Sports are competitions that are characterized by the accumulation of points, and the team with the highest total score wins. The scoring system is defined by the sport, and it can be based on abstract quantities or more natural measures such as distance or duration. Each competing athlete accrues scores based on the sport’s scoring system, and some events have a score limit that must not be exceeded, such as the high jump or tennis.

The fraction of all events won by a team, cc, provides a simple measure of balance for each game in any sport. Smoothed distributions for this quantity, which are shown in Figure 1, suggest that most games have a close to balanced scoring. However, the occurrence of long gaps between events is also common in some sports.

For the most popular spectator sports, CFB, NFL and NHL, there is very good agreement between the Poisson model and the empirical distribution of events per game. This suggests that these games are dominated by differences in latent team skill, which result in a fairly predictable pattern of event outcomes. Conversely, NBA games have less predictability and take longer to reach a maximum AUC.

Developers build applications that depend on sports scores, such as fantasy leagues and sports betting. They need access to current and historical scores, including live scoring. A sports scores API is a structure that provides this information to developers. These structures may be free for fan-based or commercial applications, and they can either update in real time (real-time APIs) or after the end of a match or game.