PicoBlog

Link to Simulator:

Project Overview:

The NCAA March Madness tournament is right around the corner and the excitement for college basketball is soon to reach its annual high. As a result, there has been a growing demand for accurate and reliable NCAA game simulators that can provide fans with a realistic preview of what to expect during the tournament, and guidance when making their brackets. Building on the success of a previously released NBA simulator, I decided to take on the challenge of creating an NCAA game simulator that can predict the outcomes and win probabilities of any D1 matchups from teams this year. By leveraging the use of Monte-Carlo simulation and a modified binomial tree, my NCAA game simulator aims to provide fans with a unique and engaging way to experience the excitement of March Madness!

How the simulator was made:

This simulator was originally made as a final project in an undergraduate statistical computing and simulation course, with the goal of simulating NBA games. Then, the same overall simulator structure was taken and modified to (1) match the play style of Men’s NCAA Basketball and (2) include teams and data from this year’s NCAA season. If you’re interested in how the original NBA simulator was made, check that out here.

Differences between the NBA and NCAA Simulators:

• The game clock in the NBA is 24 seconds (and played for four, 12 minute quarters) while the NCAA uses a 30 second game clock and two, 20 minute halves

• The overall pace of the game is different in the NCAA, and the PACE advanced stat (possessions per game) for both teams in a matchup is used to better match the game, in addition to the time of possession/game clock

• Difference in strength of schedule is factored in when two teams match up

  • If two teams have the same strength of schedule, than there are no shooting percentage boosts or reductions

  • If team 1 had a more difficult regular season schedule than team 2, shooting percentages are marginally increased for team 1 and decreased for team 2

    • Maximum boost/reduction in shooting percentage is 10% (which was tested for accuracy)

    • This may seem like a large penalty, but teams’ real shooting stats are used when simulating, and teams tend to shoot better against bad teams, so when playing a higher level of competition, shooting will expect to slightly decrease

• Instead of using relative defensive rating (like the NBA simulator) as a multiplier to slightly affect team shooting percentages, the teams’ average opposing team shooting percentages are factored in as a multiplier

Shortcomings of the Simulator:

• The simulator does not take recent success into account, for example teams like Duke who have been on a hot streak of late are undervalued

• Injuries (especially recent ones) are not included in simulation, for example Tennessee’s starting point guard’s injury overvalues Tennessee

If you are interested or have more questions about how this was made, please leave a comment below and I’d be happy to answer!

As always, thanks for giving this a read, and consider subscribing if you haven’t already (it’s free!).

ncG1vNJzZmirpJbBtMHRoJxnq6WXwLWtwqRlnKedZL1wvsSao6KrpJ6wbrrCmphmmpGouKbAwZqjpWWjnrq2uMCtpqs%3D