How our ratings work

What does the rating mean?

A team's offensive rating (oRtg) is the number of points above league average we expect them to score against a league average defense. A team's defensive rating (dRtg) is the number of points below league average we expect them to allow against a league average offense.

Offensive rating + defensive rating = net rating (nRtg).

Net rating is the core metric of Zero Sum Labs and represents how many points above league average any given team is per game.

How are the ratings calculated?

We use a zero-sum power rating system. Which means the ratings within each league always add up to zero. Positive ratings are balanced by negative ratings elsewhere in the league, making zero the league-average rating.

We begin by assuming every team has an average offense and defense, then use those ratings to predict every game of the season. Teams that outperform those predictions gain rating points, while teams that underperform lose them. We repeat this process until the ratings provide the closest mathematical fit to the season's results.

How are game predictions calculated?

Our score prediction formula is simple:

Expected home score = League-average + Home oRtg − Away dRtg + ½ home advantage.

Expected away score = League-average + Away oRtg − Home dRtg - ½ home advantage.

How often are the ratings updated?

Our ratings are recalculated daily at 4:00 a.m. ET to incorporate the latest available game results.

What does the model not account for?

Ratings are based strictly on team-level game results and do not directly account for injuries, lineup changes, trades, travel, weather, or other unique game circumstances. However, we plan to account for these factors in the future.

How are early season ratings derived?

Performance from a team varies by game. Just because a team wins by ten points in the first game of the season doesn't mean they are going to be ten points better than the average team all season long.

To account for this, early-season team ratings are blended with a normalized version of their final rating from the previous season to start the season. The influence of the prior rating slowly fades away as the season goes on and is negligible at the about halfway point, depending on the sport.