If you are still confused, I highly recommend checking out the example here, which walks you through the calculation step-by-step. In other words (and using the example data above), Kansas as an opponent will have a different WP for when calculating the OWP for UConn than it would for Wisconsin. Though calculating the RPI seems simple on the surface, there are two additional caveats that make it a bit more challenging (and a great example for functional programming).Īn additional weighting factor (1.4 for away wins/losses, 0.6 for home wins/losses ) is included for the reference team’s WP calculation, but not for OWP and OOWPįor OWP and OOWP, the reference team is not included in the win/loss record, making both of these calculations for the reference team unique to that team. That being said, it is still one of the biggest factors used by the committee for tournament selection and seeding, so it’s worth looking at and understanding. The main criticisms of the RPI is that it fails to take into account margin of victory and it places too much weight on facing strong opponents (to the point where it clearly rewards teams for losing badly to many strong opponents and penalizes teams for playing well against weaker ones). In other words, it is trying to use a team’s win/loss record and their strength of schedule to infer their rating. It is a weighted formula, where 25% comes from a team’s own winning percentage (WP), 50% from its opponents’ winning percentage (OWP), and 25% from its opponents’ opponents’ winning percentage (OOWP). The RPI is one of the most widely known (and widely criticized) rating systems used in College Basketball. Then, in future posts to be released throughout the tournament season over the next few weeks, I will be able to use the rating systems described here to answer some interesting questions about tournament teams in the past and what is happening in 2018.įor this, we will need the following R packages, and a small dataset outlining an example season. The focus here today is to describe two popular methods, the Rating Percentage Index (RPI) and the Simple Rating System (SRS), that you can easily calculate and track yourself using functional programming in R.
Unsurprisingly, there are literally hundreds of different ranking systems for College Basketball and other sports, some created by large news organizations and others by passionate fans.
With only 30 or so games in a given season and around 350 teams in Division I, being able to measure the relative strength of teams who have not played one another becomes an important challenge. March Madness is officially upon us as College Basketball teams across the US try to get a few more signature wins on their resume during conference tournaments before Selection Sunday. Calculating College Basketball rankings using functional programming in R March 10, 2018
0 kommentar(er)
