AbstractA standard view of elections is that parties should choose moderate platforms to maximize their probability of winning. However, some parties embrace more extreme positions. These parties often feature an energized activist group that is optimistic about the party's probability of winning – contradicting the standard view. We build and analyze a game‐theoretic model of electoral competition that simultaneously explains the prevalence of extreme platforms and activists' biased beliefs. We show that the activists' interests and role within the campaign induce them to optimally misperceive the interests of the electorate because such a bias shifts the party's equilibrium policy platform closer to the activist faction's ideal point. We also find that even though the more optimistic an activist faction, the more it learns from electoral defeat, the divergence between activist and elite beliefs may worsen after a loss.