金融学院知识库

    Quantitative investment is a new type of investment technology and investment way. It establishes a systematic process of thinking and decision-making, including strategic trading, risk control and many other core contents, in addition to the integration of artificial intelligence technology, but also to promote the rapid development of quantitative investment. The application of artificial intelligence in quantitative investment mainly shows that the computer program takes the place of man to screen and clean the huge amount of data, and uses the learned law to train the model, so as to establish a more rational and scientific investment trading strategy, this to a large extent overcome the investor's greed, fear, fluke psychology. With the rapid development of information technology, China's capital market is also entering a period of rapid development of quantitative investment. Under this trend, the new investment strategy which combines artificial intelligence with quantitative investment has certain research value and development space.

    Using Efficient-market hypothesis, behavioral finance theory, the capital asset pricing model, and markowitz's portfolio theory, an integrated analysis-Support Vector Machine (ica-support Vector Machine) , which combines IC analysis with Support Vector Machine algorithm, is creatively constructed, and the portfolio is constructed according to this model. This paper is divided into three parts to build the composite model and test the strategy. Firstly, 112 factors covering fundamental, technical and behavioral financial aspects were selected as the candidate database of factors, secondly, the daily data were acquired for cleaning, and the factors after cleaning were selected by IC test and return rate test Finally, the optimal factor is used as the input index of the Support vector machine to train the model, carry out the strategy test, and analyze the feasibility of the model and the applicability of the investors according to the result of the strategy.

    The model is based on a sample of hs300 Index stocks, measured back from June 1,2014, to June 1,2019, against a benchmark of the earnings of the hs300 Index index and the earnings of a strategy based on the principle of random selection, the goal is to explore the effectiveness of an investment strategy that combines the methods of IC analysis and Support vector machine. The empirical results show that: First, the annualized return of the Investment Strategy based on the compound stock selection model established by the radial basis function is 11.52% , which exceeds the return of the hs300 Index index by 1.90% ; Second, the annualized return of the Investment Strategy based on the composite stock selection model established by Sigmoid function is 12.03% , which exceeds the return of the hs300 Index index by 5.76% , and the Sharpe ratio is increased from 31.5% to 33.14% Thirdly, the returns of the compound stock selection model far exceed the returns of the random
stock selection model. Therefore, the compound stock selection model designed in this paper, which combines IC analysis with Support vector machine algorithm, has certain generalization ability in the construction of Quantitative Investment Strategy, and has a strong practical significance for the investors in the capital market.