Playing styles in starcraft

Understanding playing styles in video games may assist game designers to create entertaining game content for different players. Numerous factors determine how distinct players may approach a game, e.g., player preference, game literacy, and player motivation. In Real-Time Strategy (RTS) games, for understanding player behaviour, it is particularly important to model player preferences and adopted game strategies. As such, as a continuation of previous work, the present paper investigates how distinct human players approach the popular StarCraft game in terms of preferences and strategies that may be inferred from game observations. In particular, we investigate how distinct match-types relate to the different playable races in the game. To this end, we propose features that reflect playing style, and uncover unique variations in playing style by means of Principal Component Analysis (PCA). Findings of experiments with clustering player styles of StarCraft players reveal that playing styles can indeed be distinguished in different match types. While one may expect playing style to affect the chance of winning, results reveal win probability is not significantly affected by player style, but the length of matches is.