Research On Stock Quantitative Timing Strategy Based On Support Vector Machine

Quantitative investment strategy is a hot issue in the field of stock investment.The most important thing in quantitative investment is timing,so timely grasping the timing of buying and selling is the key to maximizing profits while controlling risks.Based on this,this paper chooses to study the quantitative timing investment strategy of stocks.Since the key of timing investment strategy lies in the determination of trading point,the key of quantitative timing lies in the prediction of turning point of stock price trend.Support vector machines have unique advantages in solving small sample,nonlinear and high-dimensional pattern recognition,which exactly meet the requirements of selected stock data.Therefore,this paper chooses SVM for quantitative timing investment strategy research.In this paper,a total of 1000 stock daily data of the top 8 stocks in the sse 50 index from January 5,2015 to February 12,2019.12 were selected,and the ratio of 7:3 was adopted.The stock data from January 5,2015 to November 15,2017 were used as the training set,with a total of 700 samples.The stock data from November 16,2017 to February 12,2019 is the test set,with a total of 300 samples.Gaussian kernel function with minimum fitting error was selected to construct SVM timing model,and the optimal parameter model of eight stocks was obtained.According to the mean square error,three stocks with poor simulation effect were eliminated,and the remaining five stocks were used as the stock pool for analysis.Based on the establishment of the timing model,three investment strategies,namely,SVM's closing price timing strategy,rolling SVM's closing price timing strategy and rolling SVM's MACD timing strategy were respectively adopted for comparative analysis.The analysis results show that SVM's closing price timing strategy is inferior to the buy-and-hold strategy.The other two strategies are better than the buy-and-hold strategy.From the perspective of risk and stability,the rolling SVM timing strategy for predicting closing price is superior to the MACD timing strategy.Regardless of individual stocks or portfolios,the optimal quantitative timing strategy has randomness in the effect of returns,but the risk control is relatively stable,and the overall performance is better than the buy-and-hold strategy.Research shows that the quantitative timing strategy is effective.The innovation of this article is in comprehensive multi-index shares based on support vector machine was used to predict,overcome the practice under the conditions are not consistent with traditional methods are limited by the drawbacks of the prediction,and on the basis of quantitative,from stocks to the portfolio,to verify the support vector machine to quantify the effectiveness of the timing of investment strategy and provides a new SVM application perspective.The deficiency is that the strategic backtesting in the empirical research is not a complete quantitative investment process in the strict sense,but a "strategy test" process of quantitative investment.It does not include a strict stop loss mechanism and a more complete backtesting program,and does not use a large amount of data for a comprehensive backtesting.The strategy backtest scheme needs to be further improved in the future.