A multi-stage race model with strategic interaction: An agent-based simulation approach

We consider a multi-stage race characterized by technological uncertainty and winner-take-all stakes. A number of studies have found that players in such a race tend to behave strategically by adjusting their effort levels according to their relative ranks. The aim of the present study was to examine the dynamics and the results of such interaction. For this purpose, we employed an agent-based computational approach that enabled us to better understand the dynamics of a race, particularly those in which players exhibit heterogeneous behaviors. The results found that strategic interaction decreases the total cost but also results in delaying the completion of a race. Also, it was discovered, larger budgets rather than strategy differences have more significant effects on the probability of players winning a race. Finally, our study suggested that early preemption is an effective strategy that can improve the probability of winning races substantially.