The Study Of Stock Investment Strategy Based On Piotroski Strategy And ARIMA-SVR Model

The reform of economic system in our country is deepening continuously. At the same time, the stock market has been consciously improved and developed. The stock market is an essential part of Chinese securities industry and financial industry, and is an area full of opportunities and challenges. Stock investment is a mean of investment with high profit and high risk. Before selecting which stock to invest, investors should have a set of scientific and effective method to find a balance between return and risk so that they can achieve their investment objectives. Therefore, the analysis and prediction about stock investment are of great theoretical significance and practical value.At present, the study of stock investment is based on financial data or trade data only. And this paper presents an investment strategy combining analysis of basic situation of companies with data mining technology. At first, build a strategy to select stock using financial indicators to pick up the stocks with good investment prospects.Then predict the closing price of stock using the technology of data mining, and gain the prediction result of a stock price trend. After that, by comparing the prediction gains and a preset threshold value of gains, so that we can know when is better time to buy and sell stocks and can obtain higher profit.Firstly, use Piotroski strategy to select 9 respective financial indicators, score them with their own criteria and calculate the final score of each stock. Then select the stocks with high investment value based on the final score. In this paper, we take these stocks as study object, and we will find the right time to buy and sell them.Secondly, predict the closing price of stock using time series model(ARIMA) and support vector machine regression(SVR) respectively. After that, build ARIMA-SVR model combining ARIMA model and SVR model. And then predict the price interval with scalar method using the result of hybrid model. At last, make a decision about whether still hold the stock based on prediction gains so that we can know the best to buy and sell these stocks.Experimental results show that, Piotroski strategy is suitable for Chinese A Stock Market. The rate of return of portfolio with high score are higher than those with low score and the maximum retracements are lower. The performance of closing price prediction using ARIMA-SVR hybrid model about China Eastern Airlines and Sichuan Road & Bridge, which one of them is stock in Shanghai and Shenzhen 300 index and the other is not, is better than that using ARIMA model and SVR model respectively. And the interval prediction based on scalar method has good performance. At last, design a comprehensive experiment to verify the effectiveness of combination of qualitative method and quantitative method. The last experiment shows that the strategy based on Piotroski strategy and ARIMA-SVR model is effective. With this strategy, we can gain some profit and make good control of risk.