p} \cdot (R_{i,p} - R_{i,b}) + \text{Other Effects} $ - $ w_{i,p} $: Portfolio weight in sector $ i $ - $ w_{i,b} $: Benchmark weight in sector $ i $ - $ R_{i,p} $: Portfolio return in sector $ i $ - $ R_{i,b} $: Benchmark return in sector $ i $ - The first term represents the allocation effect, the second term represents the selection effect, and the third term accounts for interaction effects[44][45] - **Model Evaluation**: The model effectively identifies that the portfolio's excess return primarily stems from allocation effects, with notable contributions from sectors like communication, automotive, and pharmaceuticals[44] 2. Model Name: Sharpe Style Factor Model - **Model Construction Idea**: This model identifies the portfolio's exposure to various style factors, such as size, value, and growth, by solving an optimization problem[46] - **Model Construction Process**: The Sharpe model uses the following steps: 1. Define style factors: Large-cap growth, large-cap value, mid-cap growth, mid-cap value, small-cap growth, small-cap value, and bonds 2. Set constraints based on the fund type 3. Solve a quadratic programming problem to determine the portfolio's exposure to each factor[46] - **Model Evaluation**: The model reveals that the portfolio has a preference for small- and mid-cap stocks, with a transition from growth to value style over time. However, the model may overestimate bond exposure due to the portfolio's low volatility[46] 3. Model Name: Barra-CNE5 Risk Factor Model - **Model Construction Idea**: This model evaluates the portfolio's exposure to specific risk factors, such as size, liquidity, and valuation, using weighted averages of individual stock exposures[47] - **Model Construction Process**: 1. Obtain detailed stock-level factor exposures from the Barra-CNE5 model 2. Calculate the portfolio's weighted average exposure to each factor 3. Analyze the results to identify significant factor exposures[47] - **Model Evaluation**: The model indicates high exposure to liquidity and small-cap factors, with a negative exposure to valuation factors. This aligns with the manager's preference for undervalued, small-cap stocks with growth potential[47] --- Model Backtesting Results 1. Brinson Attribution Model - **Excess Return Contribution**: Communication sector contributed 3.11% to the portfolio's excess return, with other sectors like automotive, pharmaceuticals, and oil & gas also showing significant positive contributions[44] 2. Sharpe Style Factor Model - **Style Exposure**: The portfolio shows a strong tilt towards small- and mid-cap stocks, with a shift from growth to value style post-2023. Bond exposure is likely overestimated due to low portfolio volatility[46] 3. Barra-CNE5 Risk Factor Model - **Factor Exposure**: High exposure to liquidity and small-cap factors, with a negative tilt towards valuation factors. The portfolio's weighted price-to-book ratio is significantly lower than the peer average, reflecting a focus on undervalued stocks[47]