KAIYUAN SECURITIES·2025-05-12 13:17Quantitative Models and Construction Methods 1. Model Name: Index Adjustment Prediction Model - Model Construction Idea: The model predicts the adjustment of constituent stocks in major indices (CSI 300 and CSI 500) based on the index compilation rules published by the China Securities Index Company[12][13][15] - Model Construction Process: - Data Reference: The model uses trading and financial data from May 1 of the previous year to April 30 of the current year for mid-year reviews, and from November 1 of the previous year to October 31 of the current year for year-end reviews[12] - Adjustment Date: Adjustments occur on the trading day following the second Friday of June and December each year[12] - Sample Space: - Excludes ST, *ST stocks, and delisted securities - Includes STAR Market and ChiNext stocks listed for over a year, or those with daily average market capitalization ranking in the top 30 since listing[12] - Selection Method: - For CSI 300: Securities are ranked by daily average trading volume, with the bottom 50% excluded, and then ranked by daily average market capitalization, selecting the top 300[12] - For CSI 500: Excludes CSI 300 constituents and the top 300 by market capitalization, then ranks by trading volume and market capitalization, selecting the top 500[12] - Buffer Zone Rules: - CSI 300: New samples ranked in the top 240 are prioritized for inclusion, while existing samples ranked in the top 360 are prioritized for retention[12] - CSI 500: New samples ranked in the top 400 are prioritized for inclusion, while existing samples ranked in the top 600 are prioritized for retention[12] 2. Model Name: Event Return Analysis Model - Model Construction Idea: This model analyzes the relationship between index sample adjustments and the cumulative excess returns of the adjusted stocks before and after the adjustment date[24] - Model Construction Process: - Collects data on all adjustment cycles since 2010 for CSI 300 and CSI 500 indices[24] - Calculates the cumulative excess returns of adjusted stocks relative to the index over a 60-day window before and after the adjustment date[24] - Model Evaluation: The model highlights that pre-adjustment event returns are more significant for investment strategies, as investors can trade in advance based on predictions or announcements[24] --- Model Backtesting Results 1. Index Adjustment Prediction Model - CSI 300 Index: Predicted 7 constituent stock adjustments, with additions primarily in the transportation and defense industries, and removals concentrated in the electrical equipment industry[13][14] - CSI 500 Index: Predicted 49 constituent stock adjustments, with additions mainly in the electronics, electrical equipment, and computer industries, and removals concentrated in the electrical equipment and pharmaceutical industries[15][16][23] 2. Event Return Analysis Model - CSI 300 Index: Stocks removed from the index showed negative excess returns before the adjustment date, while no significant conclusions were drawn for added stocks[5][24] - CSI 500 Index: Stocks added to the index exhibited positive excess returns before the adjustment date, while removed stocks showed negative excess returns[5][24]