Design And Implementation Of Stock Market Quantitative Trading System Based On Deep Learning

China's capital market is a market with rapid development and broad prospects.The integration of big data,blockchain,artificial intelligence and other fintech and securities industries is deepening,and the capital market environment is evolving towards intelligence,customization and virtualization.Quantitative trading refers to the investment behavior in the open capital market to avoid risks and increase the value of assets through the planned buying and selling of a series of assets.Artificial intelligence technology is involved in the analysis and judgment of the securities market,the prediction of stock price fluctuation and the simulation of stock trading and so on,which have developed rapidly,providing a new power and engine for the leapfroging-forward development of the traditional quantitative trading field.This thesis firstly reviews the successful cases of combining deep learning technology with quantitative trading in the stock market in recent years,sorts out the main applications of deep learning in the field of quantitative trading,and summarizes the advantages and limitations of current deep learning technology in this field.Then the thesis expounds the related problems of quantitative trading from the level of algorithm,abstracts the mathematics of portfolio management and quantitative trading strategy,and designs and implements the quantitative trading system combining the stock market prediction and investment decision.From the perspective of classic portfolio management,this system operates a series of stock assets,predicts stock price fluctuation by using the deep time series model,and controls trading decision by using the deep reinforcement learning algorithm,realizing the automatic trading system driven by algorithm and data.Finally,combining with the basic principles of VNPY quantitative trading framework,it explains how to embed the deep learning algorithm into the quantitative trading platform for real trading,and evaluates the investment process controlled by the algorithm using the simulated trading environment.Through simulation experiments,the quantitative trading system of stock market based on deep learning implemented in this thesis can select the portfolio with the optimal return among a series of stock assets.By means of diversification of investment risks,exploration of trading opportunities,and control of trading costs,the algorithm achieves excess return in long-term investment of stock market.