A game theory competitive intelligence solution stimulated from a Stackelberg game: A three players scenario

Abstract

Competitive Intelligence (CI) involves systematically gathering information about the competitive landscape to anticipate rivals’ decisions and secure a strategic advantage. As academic interest in CI grows, the existing literature reflects a diverse body of knowledge. However, few works address the practical aspects of CI comprehensively, presenting viable solutions to its challenges. The motivation behind this paper is not only the scarcity of practical studies on CI but also the importance of CI and its benefits within companies, including improved decision-making and its positive impact on performance and sustainable growth. This study aims to contribute to the literature by proposing a CI solution grounded in game theory. We assert that applying game theory to CI is both intuitive and natural, given that the competitive environment readily lends itself to a game model. Viewing competitors as players, each striving to maximize profit, aligns seamlessly with game theory principles. We chose to consider a scenario with one initiator and multiple followers and model it as a Stackelberg game within the extensive form game framework. The literature review showed that the computational complexity of Stackelberg Equilibrium becomes NP-hard in most game classes with more than two players. This challenge encouraged us to contribute to the literature and focus on a scenario with two followers (three players), modeling the game in the extensive game form. To find a solution to this challenge, we used a modified version of the backward induction algorithm. To assess our solution’s efficacy, we conduct tests on 26 games (26 data instances), comparing results with those obtained using CPLEX and Decision Tree, and MiniMax methods. The ensuing pairwise sign test establishes the competitiveness of our outcomes. This research significantly contributes to the CI literature, offering a game theory-based solution and demonstrating its practical applicability through evaluation.