Research On Trading Strategies From The Perspective Of Complex Financial Data In China's Stock Market

The study of trading strategy is not only a practical problem for investors,but also a theoretical problem of analyzing the investment behavior in-depth,predicting of the stock market trends scientifically,and making the stock market policy rationally.With the fantastic development of China's stock market,the great advances in computer science and information technology,the maturity of electronic trading,and the popularization of the Internet,massive and complex financial data is produced everyday.The financial data is unprecedentedly high frequency,increasingly strong correlation and with multi-sources.These new charactors in data unprecedentedly impact the classical theoreties of market efficiency hypothesis,rational person hypothesis,and complete information hypothesis,which are widely considered as the basis of trading strategies.It also inevitably brings great challenges to the research of trading strategy.How to optimize the tradying strategy for the special market environment under the background of the information age,reexamine the traditional trading strategy from the perspective of complex financial data,grasp the overall characteristics of the market and even build a new trading strategy framework and model? Specifically,what changes will be made to the traditional trading strategy in the stock market's statistical data from the typ ical medium low frequency data to the high frequency data? As a complex dynamic system,how does the correlation between the trading data and the trading data affect the trading strategy? Besides,what are the important external news data that are closely related to the stock market? Can we use the multi-sourced financial data to improve existing trading strategies in building the framework and model of automatic trading?To deal with the above problems and challenges,we study the trading strategies of the Chinese stock market,based on the classic theories and technical methods,such as asset pricing,cointegration theory,complex network,and information extraction.We study the transmission of complex financial data from three aspects of the high frequency,relevance and multisource of financial data.Our research is focused on the impact of unified trading strategy,the changing characteristics of the overall market,and how to effectively use the financial data to enhance the performance of trading strategies.Our work and innovations of this dissertation can be addressed as follows:First,to remove the limitation of low updated frequency of anomalies in stock market,we propose a strategy based on daily frequency trading information and moving average method to dynamically adjust the anomaly index.Our method can effectively measure the price trend in the sample period,and use the more abundant price information in the period of the index adjustment to avoid the problem of neglecting the important details in the portfolio of market anomalies based on low frequency data.The empirical study shows that,unlike the mature stock markets,the price trend of China's stock market is short term.Besides,the performance of portfolios based on the adjusted market anomalies can be significantly improved,especially for some of the weaker ones.In addition,the investor sentiment and information uncertainty show significant impact on the anomalies of China's stock market.Second,to analyze the impact of high frequency trading data on asset price trend and volatility,we construct pairs trading strategy for high frequency trading data based on the two-stage method.Specifically,using the minute frequency data of margin stocks in China's stock market,we construct the pairs trading strategy based on the cointegration theory and correlation method.Then,we examine the performance of the strategy in shock period,bull market and bear market measured by annual return,sharp ratio and maximum retracement.The study shows that the pairs trading strategy is superior to the benchmark strategy regardless of the transaction costs.The experimental comparison under different market quotations shows that the pairs trading as a market neutral strategy can effectively avoid the risk of market decline.In addition,the number of pairs trading in the sample period is closely related to the market quotations and the industry.Third,in view of the impact of the correlations of the stock prices on the performance of the trading strategy,we construct a cointegration network based on the complex network theory,and analyze the static and dynamic characteristics of the network.The empirical results show that cointegration network is a highly correlated small world network,indicating that the underlying stock prices are highly correlated.Besides,the stocks of the same industry or those involved in the MTSS(Margin Trading and Short Selling)programme are more likely to form cointegration relationships.In addition,the evolution of the network edge proves that the pairs trading strategy is market neutral and a large number of relationships are not stable and robust.Fourth,in view of the impact of the MTSS programme on investor behavior in China's stock market,we make use of the long-term stock price relationship data of multiple MTSS stocks,and put forward a new perspective of fin ancing margin as information uncertainty.Then,we analyze its impact on the policy of exchange.Experimental results show that the stock portfolios not included in MTSS tend to show greater information uncertainty.The MA of the portfolios of non-MTSS indicates that the earnings of average return and risk adjustment return are significantly higher than the benchmark buying and holding(BH)strategy.Unlike the mature stock markets,the market state of China's stock market has a significant impact on the performance of the technical trading strategy.Fifth,to use multi-sourced financial data to enhance the performance of the stock automatic trading system,we use the financial news data and the stock price data to analyze the impact of the multi-sourced financial data on the technology transaction.We construct a stock automatic trading framework which combines financial news signals and technical trading signals.Moreover,a stock trading strategy optimization model is constructed.We first design the corresponding model to extract the financial events from Chinese,then analyze the relationship between the financial news events and the stock price from the statisti cal point.Then,we put forward an optimization model of the stock trading strategy based on th e genetic programming algorithm,which considers the news event signal.Experimental results of the underlying stocks listed in the China Securities Index 300 show that different types of news events have significant but different effects on the stock mark et.The integration of news event information and traditional technical strategies can effectively improve the return of automatic trading.For the challenges posed by the complex financial data to the traditional trading strategy,we analyze the impact of financial data on the trading strategy from the perspective of high frequency,correlation and multi sources.This study may bring new ideas for the research on the trading strategy under the background of big financial data,and also provide useful refere nce for investors in China's stock market to choose trading strategies,build new investment models and achieve higher and more robust excess returns.