I'm a Statistician and a Friar Fan
So do I hate Kenpom?
I’m a statistician and Providence fan. Do I hate Kenpom?
No. But I wish basketball fans would stop quoting it as a ranking.
The goal of a team is to win individual games, not improve your season-long margin of victory.
It's a straightforward calculation that offers great insight and value, but it’s not a measure of how “good” a season a team is having.
As of this post, you’ll see a “30” next to Providence’s name on Kenpom, but that doesn’t mean it’s correct to claim that Providence is “30th in the country.”
What it means is that Providence’s performance to date is associated with scoring 14 more points than the average D1 offense and allowing 4 fewer, per every 100 possessions. There are 29 teams that have higher such margins and 328 that have lower.
And that’s because inference statistics tell you the “what” not the “why”. A team’s season-long offensive and defensive efficiency numbers are a starting point for analysis, not a conclusion.
As reluctant as I am to reignite the controversy, lets look at last year’s Providence team as an example.
Why did Providence fans think their team was “low”, despite having a great record and a number of wins over quality opponents?
But was Providence low in the rankings? There are 358 teams in division 1. Even at 50th, Providence was still in the 86th percentile. Additionally, the dispersion between teams is tight; the difference between 50th and 35th could be a single basket made or missed per a game over the course of a season. That’s normal variability, and not an underlying indicator that one team is “better” than the other.
Ok, but why were they 50th throughout the season when they had a better record against better opponents than other teams ahead of them?
It wasn’t because they won several close games against good teams, or because they had a high “luck” factor. It’s because they didn’t blow out their buy-game opponents.
One game in particular stands out in the data, against Central Connecticut State on December 11th. Providence went up by 20 at the U12 of the first half. They extended the lead to 30 early in the 2nd…and then they just stopped playing.
They were outscored the rest of the game, even giving up a 12-2 run at one point. They had 61 points with 15 minutes left in the game…and finished with 68. This is against a team that finished 348th in Kenpom.
So what do the inference statistics say, as baked into Providence’s AdjEM that followed them over the course of the season? It says that in 30 mins of game time, 50 possessions, that Providence was worse than the 348th team in the country. That Providence’s floor as a team extends so low, that it falls off the bottom of the chart completely.
The performance at the time buried their adjusted offensive score by at least 10 spots, offering daylight to critics that claimed Providence was overrated. But is that accurate? Was Providence legitimately severely outplayed by one of the worst teams in the country on their home floor?
Upon re-watching the game, I saw something different. I saw a veteran-laden team that went up big early, and then shut it down. To avoid injury, to not put anything on film, to save it for conference play; at that point less than a week away. I watched AJ Reeves not jump for a rebound on 3 straight possessions. I watched them sleepwalk through closeouts, and not attempt a shot on 4 straight possessions.
They weren't playing like a team protective of their analytics, but rather a veteran-laden team that decided, as a choice, to shut it down against an inferior opponent. Because it was time for the sun to set on a 10-1 non conference, even though they had 50 possessions left to play.
Incidentally, the first game of the conference schedule was at Connecticut, a team that had beaten that same CCSU team by 51 earlier in the season. And in that first game at UConn, who was ready and who wasn’t? The team that obliterated the 348th team by 51? Or the team that ended the game early, and saved it for when it mattered?
A core part of the Providence narrative is that they are better in February and March vs. November and December. Is that because Ed Cooley makes the right choices for his team? Even if that comes at the expense of their AdjEM?
