CLICK HERE TO SIGN IN OR SIGN UP
About The State of College Basketball™ Ratings
There are many different polls and models in college basketball, and "The State" is one of them. But this ratings system, which has been operational since 2007 (since 2011 on the women's side) offers a few things some of the others don't. TSoCB:

  • Rewards good on-court performance
  • Punishes bad home play
  • Uses possession-based statistics
  • Degrades older results
  • Is updated every hour during the season

Here's a quick rundown on how the ratings will be calculated during this season.

Phase 1. Standard Ratings

The Ratings Percentage Index gets a bad rap from diehard fans, but it isn't useless, not at all. A lot of NCAA scientists worked long and hard to put that formula together. We use it, along with the Strength of Schedule rating, as the basis for the first phase of TSoCBB.

((((RPI * 60) + (SOS * 40) * 100) * .01 + 50) * .5) * .01

This somewhat arbitrary math gives us a number that gives a mean value around 100, while maintaining an apples-to-apples context. Wake Forest, a 15-16 team in the tough ACC, ended 2006-07 with a 100.5518. UCLA led the way with 115.3503; a 4-14 SWAC team, Alabama A&M, was dead-last at 31.5760.

Phase 2. Location-Based Performance (LBP)

In college basketball, wins and losses are important but where they happen is much more so. Our ratings make use of a catch-basket of stats, with an emphasis on possession-based measurements. Each team is graded on a per-game basis in the following categories.

  • Points
  • Rebounds
  • Turnover Rate
  • Floor Percentage
  • Free Throw Percentage
  • Effective Field Goal %
  • Free Throw Production

Click here for more information on non-standard statistics.

For every game, the team's output in each category is divided by the opponent's, with total points included three times for emphasis. Then the nine numbers are averaged and then adjusted for the opponent's strength, resulting in a figure that is either above 100 (good performance) or below 100 (bad performance). The adjustment is such: the number is multiplied by the opponent's standard-rating value from Phase 1, so good performances against good teams aren't confused with those against clearly overmatched opponents, and elite teams are summarily punished for scheduling below their station. Likewise, games between two teams of lower stature result in both squads receiving appropriately low marks. The games in which one or both teams scored above 125 tended to be the "good games."

Finally, this figure is multiplied by the number of miles away from their home gym the game was played (Miles * .000015). This rewards teams for a.) playing on the road or at neutral sites and b.) playing well away from home. Wins are give a .2 point bonus.

Here are some random per-game ratings from this year (click for boxscores). They really are random, and they'll be different if you reload the page.

Feb 12, 2014 Villanova vs. DePaul - 85.2356213441
Feb 15, 2014 Gonzaga vs. Loyola Marymount - 79.8323212194
Dec 14, 2013 South Alabama vs. Gonzaga - 47.4239536996
Dec 29, 2013 Bowling Green vs. Wright State - 34.8207173447
Jan 26, 2014 Cincinnati vs. Temple - 76.9278582772

You can view the past day's ratings, the best of the past week, and the 25 top team performances of the season on the Real Meat Report page.

Phase 3. Efficiency

Each team's performance in each game is also graded on the two most important possession-based categories of all: points per possession, and points allowed per defensive possession. These figures are normalized to result in numbers similar to those in Phase 2.

(O-PPP * 100) - ((D-PPP * 100) - 1)

The number is then multiplied by the opponent's standard-rating value from Phase 1.

It should be noted that for Phases 2 and 3, the game scores are fixed for the remainder of the season. Teams are graded on their performance against an opponent on that particular day, and it does not change no matter what the opponent does for the remainder of the season.

Here are some random per-game efficiency scores from this season.

Nov 18, 2013 Southern vs. Florida - 56.0418025925
Nov 09, 2013 Northwestern vs. Eastern Illinois - 31.84412
Jan 29, 2014 Valparaiso vs. Green Bay - 79.6131974227
Jan 02, 2014 Tennessee-Martin vs. Austin Peay - 53.049115392
Dec 03, 2013 Duke vs. Michigan - 73.7017782207

You can view the past day's ratings, the best of the past week, and the 25 top team performances of the season on the Real Meat Report page.

Calculations and Degradation

To create the final Phase 2 and 3 numbers, the numbers are averaged... but not before a degradation process on old games is applied. The most recent game is not adjusted, but the second-oldest game score in each column is multiplied by .96, the third-oldest by .92, and so on. This ensures that the freshest results mean the most, so early-season wins won't mean as much when March rolls around.

UCLA finished first in Phase 2 (119.8052) for 2006-07, but were edged out in Phase 3 by North Carolina (112.0820). There were 45 teams over 100 in Phase 2 (No. 45 was Washington, 100.0973), and 44 over 100 in Phase 3 (Washington again at No. 44, 100.0824). The lowest teams finished with numbers in the mid-40's.

Bonus/Penalty

After the raw ratings are calculated, the final number is adjusted to reward road and conference wins, and punish league losses and unsuccessful defenses of the home court.

  • Road Wins: .5 point bonus
  • Conference Wins: .2 point bonus
  • Home Losses: 1.5 points penalty
  • Conference Losses: .2 point penalty

Of the top-level teams during 2006-07, none were as affected by this phase than Arizona. Seven conference losses and five home drops resulted in a -6.3 penalty. Duke was also hit hard, with a -6. Both teams went out quietly in the NCAA Tournament's first round, to lower seeds.

After the bonuses and penalties are added and deducted, there's one final adjustment: the number of losses is multiplied by 1 and deducted from the overall number (because losing is bad). Then each team has a final rating. Once again, you can view the results of the 2006-07 retroactive simulation here.

So there you have it. We hope you enjoy following along this season as we update The State of College Basketball on a regular basis. The ratings will be refreshed every hour, so keep checking back to see if your team is on the move!


Basketball State is an ad-free, subscription-supported website dedicated to providing advanced statistics, analysis tools, historical data and news for Division I coaches, scouts, media and fans. Visitors are limited to six free page views per day, and site access costs just $12.49 per year. Learn more or sign up.
The National Statistical Concern © 2007-2014 The National Statistical Concern  ✶  Chicago, Illinois, USA
natstat.co   (872) 221-HOOP
Women's Basketball State
Advanced statistics and analysis for women's Division I college basketball.
StatState
Custom data solutions for Division I men's and women's college basketball.