| In the financial market, there are usually two methods of data analysis: the technical analysis and the fundamental analysis. In this thesis, we use the latter method and pay attention to the fundamental variables.;In the stock market, there is a large amount of data and fundamental variables associated with it. In our approach, seven variables are selected and prove that no linear correlation exists between any two of them. Then, fundamental variable pairs are constructed from these fundamental variables and their visualization figures are built. From the results of the visualization figures, a statistical method is used to find the sub-areas with high frequency. With the help of these high frequency sub-areas, we observe all the visualization figures for every variable pair. Observations show that a sub-area in which the data samples always have a good or bad return from any variable pair can not be found. However, we can find sub-areas in which the return of stocks is better than the average return of all stocks. Based on these sub-areas, a set of rules is derived using the training data sets from 1993 to 1998. These rules are tested on the data set from 1999 to 2003. Most of the rules perform well because, except for rule 4 in year 2001, the average of returns of the rules are better than the average of returns of all stocks in S&P 500. |
