Abstract
This paper aims to formulate a model of stock classification to facilitate investors making investment decisions related to equity stock selection by two dimensions-performance shift and stock price evaluation. The former is measured by the Malmquist productivity index based upon Data Envelopment Analysis (DEA). The latter is measured by Range adjusted Measure (DEA-RAM) in DEA. With these two dimensions, the model of stock classification can be partitioned into four grids, which are classified as Value, Monitor, Speculative and Avoidance, respectively. The four stock classifications have different implications. For example, value stocks have potential to achieve higher stock returns for the next year than that of the others, while the avoidance stocks which the model designates as high risk, will likely have inferior performance to the others in the coming year. The empirical results indicate that the proposed stock classification model can predict stock returns, enabling both institutional and individual investors to reduce the risks associated with equity investments. The results of this study further demonstrate that DEA can be extremely effective in devising stock trading strategies.