There are many methods of determining the best team in Major League Baseball. You can always wait for the World Series, but that doesn’t come around until October and is reliant on a bit of the randomness and luck that come into play in these short playoff series. Many would argue that the best team doesn’t always win it all (looking at you, 2006 Cardinals). You could follow the standings all season, but standings can be unreliable and don’t consider some important aspects of baseball games like margin of victory or strength of schedule.
One method of judging MLB teams is to fit a rating system, which factors in win percentage, margin of victory, and strength of schedule, to effectively measure the best teams in baseball and predict the results of individual matchups. One such rating system is the Elo rating system. This season, we will be applying this method of rating team performance to each MLB team and track their performances over the course of the season in order to determine the team that had the most successful season.
The Elo rating system was created by Hungarian-American physics professor Arpad Elo to calculate and adjust the relative skills of chess players based on the outcomes of matches. Since then, the Elo ratings system have blossomed into use in many fields of competition, from board games to video games to sports.
The Elo rating system works by predicting the win probability of each side of a given matchup based on their pre-existing ratings and re-rates each team afterward.
If, for instance, a team that is given a 90% win probability loses, their rating will drop by much more than the rating of a team that is given only a 10% win probability and loses. These ratings can be incredibly predictive of future success, as is seen in the world of chess, and are a useful tool when comparing Major League Baseball teams.
Since this is the first year I will be tracking and publishing Elo ratings, there are many ways I could go about setting each team’s starting point:
Because each of these approaches are flawed, I decided to rate each team based on a set of run differential projections based on the performance of each team in 2021 and their outgoing and incoming talent, along with some adjustments necessary for players on the move, such as park factors and new predictions of workload. From this, we get both an offensive rating and a defensive rating for each team, along with a total Elo rating.
From these Elo ratings, I will be publishing MLB power rankings each week showcasing which teams are the league’s best performers and marking teams on the way up (or the way down).
Unlike power rankings published by other sites, these figures will be purely quantitative and objective. There will be zero subjectivity in the ranking of these teams, which will instead be based purely on each team’s performance and strength of schedule. Unlike other Elo rankings in the world of sports, my starting point will not be arbitrarily selected, and I am armed with a wealth of historical data provided by the most expansive sports statistics database in the world.
To see the first set of Elo ratings, make sure to continue checking the blog and follow us on Twitter @EliasGamePlan, as the opening day rankings come out soon, and the ratings will be updated weekly.
Hungarian-American physics professor Arpad Elo designed the Elo rating system to calculate and adjust the relative skills of chess players based on the outcomes of matches. Since then, Elo ratings have blossomed into use in many fields of competition, from board games to video games to sports.
The Elo rating system works by predicting the win probability of each side of a given matchup based on their pre-existing ratings and re-rates each team afterward. If, for instance, a team that is given a 90% win probability loses, their rating will drop by much more than the rating of a team that is given only a 10% win probability and loses. These ratings can be incredibly predictive of future success, as is seen in the world of chess, and are a useful tool when comparing Major League Baseball teams.
Each team is rated based on a set of run differential projections computed based on:
