Friday, October 26, 2012

Basketball, Brackets and Sectional Seeding

Section2StatSheet will be your place for high school basketball brackets, stats and analysis.  The central function will be to use a statistical regression model to provide a forecast for sectional seedings for each class on a periodic basis.  The model is more sophisticated than the RPI in college basketball but less definitive than the BCS in college football.  It's not meant to tell you who the best teams are, but rather who has the best chance to win sectionals.  After Section 2 releases their brackets, I'll be making predictions and will analyze the games.
The basic premise of the model is that the size of your school matters.  Additionally, the size of your school in relation to those it plays matters.  For years Watervliet has had one of the smallest enrollments in the Colonial Council but often have one of the largest in their class.  Similar statements can be made of Mechanicville and Voorheesville.
Over the past 20 years those three schools have played in 27 title games and have won 18.  In their 9 losses, 4 have come against one of the other two.
Private schools pose greater challenges (a topic I will cover more in depth later),but the basic assumptions still hold true.  Over the past three seasons Bishop Gibbons, whose enrollment is typically less than half of CBA, LaSalle and Catholic Central's and until recently Bishop Maginn's, has had a .352 winning percentage playing in the Big 10.  In sectionals, however, playing in the A bracket instead of the AA teams in the Big 10, they have won games at a .625 clip.
Another way to emphasize this is by looking at how well schools do when playing schools with larger enrollments.  Again, over the past 3 years the team with the larger enrollment beats the team with the smaller enrollment 58% of the time.  The greater the variance between them, the greater the winning percentage.  When two teams are separated by only 31 in their enrollment numbers, the team with more students wins only 51%.  Move up to between 200-400 and it goes up to 59%.  Beyond 400 and the team with more students has won nearly 3 of every 4 games played.
To put that in perspective, by numbers alone a team at the top of class A's enrollment of 924 has on average a 70% chance of beating a team at the bottom with 525.
All of this isn't to say that I can predict how a team is going to do at the beginning of the year.  What I'm hoping to accomplish is to be able to tell you how they'll do in sectionals based on how many games they won, how many they could be expected to win based on their size and schedule and their scoring differentials.
I anticipate putting up the first round of brackets after every team plays at least five games and again after 10.  From there I'll try to put them out every Sunday until Sectionals are announced.  I'm also aiming to post midweek starting after the Thanksgiving tournaments and on the Sundays when no brackets are released.  The content will be statistically driven similar to what's above.  The more information I can get from the schools, the more varied the posts will be.  I'm also working on a post describing how the whole model works and the data that goes into it.  I'll try to get that out between now and Thanksgiving.
Things may change as the season progresses so stay tuned.