PROJECT: A Stock Investment Trading System Based on Markov Chains

This project is my undergraduate thesis completed in Tsinghua University, Beijng, China, under the supervision of Prof. Li Xia.

The stock market always goes with high risks because of its fluctuating and uncertain characteristics. In order to provide the reference for users' rational investment, we hope to develop a new stock investment trading system. The system can not only analyze the situation of stocks, but also complete the program trading on the automated trading platform according to real-time market data.

In this paper, a stochastic optimization algorithm based on Markov chains was proposed. The stock price was decomposed into Markov sequence by time series to establish the Markov chain model. With the help of stochastic optimization algorithms based on policy iteration, a rolling optimization of portfolio was achieved. Eventually, a set of investment strategies based on state feedback can be determined. An empirical study on the Chinese A-Share Market was used, whose results show that higher efficiency and return of investment can be achieved under the new model and algorithms.

Meanwhile, a new type of stock investment trading system was designed. According to real-time stock market data, the system can complete the program trading on the automated trading platform intelligently. In addition, the system is user-friendly and easy to operate, with flexible design patterns and stable performance.

An explorative method was proposed, in which a Markov chain model of stocks portfolio was constructed through historical data. On the basic of instantaneity and accuracy, policy iteration made the computation efficiency greatly improved. Since random factors were taken into consideration, the method provided investors with more opportunities and higher income.