New College Basketball Model

College Basketball Best Bets:

New Model leads to 56% Best Bet winners last season!

I have been very profitable on my College Basketball Best Bets over the years (+368.8 Stars the last 23 seasons), and especially in the last 4 years when I started releasing most plays in the morning when lines are softer (54.7% and +81 Stars the last 4 seasons).

The results were even better last season, which was the first season that the College Basketball plays were based on a new model that tested incredibly well (832-636-29 for 56.7%) over the previous 4 seasons. The model was tested on 4 seasons not used to derive the model to avoid back-fitting. The first season using that predictive matchup model produced very good results on Best Bets (55.7%) despite winning only 40% of the close games (bets decided by less than 3 points).

2020-21 Season: Best Bets were 165-131-3 (55.7%) and Opinions were 153-131-5

 

College Basketball model – 56.7% on 4 test seasons!

When the 2020 college basketball season abruptly ended we set out to program a college basketball model to improve upon the methods I’ve been using for years, which were less sophisticated in terms of priors and in-season blending with priors, while automating the player adjustment factors that I’ve been calculating by hand for years.

The results on 4 test seasons not used in the derivation of the model were even better than we expected, going 832-636-29 (56.7%) on side plays, while having a play on 11.7% of the games in the sample (see the full results below).

The key to having a good college basketball model is having good priors, which are the expectation of each team’s performance level entering the season. Our college basketball priors are derived from a blend of program, coach, and player ratings. The program and coach ratings are based on year-to-year correlations across a variety of metrics. The player ratings are derived from recruiting rankings and aging curves. We project every player’s contribution per 100 possessions and then predict playing time for each player to come up with all of the player value priors. The combination of all three ratings (program, coach, player) measure the changes of the offseason, which mainly come from transfers, new players (freshmen and JUCO), and coaching moves.

Once the priors were perfected the model was built using the same model building principles used in building the current NFL model, which has produced great results (58.5% over 6 seasons).

The college basketball model was initially built using only data from the 2003-04 season through the 2015-16 season and then was tested on the next 4 seasons, which were not included in the derivation of the test model to avoid back-fitting.

The results of the model on the 2016 to 2020 seasons were incredibly good based on significant differences between our model prediction and the line (i.e. 3 ½ points or more difference).

 

2016-17: 217-149-5 (59.3%)

2017-18: 170-139-11 (55.0%)

2018-19: 244-18-7 (57.0%)

2019-20: 201-164-6 (55.1%)

 

Total: 832-636-29 (56.7%)

 

The model went live last season with good results in a difficult environment with added variance due to Covid.

2020-21 Season: Best Bets were 165-131-3 and Opinions were 153-131-5

 

The results of the test seasons exceeded expectations, especially considering that the model was tested on closing lines and we will be releasing most games in the morning when the lines are soft. The model test also did not include any adjustments for player personnel changes (injuries, lineup changes, etc), which is something that will be adjusted for and has been a big part of the success of the NFL model.

The model parameter weightings have been updated to include the last 4 seasons that were initially excluded, and improvements were made to the model this offseason with the addition of offensive and defensive play style parameters and shot type distribution – so the model is fine-tuned for the upcoming season. I’m really excited about the model and expect to have great results in college basketball going forward.

We’ll have plays starting opening night based on priors which are certainly better than the market. The early season model plays (i.e. the first 25 days) based solely on our priors were 250-185-4 on sides (57.5%) in the 4 test seasons and last season, using the new model, the early season plays based solely on the priors were 41-26-1 on sides and 21-13 on totals.

 

Model test results for totals:

Prior to last season we didn’t have totals in the database going back far enough to test how the model predicted totals. This offseason all closing totals going back to the 2003-04 season were uploaded into our database, which allowed us to test the performance of the model on the closing total.

The results of the totals model was a very good 808-562-18 for 59% winners on qualifying plays, although in this case the totals tested were part of the training set and aren’t expected to win 59% going forward. However, the totals were tested using closing lines, which are not as easy to beat, and did not include the rotation/injury adjustments that adds value.

A realistic expectation for totals going forward is 55% and we will have Best Bet totals this season after only releasing totals as opinions last season (only 85-75-2 on those but the average line differential of +2.8 points suggests that our record should have been better).

 

 

 

  • Dr. Bob College Basketball Season

    2020-21 Season: Best Bets were 165-131-3 and Opinions were 153-131-5

    More information on the College Basketball model.

    Receive all of my NCAA Basketball Best Bets and Opinions through the 2022 NCAA Tournament Championship game – and predictions on every NCAA Tournament game.

    In addition to the emails, you can also choose to receive texts of each Best Bet. The texts are sent to everyone at precisely the same time, which allows you to beat the line moves. You can also view the Best Bets on the site.

    $1,145.00 Subscribe
  • Dr. Bob College Basketball 4-Weeks

    Get College Basketball Best Bets for 28 days. In addition to the emails with analysis, you can also choose to receive texts of each Best Bet. The texts are sent to everyone at precisely the same time, which allows you to beat the line moves. You can also view the Best Bets analysis on the site.

    $345.00 Subscribe
  • Dr. Bob College Basketball 7 Days

    Receive the Dr. Bob Sports College Basketball Best Bets and Opinions via text and email each morning for 7 days. Plays will be released each morning starting at 6:30 am Pacific, so make sure your book has lines available that that time before signing up.

    The Day Pass is no longer available. The large majority of Best Bets I release to subscribers move out of Best Bet range after the release, so it’s a disservice to you to have a day pass selling plays that are not as likely to win. The 7-day pass will enable you to get the plays as they are being released.

    $125.00 Subscribe
  • NBA Guru Basketball Season

    Receive all NBA Guru Best Bets and Strong Opinions through the 2021-22 NBA Playoffs delivered via email (and via text) and available to view on the website.

    In 9 seasons with Dr Bob Sports the NBA Guru is 1218-1014-34 on his Best Bets and 2512-2086-69 on a Star Basis (76-58-1 on 3-Stars and 1142-956-33 on 2-Stars) for a profit of 213.9 Stars at -110 odds (with an extra -3.5 for added juice).

    $925.00 Subscribe
Share this