THE EVALUATION OF PORTFOLIO PERFORMANCE BY USING DATA MINING PROCESS AND AN APPLICATION ISE STOCK MARKET
Assistant Prof. Dr. Engin KÜÇÜKSILLE, Prof. Dr. Durmus ACAR

Abstract
In this study, the concept of portfolio has been examined and different portfolios have been formed by using data mining process and Sharpe, Treynor and Jensen portfolio performance measures from 122 different stocks which were traded in Istanbul Stock Exchange (ISE) in the period of 1995 – 2007/06 permanently. Afterwards, the performances of the portfolios’ formed using data mining process has been compared with the performance of the market (ISE National 100 Indices) in the same period and the period of 2007/07 – 2008/12. In the base of this study, it has been assumed that a portfolio could be formed by using data mining process and the developed model has been applied to ISE Stock Market. In the application process, different portfolios have been formed by using genetic algorithm and the average monthly return of 122 different companies’ stocks in the period of 1995 – 2007/06. The result of the application is that Sharpe, Treynor and Jensen performances of different portfolios were higher than the market’s. The main reason for this situation is that the quotation realized a profit under the risk free rate of interest. However, the same portfolios showed lower performance than the market in the period of 2007/07 – 2008/12.

Full Text: PDF