Abstract

We analyze the logs of an online mathematics game tournament, played simultaneously by thousands of students. Nearly 10,000 students, coming from 356 schools from all regions in Chile, registered to the fourth tournament instance. The children play in teams of 12 students from the same class, and send their personal bets to a central server every 2 minutes. Each competition lasts about one clock hour and takes place within school hours. Students are pre-registered and trained by their school teacher. The teacher is responsible for reviewing curriculum contents useful for improving performance at the game and coaches students participating in trial tournaments taking place a few weeks before the national tournament. All bets are recorded in a database that enables us to analyze later the sequence of bets made by each student. Using cluster analysis with this information, we have identified three types of players, each with a well-defined strategy. © 2011 Springer-Verlag.