There are three major approaches to studying competition in dynamic markets: theoretical studies, field studies (case studies), and laboratory experiments (which include computer simulations). We believe any single approach cannot provide a full picture or a complete solution. Therefore, in this study we used a combination of all the three to attack the problem. We first investigated how corporations competed in oligopolistic markets. We then used the theoretical approach to develop prediction schemes to predict the behavior of competitors. Finally we designed and implemented an artificial market with personal computers and local area networks to study how subjects learn, develop their strategies, and react to market changes. Such a combination led us to new insights impossible to reach by any single approach. Based on these findings, we propose guidelines for a decision model which can be used as the foundation for a decision support system to support competition strategy selection.