Information quantities in financial markets

We propose a new method to characterize real stock price movements and its application to the portfolio selection problem. It is shown that the difference between a real price process and an ideal random walk, like a Brownian motion, can be detected by using information quantities. This fact infers that there exists a time serial correlation in real stock price movements on financial markets. For example the Bubble Break point in the Japanese market in 1990 is characterized by our method. Classifications for several stocks are done through the similarities of their serial correlations and we apply our classifications to the portfolio selection problem.