Markov Decision Processes Prediction On Stock Price

Markov chain is an extensively applied stochastic process model which is to quantitatively analysis a system transferring from one state to another, many dynamic systematic problems in the economic and social phenomenon can be described by Markov chain. The paper first introduces several basic methods of Markov chain and some more important branches of Markov chain especially the birth-death process issue and its structure. The main body of the paper is to analysis the stochastic mathematic model applied in the economic and management field which is based on the transition probability, the "Markov property" test of random variable series, the Markov chain prediction method based on absolute distribution and the Markov chain prediction method based on probability summation. Through the practice, it has been proven that the model is practical and can be applied in the economic and management field, which can optimize the long-term benefit.The paper discusses some methods used in prediction on stock price, and a new method using Markov Decision Processes prediction on stock price is proposed , the efficiency of the method being explained with examples. Finally, the comparison between this new method and other traditional methods is proposed, and the characteristics of this method are summarized.The article has set forth Markov Decision Processes prediction on stock price from five parts. Chapter I has started from analyzing studied current situation and the developing trend at home and abroad, have drawn forth the main body of article , the main studying content and the purpose significance studying. Chapter II has introduced the knowledge affecting the share price factor and some relevant knowledge. Chapter III has introduced relevant concept the random process and Markov Decision Processes theory. This chapter has introduced the relevance concept of Markov chain and the Kolmogorov theorem, ergodicity and Smooth distribution. Has introduced the Markov Decision-making model, the model is composed of five parts of the following: State set S , reward function R , criterion function V making policy set A, transferring probability P .And the model based on the definition of the optimal value functions, and gives Markov decision-making matrix algorithm. Chapter IV has studied Markov decision model price forecast algorithm. This chapter has completed on the transfer probability matrix calculation, the transfer probability matrix Markov test from the algorithm, and the state-set of algorithms and decision-making. Chapter V has predicted the outcome of the case and concluded evaluation. From the results of our forecast, we can see that the overall trend of the stock is rising, rose sharply and down slightly. This is our forecast period to basically consistent with the stock market. Use Markov decision on prediction stock prices, the results from the prediction are more accurate and error rate is in more behind the transaction, which is a reflection of the unpredictable changes in the stock market and difficult predictability.