A Stock Selection Method for ETF-Abstracted-Feature Based Trading Strategy

In the current financial market, ETF is a growing investment approach, and sector index-based ETFs are often the investment products that build a bridge between broad market data and individual stocks. Inspired by this, we tried to build a trading strategy based on a machine learning model that has the characteristics of an exponential ETF, that is, simply use market factor data, and use SVR machine learning models to predict and formulate investment plans for suitable stocks to obtain returns. Using this strategy as a basis, we select stocks in the stock market, evaluate the stocks linked to the broad market using various correlation, sliding window, and multi-objective optimization measures on the time series of stocks and the broad market, evaluate 10 stocks that are close to the market index, and verify their suitability using the above strategy.