A team-based competitive environment is a complex multi-agent environment, in which agents are required to coordinate among each other not only to enhance their collective behavior, but also to compete with other teams. Based on the Minority Game (MG) model, we have provided a strategy for assisting a team to win in RoboCup and in a more general environment, i.e., DynaGrid. In this paper, we aim to examine the effectiveness of the MG-based strategy for DynaGrid in more complete situations, e.g., both regular and irregular situations. We also propose a method for measuring the irregular complexity of a dynamic environment. Thus we are able to quantitatively figure out the typical situations in which the MG strategy works. Through experimental validation, we have found: (1) the MG strategy can generally speaking help a team of agents to enhance their competitiveness in a dynamically-changing environment, e.g., the target object is in a nonlinear or irregular motion; (2) the MG strategy does not have an edge over a commonly-used greedy strategy under some specific circumstances where a learning window is not large enough.