金工ETF点评：跨境ETF单日净流入22.21亿元，美护、银行、军工拥挤变幅较大

Quantitative Models and Construction Methods 1. Model Name: Industry Crowding Monitoring Model Model Construction Idea: This model is designed to monitor the crowding levels of Shenwan First-Level Industry Indices on a daily basis, identifying industries with high or low crowding levels and tracking changes in crowding over time[3] Model Construction Process: The model calculates crowding levels for each industry index based on specific metrics (not detailed in the report) and ranks them accordingly. It provides daily updates on crowding levels and highlights industries with significant changes in crowding[3] Model Evaluation: The model effectively identifies industries with high or low crowding levels and tracks significant changes, providing actionable insights for investors[3] 2. Model Name: Premium Rate Z-Score Model Model Construction Idea: This model identifies potential arbitrage opportunities in ETF products by calculating the Z-score of premium rates for ETFs over a rolling window[4] Model Construction Process: - The premium rate for each ETF is calculated as the difference between the ETF's market price and its net asset value (NAV), divided by the NAV - The Z-score is then computed as: $ Z = \frac{(Premium\ Rate - Mean\ Premium\ Rate)}{Standard\ Deviation\ of\ Premium\ Rate} $ where the mean and standard deviation are calculated over a rolling window[4] Model Evaluation: The model is useful for identifying ETFs with significant deviations from their historical premium rates, which may indicate potential arbitrage opportunities. However, it also highlights the need to be cautious of potential price corrections[4] --- Model Backtesting Results 1. Industry Crowding Monitoring Model: No specific numerical backtesting results were provided in the report[3] 2. Premium Rate Z-Score Model: No specific numerical backtesting results were provided in the report[4] --- Quantitative Factors and Construction Methods No specific quantitative factors were detailed in the report --- Factor Backtesting Results No specific quantitative factor backtesting results were provided in the report