Quantitative Models and Construction Methods - **Model Name**: Diffusion Index Timing Model (First Threshold Method) - **Model Construction Idea**: The diffusion index measures the proportion of constituent stocks in the Wind Micro-Cap Index that have been in an upward trend over the past 20 trading days. This model uses a left-side trading strategy based on threshold values to determine market timing[64][65] - **Model Construction Process**: 1. Define the diffusion index as the proportion of stocks in the index that have risen in the past 20 trading days 2. Set thresholds: when the diffusion index is above 0.9, the portfolio is fully in cash; when it is below 0.1, the portfolio is fully invested 3. Fill in signals for other periods based on the previous period's signal 4. Ensure that a full investment must precede a full cash position, and vice versa - **Formula**: Not explicitly provided in the report - **Model Evaluation**: The model failed to outperform the Micro-Cap Index in-sample but demonstrated controlled drawdowns. However, it struggled to identify major bear markets in 2024 and exited too early during the bull market in September 2024, missing significant gains. It performed better in 2025 by capturing the bottom and subsequent upward trends[65][66] - **Model Name**: Diffusion Index Timing Model (Delayed Threshold Method) - **Model Construction Idea**: This right-side trading strategy modifies the first threshold method by using momentum instead of reversal logic. It aims to address the shortcomings of the left-side approach by focusing on momentum[67] - **Model Construction Process**: 1. Define the diffusion index as the proportion of stocks in the index that have risen in the past 20 trading days 2. Set thresholds: when the diffusion index drops below 0.9 after being above it the previous day, the portfolio moves to cash; when it rises above 0.1 after being below it the previous day, the portfolio becomes fully invested 3. Fill in signals for other periods based on the previous period's signal - **Formula**: Not explicitly provided in the report - **Model Evaluation**: The model failed to outperform the Micro-Cap Index in-sample but demonstrated controlled drawdowns. It avoided early exits during the September 2024 bull market, preserving gains and achieving a new high in strategy net value. Its performance in 2025 was mediocre, capturing some upward trends but overall underwhelming[67][69] - **Model Name**: Diffusion Index Timing Model (Dual Moving Average Method) - **Model Construction Idea**: This adaptive trading strategy uses short-term and long-term moving averages of the diffusion index to determine market timing. It aims to reduce reliance on fixed thresholds and adapt to market trends[68][70] - **Model Construction Process**: 1. Define the short-term moving average as the 10-day moving average (MA) of the diffusion index 2. Define the long-term moving average as the 20-day moving average of the short-term MA 3. When the short-term MA crosses above the long-term MA, the portfolio becomes fully invested; when it crosses below, the portfolio moves to cash - **Formula**: Not explicitly provided in the report - **Model Evaluation**: The model failed to outperform the Micro-Cap Index in-sample but demonstrated excellent drawdown control. It avoided significant losses during the 2024 bear market and preserved gains during the September 2024 bull market. However, it showed limitations in trend recognition due to insufficient parameter smoothing, leading to suboptimal performance in 2025[70][71] Model Backtesting Results - **First Threshold Method**: - Triggered a full investment signal on April 7, 2025, at a diffusion index value of 0.0225[65][66] - **Delayed Threshold Method**: - Triggered an opening signal on April 22, 2025[67][69] - **Dual Moving Average Method**: - Triggered an opening signal on April 30, 2025[70][71]