Welcome to year two of the experimental PiRate College Basketball Ratings. In year one, our three ratings performed remarkably well with the Red and White ratings hitting around 77% accuracy and the Blue rating coming in at 72%.
Because of time constraints, we will be issuing picks of games played on Saturdays from the Atlantic Coast, Big 12, Big Ten, Pac-12, and Southeastern conferences, plus some additional key games involving other teams highly-ranked. For instance, this week, we will include the Butler-Xavier game, which just might be the game of the week.
Our college basketball ratings are not fluid like our football ratings. We have to recalculate them every game as if it is a brand new rating, so to calculate every game in Division 1, we would have to recalculate upwards of 350 teams a day, and that my friends is just not possible for the small staff we have here.
Here is a brief explanation of how we calculate these ratings.
RED: Our Red Rating uses an algorithm based solely on basketball’s “Four Factors,” popularized by basketball metric genius Dean Oliver. The Four Factors are: effective field goal percentage, turnover percentage, rebounding percentage, and free throw rates. Oliver stated that the outcomes of games are decided by the offensive and defensive rates of these four factors with shooting counting for 40%, turnovers 25%, rebounding 20%, and free throw rates 15%. We have tweaked his formulas just a tad, especially the philosophy of how to count free throw rates, and we have also separated turnovers into two sub-categories: steals and all others. We then used the great back-testing tool to find an algorithm that made a “best fit” for the data into actual point spreads.
WHITE: Our White Rating uses all the data from the Red Rating and then adds a “least squares” approach to fit the teams into an order of best to worst. Because we do not plan on rating all 351 D1 teams, we only perform this for the five conferences previously mentioned plus any other teams in our weekly report. For instance, this week, we rated the 65 teams in the ACC, B12, B10, P12, and SEC plus the 3 extra teams that we have included in this report, ranking them from #1 Kansas to #68 Boston College We then take the Red Rating and adjust it up or down by a formula based on how much higher or lower each team is in our least squares rating.
BLUE: This is another statistics’- based formula relying a lot more on scoring margins and strength of schedule. The Four Factors do not come into play in this rating. Who you beat or lost to, and by how much is the base for this rating, but we break it down into a unique manner. Some teams will win games by large margins when they are superior but may not be as superior overall as other teams that win more consistently buy by smaller margins. For instance, let’s take three conference teams in a good league: Team A, Team B, and Team C. A is the best of the three and will eventually win the conference title, while C is a rebuilding team that will finish last in the league. B will finish in 4th place in this mythical 12-team league. A is good enough to sweep both B and C, while B will sweep C. You would normally expect A to beat C by more points than B beats C, but B may be that sort of team that runs the score up on teams like C, while A wins by fewer points. If A beats C by 12 points, while B beats C by 24 points, you might get the impression that B is better than A. This Blue rating does not automatically believe that to be the case.
One drawback with these formulas is that it cannot account for injuries or other personnel changes. That is why it is still experimental. We will not remove the experimental tag until we have a way to account for injuries. For instance, earlier today, Xavier lost its star point guard Edmund Sumner to an injury that required his removal from the arena by a stretcher. It is undetermined when he can return to action, and in his absence, XU goes from a Final Four contender to possibly an NIT participant if he is lost for the season. However, in the Butler game for Saturday, his absence will not be factored into the ratings.
Okay, now that we have either confused you or put you to sleep, here are our opening ratings for Saturday, January 2, 2016. Yes, get used to “2016.” If it’s 2016, there will be a little national election, and we will supply our metric analysis to this pennant race as well.
|Games Schedule for:||Saturday, January 2, 2016|
|North Carolina||Georgia Tech||16||12||16|
|Virginia Tech||North Carolina St.||-1||1||-2|
|Kansas St.||West Virginia||-4||-1||-1|