Why Doesn't KenPom Like Indiana?
Coach Adragna goes in depth on KenPom's ranking system and why it doesn't like Indiana thus far.
For college basketball fans, KenPom is a staple throughout the season. It provides advanced statistical analysis and ratings that go beyond traditional box score analytics.
Oftentimes,
’s ratings will provide a clearer picture of the college basketball landscape than polls do.It’s essential to keep this distinction between KenPom and traditional polls in mind:
KenPom uses adjusted efficiency metrics to be predictive; in other words, it is using past data to predict what a team will do moving forward.
Traditional polls and rankings tend to be more resume-focused, rewarding the actual results that have happened more than trying to predict future results.
So as Indiana sits at 7-1 and 2-0 in the Big Ten, why do they find themselves all the way down at 65th in the country on KenPom — 15 spots lower than they started?
Let’s dive in.
Preseason Rankings
The first piece of the equation is KenPom’s preseason rankings. While the specific details of the preseason rankings aren’t publicly known, Pomeroy has provided some general insights into how his preseason rankings are calculated and how they’re used in his system.
Initial Ratings
Teams are given an initial rating at the beginning of the season based on a variety of factors. These include:
Previous season’s performance
Roster continuity
Coaching changes
Recruiting rankings
Last year’s Indiana team finished ranked 30th in the KenPom rankings. They didn’t have a coaching change and their recruiting rankings were pretty solid.
The piece that hurt them in the preseason rankings was the roster/minutes continuity. Indiana has ten new faces on its roster. Plus, only 28% of last season’s minutes played returned for this season. The bulk of that is Trey Galloway and Malik Reneau.
Because of the roster uncertainty, Indiana found itself with a KenPom preseason rating of 50.
Preseason Weighting
The thing that differentiates KenPom from other analytics systems is that he doesn’t just drop the preseason factors as soon as games start being played. As the season progresses, the initial preseason ratings are gradually phased out and the ratings become more heavily influenced by actual in-season performance.
In other words, the more games teams play, the less impact the preseason ratings have on their overall ratings.
This weighting system helps to ensure that extreme preseason expectations don’t overly influence the ratings as the season progresses while also helping to avoid biases caused by early-season outliers.
How Are Teams Ranked?
When you look at KenPom’s homepage, you’ll see every college basketball team in Division I ranked from 1 through 362. These rankings aren’t arbitrary or done like an AP poll.
Instead, teams are ranked on their adjusted efficiency margin.
In short, the adjusted efficiency margin takes into account both offensive and defensive performance — assessed on a tempo-independent, per-possession basis — while adjusting for the quality of the opponents faced.
The basic idea is to measure how many points a team scores and allows per possession, factoring in the strength of the competition, with other variables normalized across all teams to make it an objective comparison.
Adjusted Offensive Efficiency (AdjO)
This metric represents the number of points a team would be expected to score per 100 possessions against an average Division I defense.
First, let’s dive into how raw offensive efficiency is calculated. That calculation is:
(Total Points / Total Possessions) x 100
The total points are easy to calculate. The calculation for figuring out total possessions is:
FGA + (0.475 x FTA) - oREB + TO
That might seem like a difficult calculation when staring at it, but it makes sense when you think about it.
Start with field goals attempted.
The take away offensive rebounds because those don’t add an extra possession.
Then add turnovers because a field goal isn’t being attempted even though you had possession.
The free throw calculation uses the 0.475 factor to account for the fact that not all free throw attempts result in a new possession.
Then, multiplying by 100 helps to express the offensive efficiency as a per-100-possessions value, making it more easily interpretable and comparable across teams.
The adjustments to the raw efficiency metric are made to account for the quality of opponents faced. Teams that perform well against strong opponents, win or lose, are given more credit than those performing equally well against weaker opponents. (So for example, if Indiana plays well against Kansas loses by one point, it will have a more positive impact on their KenPom ranking than playing poorly but surviving against Army.)
The adjustment involves considering the defensive strength of the opponents and how well the team is scoring.
KenPom also slightly adjusts based on the location of the game. A team is expected to perform better at home than they are on the road.
Adjusted Defensive Efficiency (AdjD)
Adjusted defensive efficiency is much like the offensive efficiency I just described above, except it’s on the other side of the ball.
It represents the number of points a team would be expected to allow per 100 possessions against an average Division I offense.
Raw defensive efficiency uses the same calculations described above, except it’s:
(Total Points Given Up / Total Possesions) x 100
Then, the adjustments are calculated to account for the quality of opponents faced. Teams that perform well defensively against strong opponents receive more credit than those performing equally well against weaker opponents. The location adjustment is also made, just like adjusted offensive efficiency.
Adjusted Efficiency Margin (AdjEM)
This is also an easy calculation. It is:
Adjusted Offensive Efficiency - Adjusted Defensive Efficiency
A positive margin indicates a team is more efficient than its opponents, while a negative margin suggests the opposite.
This calculation is how teams are ranked on KenPom. The team with the highest positive margin is ranked first, and the rankings descend from there.
As of today’s writing, Houston is #1, with an AdjEM of +30.86. Mississippi Valley St. is last (362nd) with an AdjEM of -29.30.
What Does This All Mean for Indiana?
As the above explanations indicate, KenPom, at its core, is an efficiency metric. It is a system that is designed to be purely predictive of future success.
It doesn’t look at a team’s win-loss record and go from there. Instead, it looks at how efficient teams are on both sides of the ball, relative to the opponents they are playing and with tempo normalized across all teams. Actual wins and losses don’t matter.
Another example to illustrate this: a 2-point loss to UConn at a neutral site would certainly be a more positive indicator for this system than a 6-point win at home against Florida Gulf Coast.
Another example is the aforementioned Army. The Black Knights have the 3rd-worst adjusted offensive efficiency in the country but scored almost a point per possession against Indiana. In KenPom’s algorithm, that was a negative indicator about Indiana’s defensive ability, so it’s adjusted defensive efficiency dropped.
In fact, all of Indiana’s first several opponents were performing better than the predictive metrics were suggesting, which was dinging Indiana.
Going into the Harvard game, the Hoosiers were 80th on KenPom. The Hoosiers performed relatively well compared to the predictive metrics and moved up to 72nd by the Maryland game.
As the Hoosiers have been stacking solid performances together, they’ve improved their KenPom ranking by 15 spots to 65.
However, that beginning stretch of the season, alongside the pessimistic preseason ranking for Indiana is ultimately what finds the Hoosiers in a spot on KenPom that has left some Indiana fans scratching their heads.
With that said, as more and more games get played, Indiana will have ample opportunities against formidable opponents to improve their rankings, and the preseason rankings weight will continue to get phased out.
Now, let’s hear from you.
How often do you look at KenPom?
Were you wondering why Indiana was so far down the rankings?
What’s your prediction for will Indiana finishes in KenPom at the end of the year?
I like the explanation from Tony, but one of my pet peeves about all the analytics is that they don't account for actual performance on the floor. A team may not be playing to its potential, but all that matters at the end of the day is the W or the L. I know, I know, that is heresy to those of us who grew up with Bob Knight, but you are ultimately judged by your wins and losses. As Bill Parcells famously said, "you are what your record says you are". Even when I coached, I would rather win ugly than lose pretty.
Thanks for a great, clear explanation. I look at KenPom occasionally, but now understand it better. IU will be top 20, but I’m an optimist.