clock menu more-arrow no yes

Filed under:

Charts! This season in win probability

Resident nerd Luke Benz uses math to show that mid-major games are more exciting

North Carolina v Gonzaga Photo by William Mancebo/Getty Images

The calendar turning to a new year may lead some to shift their focus to new beginnings. In college hoops, January means teams’ focuses shift towards conference play. With league games on the doorstep, now is as good a time as any to recap the first eight weeks of one of the most chaotic seasons in recent memory. The last two months have proven that college basketball starts long before the end of college football.

During this season of looking back, there are numerous lenses through which one can reflect. One might choose the lens of the legitimate player of year candidate (like Obi Toppin), a conference with hopes for multiple at-large bids (#4BidA10), or David slaying Goliath on an epic overtime buzzer beater (looking at you Stephen F. Austin). While these are all great stories, a season look-back really requires a breadth of viewership that surpasses what is possible for even the most ardent of fans.

That’s why I’ve chosen to look at win probability.

Examining win probability curves, which show how likely a team is to win a particular contest at any point in time over the course of a game, allows us to quantitatively answer questions like “which games have been the most exciting?”, “which teams have been most dominant?”, and “which conferences are most likely to cause fans heart attacks watching their team?”

The most exciting games

Different people might have their own views on what constitutes an exciting game. Some might prefer a nail-biter, while others may root for an epic comeback. One way to capture both of these archetypical games is through a metric I track called “Game Excitement Index” (GEI). GEI is calculated as the sum of swings in win probability over the course of a game. For example, a Game Excitement Index of 5 indicates that the win probability changed by a total of 500% over the course of that game.

There have been only nine games through Christmas with a GEI over 10.0, with eight of those games involving mid-major schools. For context, the average GEI of the over 2,300 games (D1 vs. D1) through Christmas Day was 2.65, with 95% of games having GEI less than 6.55.

Hopefully you got the chance to see one of these incredible games in real time. If not, be sure to keep your eyes on them in the new year, and if you’re a fan of one of these schools, be sure to take your blood pressure medications. Yale has already played in a whopping four overtime games, while Sam Houston State’s average GEI of 4.72 leads the entire country.

Game Excitement Index: 2019

Game Score Date Game Excitement Index
Game Score Date Game Excitement Index
Colorado St. @ Tulsa 111-104 (3OT) 12/21 14
Western Michigan @ Milwaukee 115-110 (3OT) 11/9 11.4
Texas Southern @ Northern Kentucky 98-96 (2OT) 11/23 11.4
The Citadel @ Longwood 102-99 (3OT) 12/19 11
Siena @ Yale 89-100 (3OT) 11/20 10.8
Boise State @ Pacific 82-76 (3OT) 11/23 10.7
South Dakota State @ CSU Bakersfield 93-91 (2OT) 11/9 10.7
Sam Houston St. @ UTRGV 86-90 (2OT) 12/3 10.6
SMU @ Georgia 85-87 (2OT) 12/20 10

Most exciting conferences

Another way we can use Game Excitement Index is to rank which conferences have been the most exciting thus far. In fact, by computing the average GEI for games involving at least one team from a given conference, we see that mid-major conferences occupy 21 of the top 22 spots on the leaderboard. This is perhaps not so surprising as power conferences play lots of buy games in their non-conference schedules, many of which are non-competitive. In fact, such a trend almost certainly explains why the MEAC and SWAC register as so much less exciting in the non-conference than any other conferences. Nevertheless, it’s nice to say that math proves mid-major basketball is more fun to watch.

As someone who has been an avid follower of Ivy hoops over the past half decade, it comes as little surprise to see the Ivy League at the top of the most exciting conferences leaderboard. In fact, for much of the last decade, the Ancient Eight has topped Ken Pomeroy’s leaderboard for close games. Some initial research of mine suggests that regardless of conference, conference games are on average more exciting than non-conference games, so it will be worth keeping extra tabs on an Ivy League with three front-runners (Yale, Harvard, Penn), a Sun Belt without an obvious frontrunner, and a MAAC where it’s anyone’s guess for the second-best team in the conference.

Most dominant teams

Combing through win probability to finding exciting games and conferences doesn’t quite paint a complete picture of the last eight weeks of the college hoops season. Some teams might be penalized by GEI because they are blowing opponents out, meaning many of their games are not particularly close. But some of what makes a team exciting may be its dominance — so thoroughly out classing opponents that the game is never in doubt. To capture which schools have been the most dominant thus far, we can look at average win probability. The best way to interpret this metric is the win probability a team would be expected to have if you turned on a random game of theirs at a random point in the season. Note that for the purposes of average win probability calculations, we start win probability at 50% for all teams (that is, win probability based solely on score differential and time remaining).

Average Win Probability

Team Conference Average Win Probability
Team Conference Average Win Probability
Dayton A-10 75.40%
Gonzaga WCC 75.00%
Liberty A-Sun 73.40%
San Diego State MWC 73.10%
Northern Iowa MVC 71.20%
Akron MAC 70.60%
Duquesne A-10 69.60%
Santa Clara WCC 68.30%
Pacific WCC 65.80%
BYU WCC 65.70%

It’s no surprise to see Gonzaga, Dayton, San Diego State and Liberty at the top of this leaderboard. With the exception of its first game against Indiana State, all of Dayton’s wins have been by double digits. To see Gonzaga this high given the level of difficulty of its schedule is quite remarkable. Liberty has taken care of business, albeit against a slew of weaker opponents. San Diego State has been arguably the most consistently dominant, but its dominance metrics are driven down by a close call against San Jose State. Again, this is perhaps somewhat driven by strength of schedule, but the WCC makes a strong showing. BYU is not talked about nearly enough in my opinion, and the middle-of-the-pack teams make the WCC arguably the best mid-major conference this year from top to bottom.

One way to visualize different teams dominance is by plotting average win probability over time. Comparing the top four teams from the table above, we see subtle-but-noticeable differences arise where teams really dominate their respective games. Disregarding Dayton and Gonzaga’s overtime minutes, we notice that both the Flyers and the Zags start out quickly but often only put the game away for good in the final few minutes. Liberty seems to be in cruise control the entire game, while SDSU “stumbles” more than the other three schools late in the first half, before really turning on the heat in the second half. Overall, win probability helps us paint different profiles of exactly how and when four good teams win games.

Hopefully this win probability deep dive has given a little insight into one of the ways I like to analyze college basketball, especially given that I can’t watch 120 games every weekend (as much as I wish I could). Conference play is coming in earnest. The best, most exciting games (and win probability graphs) are still to come.