College Basketball Vegas Rankings

Betting market rankings for College Basketball are now up and running and will update daily throughout the season. As with my NFL, NBA, and College Football rankings, these are an attempt to distill the wisdom of the betting market. I do this by reverse engineering an implied ranking from the market point spreads and over/unders (more background at my Methodology page).

The key metric is what is known as "Generic Points Favored", or GPF. It is what you would expect a team to be favored by against an average Division 1 opponent on a neutral court. With the help of the over/unders, a team's GPF can be decomposed into its offensive and defensive components: oGPF and dGPF. One callout is that these offensive and defensive numbers are not pace-adjusted - meaning that they represent total points scored or allowed per game. If a team tends to play at a quicker than average pace, it will tend to score more points and allow more points, thus inflating their oGPF and deflating their dGPF. For a more "pure" representation of offensive and defensive talent, you need to adjust stats to a per-possession basis, such as those published at

The "Projected Wins" column is based on a 10,000 run simulation of the remainder of the regular season. The bar graph shows the probability distribution of that win total (tournament play not included). Just a note that that my data source may not be complete when it comes to win-loss records. The "pSOS" and "fSOS" columns summarize each team's strength of schedule. pSOS is the average GPF of the opponents already played by each team. fSOS is the average GPF of the opponents remaining on each team's schedule.

Here is the top 10 as of January 10:

At this point in the season, the rankings align pretty closely with the general consensus. As the season progresses, I would expect my rankings to diverge from the polls, as the polls tend to focus on "rewarding" wins, rather than predicting future performance. As Ken Pomeroy has pointed out, the AP rankings get less accurate as the season goes on. The pre-season AP ranking actually does a better job of predicting the tournament champion than the final AP ranking. In my analysis of NFL power ranking accuracy, I discovered a similar phenomenon with ESPN's NFL power rankings. While stat-based rankings improved in accuracy going from four weeks of data to eight weeks of data, there was no corresponding improvement from ESPN.

Based on these rankings, here are the top 10 games of the remainder of the regular season, when ranked by combined GPF of the opponents:

game date away GPF home GPF predicted spread
Duke @ Virginia 20150131 22.0 20.3 -3.0
Duke @ Louisville 20150117 22.0 18.7 -1.0
North Carolina @ Duke 20150218 18.4 22.0 -8.0
Duke @ North Carolina 20150307 22.0 18.4 -1.0
Kentucky @ Florida 20150207 26.4 13.6 8.5
Florida @ Kentucky 20150307 13.6 26.4 -17.5
Louisville @ Virginia 20150207 18.7 20.3 -6.0
Virginia @ Louisville 20150307 20.3 18.7 -3.0
Virginia @ North Carolina 20150202 20.3 18.4 -2.5
Arkansas @ Kentucky 20150228 12.0 26.4 -19.0

My simulations give the Kentucky Wildcats a 38% chance of finishing the regular season undefeated at 31-0. Their toughest test looks to take place February 7, when they visit 21st ranked Florida - though the Wildcats are still a projected 8.5 point favorite in that game. 
Powered by Blogger.