In past years, I’ve written in detail about year-to-year improvement among college basketball teams. Back in the 2013-14 season, I observed a very experienced North Dakota State team. The Bison had nearly everyone back from 2012-13 and were expected to have a big season largely thanks to a great defense. However, the defense took a big step back in 2013-14 and instead the offense (which would finish the season ranked 29th in AdjO) was what brought NDSU to the round of 32 in the NCAA tournament.
It turned out NDSU was not alone among teams with tons of experience being unable to take a step forward on the defensive end. Last pre-season I speculated about possible reasons for why this is the case and tried to apply the findings to project Wisconsin’s defense.
In this post, I’m looking at a different element besides experience that could be useful in season-to-season projections: consistency. We tend to think of consistency as a good thing, but this is not necessarily the case. Obviously teams can be either consistently good or consistently bad. Furthermore, consider a team with a good but not great level of play over the course of a season. If this team was inconsistent, that means they had some great performances and some very bad performances. Should the higher ceiling for this team affect our projection for the following season? That’s the theory I am looking at in this post.
To measure consistency, I created a “performance” metric (like I’ve done in many posts in the past) for every game from 2004-2013 that is an estimate of KenPom’s AdjO and AdjD ratings. In other words, points per possession adjusted for opponent and game location. Then, I took the standard deviation of each team’s game-to-game performance to measure consistency.
Of the 3,372 teams from 2004-2013: 309 met my definition of very inconsistent, 266 met my definition of very consistent, and the remaining 2,797 were simply defined as somewhere in the middle. Next, we want to look at team performance the year after. The results are below:
39% of inconsistent teams jumped up 20+ spots in KenPom rank the following season. That 39% compares favorably to the 35% by consistent teams and the 33% by teams somewhere in the middle. That appears to be some evidence for the theory that teams that have shown a high ceiling the previous year are more likely to make a jump in year two. The same can be said of consistent teams being more likely to avoid a drop in performance, but that’s for another post.
With this data in mind, let’s take a look at some of the most inconsistent teams from the 2013-14 season. I limited my search to teams that finished in the top 50 in KenPom. The idea here is to identify teams with the possibility of going from good to great:
Last year’s Miami team is the model example for a good but not great team with a high ceiling. Jim Larranaga’s squad failed to make the NCAA tournament despite some impressive performances, most notably beating Duke by 16 on the road. Of course Miami also lost to Georgia Tech by 20 at home.
Flash-forward to this season and Miami, Iowa, and Davidson are a combined 11-0 to start the season. Obviously we are dealing with an extremely small sample this early in the season. And for full disclosure, Davidson’s record is a bit misleading. The Wildcats have probably underachieved a bit to start the year with 3 close wins at home against mediocre opponents (albeit with a Jack Gibbs injury mixed in). But the ceiling for these three teams remains high. Miami has been one of the biggest surprises in the country after a 5-0 start and Iowa has a 28-point road victory at Marquette to its name. Could these teams’ strong (but inconsistent) performances last season help us explain the surprise starts?
Returning minutes, injuries, and transfers all must go into a pre-season projection and help determine the mean expectation for a team. But the idea of consistency and a ceiling is something that might be useful to consider for potential breakout candidates. Miami won’t finish the season shooting their current three-point percentage of 45% and the offense will certainly regress to the mean to some extent, but they are also a team with a high ceiling that might just have finally figured out how to put it all together.