Research On Alpha Arbitrage Strategy Based On LSTM

With the launch of the CSI 300 stock index futures,China's stock index futures market has developed tremendously,and quantitative investment technology has also developed rapidly in China.Alpha hedge arbitrage is a trading strategy based on the reverse operation of the stock spot market and stock index futures to build a portfolio,which can effectively avoid the systemic risks of the securities market and obtain stable returns.This paper redefines the Alpha value of securities based on the CAPM model,and attempts to use the machine learning model to construct the Alpha arbitrage strategy portfolio.The effectiveness of the Alpha arbitrage strategy depends on two aspects:First,whether it can effectively avoid the hedge.The market risk of the securities market,and the second is whether the investment strategy after hedging can achieve a higher rate of return.In the empirical part of this paper,we will select different investment portfolios based on the Long Short Term Memory(LSTM)model and the Linear regression model,and back-test their effectiveness,and evaluate them according to different methods of arbitrage effects.The perspective confirms the feasibility of machine learning in the Alpha arbitrage strategy,and hopes to help investors make investment decisions.