- **Quantitative model and construction method** - **Model name**: Three-dimensional timing framework - **Model construction idea**: The model integrates liquidity, divergence, and prosperity indicators to assess market trends and provide timing signals [7][11][12] - **Model construction process**: 1. **Liquidity index**: Calculated based on market trading volume and other liquidity-related metrics [18] 2. **Divergence index**: Measures the degree of disagreement among market participants [16] 3. **Prosperity index**: Reflects the overall economic and market health, scaled to match the dimension of the Shanghai Composite Index [20] 4. Combine the three indices into a unified framework to evaluate market conditions and predict trends [12] - **Model evaluation**: The model maintains a stable performance in predicting market trends, with historical data showing its effectiveness in identifying periods of market oscillation and downturns [14] - **Quantitative factor and construction method** - **Factor name**: Growth factor - **Factor construction idea**: Measures the growth potential of stocks based on financial metrics such as revenue and profit growth [39][40] - **Factor construction process**: 1. Calculate the growth rate of key financial metrics, such as revenue, profit, and liabilities [42][44] 2. Normalize the metrics by market capitalization and industry to ensure comparability [41] 3. Construct the factor by aggregating the normalized metrics into a composite score [42][44] - **Factor evaluation**: The growth factor demonstrated positive returns, with high-growth stocks outperforming low-growth stocks in the recent week [40][42] - **Factor name**: Size factor - **Factor construction idea**: Evaluates the performance of stocks based on their market capitalization [39] - **Factor construction process**: 1. Divide stocks into groups based on market capitalization [39] 2. Calculate the average return for each group [39] 3. Compare the performance of large-cap stocks against small-cap stocks [39] - **Factor evaluation**: Large-cap stocks outperformed small-cap stocks, with the size factor recording positive returns [39] - **Factor name**: Beta factor - **Factor construction idea**: Measures the sensitivity of stocks to market movements [40] - **Factor construction process**: 1. Calculate the beta of each stock based on historical price movements relative to the market [40] 2. Group stocks into high-beta and low-beta categories [40] 3. Compare the performance of high-beta stocks against low-beta stocks [40] - **Factor evaluation**: High-beta stocks outperformed low-beta stocks, with the beta factor recording positive returns [40] - **Factor name**: Alpha factors (multiple) - **Factor construction idea**: Focuses on growth-related metrics and analyst adjustments to predict stock performance [42][46] - **Factor construction process**: 1. Calculate metrics such as single-quarter ROE growth, revenue growth, and analyst forecast adjustments [42][46] 2. Normalize these metrics by market capitalization and industry [41] 3. Aggregate the metrics into individual alpha factors [42][46] - **Factor evaluation**: Alpha factors such as single-quarter ROE growth and analyst forecast adjustments showed strong performance, particularly in small and mid-cap stocks [46][47] - **Model backtesting results** - **Three-dimensional timing framework**: Historical performance indicates stable prediction of market oscillations and downturns [14] - **Factor backtesting results** - **Growth factor**: Weekly long-side excess return of 0.42% [40] - **Size factor**: Weekly long-side excess return of 1.57% [39] - **Beta factor**: Weekly long-side excess return of 1.08% [40] - **Alpha factors**: - Single-quarter ROE growth (considering quick reports and forecasts): Weekly excess return of 1.61%, monthly excess return of 10.17% [44][47] - Analyst forecast adjustment (np_FY1): Weekly excess return of 7.14% in CSI 300, 5.60% in CSI 500, 9.54% in CSI 1000, and 4.19% in CSI 2000 [47] - Single-quarter ROE growth (report): Weekly excess return of 7.47% in CSI 300, 3.84% in CSI 500, 8.11% in CSI 1000, and 3.09% in CSI 2000 [47]