ESPN’s Sports Analytics department introduced its College Basketball Power Index, or BPI, in 2011 as a way to rate college basketball teams based on how well they had performed and how powerful they would be.
Originally designed by Alok Pattani and Dean Oliver, the author of "Basketball on Paper," after a number of ESPN analysts came to them seeking an alternative to the outdated RPI ratings, BPI was created to provide an objective ratings system to aid with the selection and seeding of NCAA tournament teams.
Five years later, our group is proud of how BPI has performed and shaped the NCAA tournament debate. The NCAA selection committee uses BPI along with other predictive systems, such as Jeff Sagarin’s and Ken Pomeroy’s ratings, to discuss the tournament field. BPI has also proved to be accurate in its game predictions, with the BPI favorite winning 75.6 percent of games the last five years.
Like any good system, however, BPI can always be improved. We believe that by splitting BPI from one all-encompassing résumé and power rating into two metrics, we can improve its predictive accuracy and better shape the debate over the most-deserving tournament teams.
Best versus most-deserving
The debate over best and most-deserving has been a part of the college basketball landscape for years. Every March, we line up résumés to determine who should be in the tournament based on strength of schedule, quality wins/losses, home/road records, head-to-head results and more. Once the field is selected, we flip a switch to project who will win the title based on each team’s projected strength.
Snubs and undeserved bids, overseeded and underseeded teams and championships odds are all a part of the discussion and at times part of answering two competing questions: Who is best, and who is most-deserving?
The 2015-16 Wichita State Shockers are a great example of the difference between these questions. Based on its résumé alone, Wichita State likely did not deserve a tournament bid. The Shockers were 24-8 against the nation’s 103rd-ranked schedule and lacked a number of quality wins. With one of the top net efficiency ratings in the country, however, Wichita State was better than its record suggested and had the talent to make a deep run in the tournament.
To answer these sometimes conflicting questions, ESPN’s Sports Analytics team has created separate metrics. BPI remains the main power rating, but it's now more predictive than ever. Strength of Record is a backward-looking résumé rating designed to measure a team’s accomplishment. Let’s break down each metric.
BPI determines the best teams
At its core, BPI is a prediction system for college basketball. It is a forward-looking measure of team strength that is designed to answer the question, “If two teams met on a neutral court, who would win and by how many points?”
Like most power ratings, BPI looks beyond a team’s win-loss record to determine which teams are the most powerful. It accounts for how a team won or lost its games, adjusting for pace, whom those wins/losses came against and where they occurred. We encourage everyone to read our full BPI explainer, but ultimately, BPI’s main goal is to correctly predict games and season outcomes.
Therefore, BPI doesn’t care if Wichita State has two more losses than Oregon. It still might rate the Shockers higher than the top-seeded Ducks heading into the tournament. That doesn’t mean that Wichita State deserves to be a No. 1 seed; rather, it means that if those two teams had met on a neutral court, the Shockers would have been favored by BPI.
Although some analysts might value the power ratings when selecting and seeding the tournament field, BPI is best-used to project the tournament once the field is selected. Because BPI is the truest measure of team strength, it is also used to determine opponent strength in its strength of schedule and ESPN's Strength of Record ratings.
Strength of Record determines the most-deserving
Strength of Record was released late in the 2015-16 season as a way to compare résumés. SOR is a backward-looking measure of team accomplishment designed to answer the question, “Which team’s W-L record is most impressive, given its schedule?”
For years we have evaluated a team’s record and its strength of schedule when breaking down the tournament field. We all know that going 28-5 against an ACC schedule is not the same as finishing with the same record in a mid-major conference. Similarly, playing the hardest schedule in the country doesn’t mean anything unless that team wins enough games to be in the discussion.
Strength of Record marries these two concepts. It accounts for a team's opponent, the game site and whether it won or lost to determine the chance that a typical top 25 team would have each team’s record or better, given its schedule.
Unlike BPI, Strength of Record doesn’t care about how a team won its games; it simply cares about the difficulty of a team’s schedule and the result (win or loss). Everything that goes into BPI’s SOS rankings -- opponent strength, game site, distance traveled, rest and altitude -- is used to capture the difficulty of a team’s schedule. Thus, amassing a number of “good wins,” no matter how the game was won, will boost a team’s SOR.
Last season, Wichita State lacked those “quality wins” and consequently ranked 64th in Strength of Record (despite its BPI rank) heading into tournament selection. If Strength of Record seeded the field, the Shockers would have been left out, once automatic bids were taken into account.
Based on recent seasons, Strength of Record has correlated more closely with the NCAA tournament field than BPI. It should be used when discussing the teams most-deserving of a certain seed or inclusion in the Big Dance, but is less accurate when predicting the actual games.
We are confident that this duo of metrics will only expand the conversations that basketball enthusiasts have been having for the last 25-plus years. Which team is best and which is most-deserving ...