Based On Event-driven Quantitative Stock Selection Strategy Research

With the advancement of computers and the development of people's investment ideas,compared with the traditional way of relying on personal observation and operation for qualitative investment,quantitative investment has become an increasingly popular investment method due to its own science and rationality.It is widely used in the investment field.Event-driven investment strategy,as a supplementary strategy for quantitative investment,is to find the impact on stock prices after a specific event occurs.As one of the mainstream strategies,the event-driven investment strategy has a bright market performance in the investment community.At present,most of the common event-driven strategies in the market are for micro-events(individual stock events),such as the increase and decrease of executives,equity incentives,reorganizations and mergers,and restrictions on the sale of restricted shares Waiting for individual stock events.But there is no event-driven trading strategy for macro events and meso events.This article selects the less researched policy event investment theme "Belt and Road" as the research object of driving events,and creatively applies the machine learning stock selection model to the event-driven investment strategy to construct a package of event-based theme investment.Stock portfolio.First,analyze the sample events selected in this paper.Through the event research method,taking the "Belt and Road" events from 2013 to 2018 as a sample,to study whether the stock price of listed companies has changed due to the occurrence of the event when the relevant subject events occur,and the problem of abnormal reward CAR appears,and use The method of statistical analysis tests the significance of the event.The research results show that when the analysis of the “Belt and Road” theme event occurs,the relevant stocks will have a significantly positive excess rate of return,which verifies that the company 's stock price of the relevant event can obtain a guess that exceeds the market index 's rate of return in the short term.Then,based on the specific “One Belt,One Road” theme event that occurred on March 25,2019 as a driving event sample,a machine learning model for recognition of three classification patterns of support vector machine,random forest,and neural network was constructed,and the model was optimized by different methods.Then,by comparing and analyzing the hit rate of the three models on the ups and downs classification and the backtesting return of the stock portfolio,the BP neural network model optimized by genetic algorithm is finally determined as the quantitative stock selection model in this paper.The beneficiary company invests and builds a package of event-themed investment stocks.Finally,by summarizing the previous two research results,an event-driven trading strategy plan is constructed,and the seven “Belt and Road” theme events that occurred in 2019 are used as event samples for simulated transactions,and historical data is brought in for backtesting.Three dimensions of risk and comprehensive performance are used to construct a variety of evaluation indicators to verify the effectiveness of the final construction of trading strategies.The calculation results show that the strategic investment portfolio has better returns,higher risks,and better overall performance than the broader market index.The event-driven trading strategy finally constructed in this paper is a short-term investment strategy.The research framework and results of the event-driven strategy can be used as an auxiliary tool for investors to make investment decisions,providing investors with a way to study event-driven strategies.New investment ideas and options.