We've developed the MPI system to share clear insights on the relative strengths of sports teams. We aim to cover a wide range of sports and levels of competition.
The system currently ranks teams in select U.S. professional, college, and high school sports. If you do not see a ranking for your sport, state, or team, check back soon, as are working to add more.
The core of the MPI system is a power ranking that we calculate daily for every team in a "rankgroup." A team's MPI value can be thought of as its expected score (points, runs, goals, etc.) in a future game, and it's best use is in comparing teams. For an upccoming game, the difference in the teams' MPI values is the predicted final game margin.
The MPI algorithm accounts for past game scores and strength of schedule. We emphasize recent games, overtime games, and playoff games. We also consider margin of victory (or loss), but we progressively discount larger margins. Thus, a team will get more credit for a 16 point win than for an 8 point win, but not twice as much.
MPI calculations are strictly math-based and do not evaluate injuries, trades, and other factors that could change a team's personnel or emotional momentum. In essence, MPIs are predicting future game margins assuming unchanged rosters.
MPI's are refreshed at the start of every season. They do not consider past seasons or roster changes from recruiting, graduations, retirements, etc. At the start of a every new season, the system assigns all teams a default MPI equal to the average game score for the sport. As games between rankgroup teams are completed, MPIs are adjusted. In the first couple of season weeks, MPI's will be relatively uncertain and volatile, but this settles out as more games are played.
Within a season, you can expect a team's MPI to change after it plays a game, but you might also see changes when a team is idle. This happens because the team's opponents, or opponent's opponents, or others have played, and this changes the value of the team's past performance. For this reason, we update MPI's daily.
Rankgroups are groups of teams competing in the same sport at roughly the same level. Currently, the MPI system sets high school rankgroups by state, NCAA rankgroups by division, and a single rankgroup for every pro sport. These could be adjusted and combined in the future. For instance, we could build rankgroups that compare all high school teams in 4-state region, or all public high schools in a given state.
MPI's are only relative within a rankgroup. For instance, and maybe obviously, a professional basketball team with an MPI of 90 is not the same strength as a high school basketball team with an MPI of 90. This same logic applies to comparison of teams between rankgroups based on division or state. So, a NCAA D3 womens basketball team with an MPI of 90 is not (necessarily) the same strength as a NCAA D1 womens basketball team with an MPI of 90. Similarly, a Texas high school boys basketball team with an MPI of 90 is not (necessarily) the same strength as a Vermont high school boys basketball team with an MPI of 90. But that Texas high school basketball team can be considered 10 points stronger than another Texas high school boys basketball team with an MPI of 80.
In a future edition, we may calculate adjustments to account for relative strengths of rankgroups.
Our power indices make it easy assign each team a "rank". But we find it more useful to focus on tiers that split the range of MPI values into ten bands, or tiers, with each tier spanning one tenth of the full MPI range. The numbers of teams in a tier will vary depending on the distribution of MPI values.
Tiers better address the common situation where many teams have similar strengths. A ranking system assigns teams a sequential ranking, no matter how close they are. This gives the impression there is an equal amount of separation between all teams, even the ones that are closely matched. A tier approach puts teams of a similar strength into a single tier, which more accurately portrays the similarity of the teams.
We consider this a work in progress. The rankings are accurate and working, but we expect occassional bugs. Also, we have many ideas for upgrades (e.g., more high school classifications, more sports, more school information, easier website menus, archive rankings, and more analytics) that we plan to add in the coming months.
If you have comments, suggestions, or questions, please email us at info@mpirank.com.
Thanks for visiting.