Popular location-based games (LBGs) such as Pokémon GO have been downloaded hundreds of millions of times and have been shown to have a positive impact on mild exercise and social well-being of their players. Several currently popular LBGs introduce a gamified implementation of territorial conflict, where players are divided into teams that battle for the ownership of geographically distributed points of interest. We investigate how social factors and territoriality influence playing intensity in the context of Pokémon GO. Using reasoning from social identity theory, we propose a structural model connecting territoriality, sociality and playing intensity. To test the model, we analyze data collected from a global sample of Pokémon GO players (N=515). Our results show social self-efficacy, territorial self-efficacy and altruism to influence players' identification with their team. Team identification in turn predicts both in-game territorial control behavior and playing intensity.