NCAA Basketball Ratings Methodology
- The number of points better the team is than an average team. This rating is the best predictor of future results. Subtracting two teams' predictive rating gives the expected scoring margin on a neutral floor.
- The quality of the outcome (win or loss) of games each team has played. It's determined by calculating the predictive rating each team would need in order to expect to win the number of games that the team won against its schedule. Since only wins and losses matter, margin of victory does not play a factor in this rating.
- The offensive component of the Predictive Rating, based on points per play adjusted for level of competition. How many points better the team is on offense compared to an average offense.
- The defensive component of the Predictive Rating, based on points per play given up, adjusted for level of competition. How many points better the team is on defense compared to an average defense.
Breakout of Offense and Defense
- Offense and defense are further broken out on team pages by how much more is scored by different types of shots (layups/dunks, 2 point jumpshots, 3 pointers, or freethrows) than an average team during a game of average pace. This is calculated using the points scored per possession by each type of shot, adjusted for the level of the opposing teams' defense for each type of shot. These individual components sum to the team's total offense or defense. In addition, a similar breakout of the offense/defense is calculated based on how many points is scored or given up on the first shot vs after an offensive rebound.
- Possessions per game that the team wants to play. Calculated based on possessions per game, adjusted for pace opponents play at.
- An estimate of the number of points the team improved from beginning to end of season, relative to all other teams. This is calculated based on how the team trended throughout the season.
All-Time Rankings are constructed by merging the ratings of all individual years. Since all ratings are points above or below the average team that year, the all-time ratings should be seen as quality of the team in the context of the era in which they played, and not necessarily a measure of who would win if matched up. The absolute level of play in college basketball likely changes over time due to resources available to athletes, the number of early exits to NBA, rule changes, and many other factors.
Offensive and defensive ratings require an estimate of number of possessions to be known in order to calculate the efficiency of the offense and defense. For recent years, the estimated number of possessions is calculated using field goal attempts, free throw attempts, turnovers, and offensive rebounds. When complete statistics aren't available for some teams/years in the past, the required stats are estimated using regression analysis on the statistics that are available. Analysis shows that in most cases, estimating these statistics does not have a significant effect on the offensive/defensive split. As the Predictive Ratings and Achievement Ratings do not require number of possessions, they are calculated the same way for older seasons as they are for more recent seasons.
A simulation of the remainder of the current season is performed to determine the chance of various outcomes occurring. It simulates every aspect of the season including each regular season game, conference tournament games, ncaa tournament selection, and lastly the ncaa tournament. This produces every team's chances of many different outcomes including outcome of each game, selection into NCAA tournament, tournament seeding, conference championship, national championship, and many more that can be found on the team pages.
Also recorded is the projected chance of each outcome if the team wins or loses the next game, which can also be found on the team pages. Actual chances shown after the next update might differ slightly (or in some cases greatly) depending on how the team played against their next opponent and the outcome of other teams' games that might affect their chances.
Any other questions, please contact me at email@example.com