The goal of this paper was to introduce a stock ranking method and to show how this method could be incorporated into stock trading systems. The proposed method analyses a large number of securities and rankes them according relative change in price during the defined interval of time. Then the values of ranks arc being normalized and assume the values from -1 to +1. The securities with ranking values close to -1 are good candidates for an investor portfolio, because these securities historically had shown a statistically significant price increase by the following days. The proposed method was tested using experimental data from US security markets. Two groups of securities (30 companies from Dow Jones Industrial Index and 292 companies from SP500 Index) were tested during time interval between 1992-2002. First investigations had shown efficiency of the proposed method. The stock’s rank indicator exhibits a mean reverting behavior. For the stocks with strong negative rank a positive stock price change in mean is followed. Furthermore, a negative stock price change on average is followed for the high ranked stocks. So these results contradict the statements of the effective market hypothesis and motivate the creating of stock trading systems, based on stock’s price rank. In this paper we analyze the usefulness of this method while applying it in a virtual stock trading system. The trading simulation is executed using historical data from USA stock market (1992-2002). The trading system has given significantly higher return. relative the benchmark (SP500 index). However, there are some practical problems that must be overcome if we arc going to apply the stock’s rank method in real stock trading process. The most important of these is the taxes rate during stock trading operations. In the paper we analyze the influence of taxes rate on the systems profit and we give some suggestions how to make this system more suitable for the practical applications.