Tom Stein

[Tom Stein @Tom326]

@ShaneRyanHere Thumbs up for the @OhioState hat.
Thumbs down for going 0 fer with it.

[Tom Stein @Tom326]

Good summary from the great @ShaneRyanHere

[Shane Ryan @ShaneRyanHere]

trying to summarize everything LIV (at least from the last month) in three minutes

[Tom Stein @Tom326]

@ShaneRyanHere The data charts are fire.

[Tom Stein @Tom326]

@ShaneRyanHere I am slowly being sucked into a Rigamarole obsession.

[Tom Stein @Tom326]

Ohio Stadium, where Heisman campaigns go to die.

#theohiostate #BuckeyesNation #GoBucks

[Rebecca Donner @RRRDonner]

#OTD My great-great-aunt Mildred Harnack was beheaded on Hitler’s direct order. Born in Milwaukee, she graduated from the University of Wisconsin & moved to Germany to pursue a PhD. As an American grad student in Berlin, she witnessed a rapid increase in Hitler's popularity. 1/12

[Rebecca Donner @RRRDonner]

Please take a moment and honor my great-great-aunt Mildred Harnack. During this fraught time in the world, when authoritarianism is worryingly on the rise, her story is more topical and urgent than ever. May her courage be an inspiration to you.
12/12

[The Ohio State University Marching Band @TBDBITL]

Natty Bound 🏆

[Ohio State Football @OhioStateFB]

🫂🫂🫂

[Tom Stein @Tom326]

@ALostrich Alex: What are Tweets that didn't age well for $600?

[Tom Stein @Tom326]

This sums it up perfectly.
#GoBucks

[Ari Wasserman @AriWasserman]

You really want to know why people get so mad at Ryan Day? Because with the way this team is constructed, everyone knew this Ohio State team was in there somewhere and nobody could understand what took so long for it to show up. It's really that simple.

[Tom Stein @Tom326]

@PlayStation it is ridiculous that there is no way to email customer service for issues that don't fit what your limited AI Customer service bot can handle.

Please reply to me with a direct contact or I will escalate this higher and higher.

[Jason Fortin @FortinAnalytics]

PRIMER ON KENPOM RANKINGS

Here is my attempt to explain the KenPom rankings.

I would welcome @kenpomeroy to chime in if I am misstating anything.

The KenPom rankings most referred to in the college basketball industry are based on KenPom’s metric called Adjusted Efficiency Margin.

It is worth noting upfront the difficulty in developing a metric to evaluate college basketball teams vs. say the NBA.

In the NBA, there are 30 teams in which each team in the two conferences plays a very similar schedule.

Wins and losses are typically a very good indicator in evaluating teams.

There is a luck factor in wins and losses so many will look at Efficiency Margin in measuring a team’s performance.

A team’s Offensive Efficiency is the number of points they SCORE per 100 possessions.

A team’s Defensive Efficiency is the number of points they ALLOW per 100 possessions.

The difference between the Offensive and Defensive Efficiencies is the Efficiency Margin.

The issues in evaluating Division I college basketball teams by simply looking at win-loss record and Efficiency Margin are the disparities in strength of (1) teams and (2) schedules.

Strength of Teams

There are 362 DI teams and the disparity of ability among teams is ENORMOUS.

For example, look at the extreme ends.

In the 2023-24 season, UConn was the best team with a 37-3 record and won the National Championship.

The worst team was Mississippi Valley State (1-30 record).

Note: UConn beat MVS on Nov 14th, 87-53. Based on how UConn steamrolled through the NCAA tournament, kudos to MVS for keeping the game that close.

Strength of Schedules

Teams have very different and distinct strengths of schedules (SOS).

Each DI team has a non-conference schedule which is typically 11 to 13 games.

There are many different philosophies in scheduling non-conference games ranging from:

(1) scheduling difficult games to challenge a team and prepare them for conference play to

(2) scheduling easy games, including Non Division I opponents, to gain as many wins as possible and minimize losses heading into conference play.

There are 33 different conferences in DI, also with a wide range of competitiveness.

Each team typically plays 19 to 23 conference games including the end of season conference tournament.

So, comparing win-loss records in DI basketball without really digging behind those numbers is a meaningless exercise.

Example of Three Identical Win-Loss Records

Tennessee, Arizona, and High Point ended the 2023-24 season with identical 27-9 records.

Arizona and Tennessee played the 12th and 19th toughest non-conference SOS out of 362 DI teams, respectively, according to KenPom.

Arizona and Tennessee played in the Pac 12 and SEC conferences which were ranked the 6th and 4th toughest conferences, respectively, according to KenPom.

High Point had the 308th hardest non-conference SOS and played in the Big South which was the 18th toughest conference according to KenPom.

Obviously viewing High Point’s 27-9 record as comparable to Arizona and Tennessee’s 27-9 records would be silly.

Just to be clear, the point of this example was not to discount High Point's terrific season.

Congratulations to High Point head coach GG Smith on taking a team that was 14-17 last season and improving to 27-9 this season.

What is KenPom’s Adjusted Efficiency Margin (“AdjEM”)?

KenPom calculates an AdjEM for each basketball game between DI schools.

Like the NBA, it calculates a Net Efficiency Margin (number of points scored per 100 possessions LESS the number of points allowed per 100 possessions).

But KenPom further adjusts those numbers based on the strength of the opponents.

KenPom rankings are based on the average AdjEM per game for each team.

UConn ended the season with an AdjEM of +36.43, which ranked 1st in the country.

What this number signifies is that if UConn were to play an average NCAA DI team, it would expect to outscore that opponent by 36.43 points over 100 possessions.

What affects AdjEm for each game?

The four factors that determine the KenPom AdjEm for each game are:

1) Scoring margin.

2) Level of competition: KenPom looks at the average Adjusted Offensive Efficiency and Adjusted Defensive Efficiency of an opponent throughout the season and factors that in the game algorithm.

3) Pace: a 10 point win in a game with 60 possessions will result in a higher AdjEM than a 10 point win in a game with 70 possessions.

4) Site of the game: home/neutral/away. Against the same opponent, a 10 point win on the road will result in a higher AdjEM than a 10 point win.

It is important to note that the two PRIMARY factors in AdjEM for a particular game are (1) scoring margin and (2) level of competition.

Ken Pomeroy is very upfront that KenPom is a predictive algorithm and not resume based.

So, wins and losses are not factored in the algorithms.

The thought being if a team lost to a team by a point vs. win that game by a point, the team’s predictive performance over the course of a season and going forward would not be materially different by those two points of scoring margin.

Alabama's KenPom Ranking After Non-Conference Play

A good example of how scoring margin and level of competition affect AdjEm and the KenPom rankings is that Alabama was 8-5 in non-conference play this season with the 16th hardest non-conference SOS according to KenPom.

Alabama’s five non-conference losses were to Purdue (national runner-up), Clemson (Elite 8), Arizona (Sweet 16), Creighton (Sweet 16), and Ohio State.

Alabama entered conference play in January with an average AdjEM that KenPom ranked 6th in the country.

Obviously that KenPom ranking at the time looks great in retrospect with Alabama making the Final Four and giving UConn its most competitive game in the NCAA tournament.

Ken Pomeroy recently published a Substack article highlighting some tweets that he saved that were less than kind of what they thought of Alabama’s KenPom rating at the time.

What is a perceived negative issue with KenPom?

A frequent knock of KenPom is that a team can positively affect its AdjEM by scheduling weak teams and blowing them out by 30 to 50 points.

I previously mentioned Mississippi Valley State. It turns out they had the 2nd hardest non-conference SOS according to KenPom.

The following schools scheduled non-conference games against MVS this season: UConn, Gonzaga, Baylor, TCU, Oklahoma, San Francisco, North Texas, LSU, Santa Clara, Liberty, Cal State Northridge, and Tulsa.

They all won with an average scoring margin of 39.4 points.

Ken Pomeroy will tell you that though not ideal, being able to beat a weak team by a substantial margin has historically been shown to be a good predictive indicator of future performance.

Future Posts

In a future post, I will get into how this issue was a widely discussed topic through the college basketball landscape this season.

I will also post separate primers on two other ranking metrics:

(1) the NCAA NET - the primary ranking tool used by the NCAA since the 2018-19 season

(2) KPI - a results based metric that appears to have played a larger role with this year's NCAA selection committee than in prior years.

[Tom Stein @Tom326]

Pitt and SJU should have been in the dance.

They both also should have proven that in the NIT like Indiana State and Seton Hall have.

[Jason Fortin @FortinAnalytics]

The NCAA tourney expanded to 68 teams in 2011.

Pre-tourney, here are the lowest KenPom ranked teams not to make the field of 68 for each season:

2024 St. John's 25th (lowest KP ranking of any team over this period)
2023 Rutgers 35th
2022 Oklahoma 30th
2021 Duke 33rd
2020 no tourney
2019 Clemson 29th
2018 St. Mary's 28th
2017 Clemson 35th
2016 St Mary's 29th
2015 Florida 35th
2014 SMU 33rd
2013 Iowa 33rd
2012 Miami 42nd
2011 St. Mary's 37th

[Jason Fortin @FortinAnalytics]

The NCAA tourney expanded to 68 teams in 2011.

Pre-tourney, here are the lowest KenPom ranked teams not to make the field of 68 for each season:

2024 St. John's 25th (lowest KP ranking of any team over this period)
2023 Rutgers 35th
2022 Oklahoma 30th
2021 Duke 33rd
2020 no tourney
2019 Clemson 29th
2018 St. Mary's 28th
2017 Clemson 35th
2016 St Mary's 29th
2015 Florida 35th
2014 SMU 33rd
2013 Iowa 33rd
2012 Miami 42nd
2011 St. Mary's 37th

[Jason Fortin @FortinAnalytics]

The KPI metric doesn't think much of UConn.

I know that I sound like a broken record citing the stats below, but then again so is UConn after their 5th straight blowout win since beating St. John's 95-90 in the BE semis. Avg margin of victory is 25.4 points (played Marq, NW, SDSU and Illinois).

Samford lost to Purdue by 53 pts this season.

Samford had a higher KPI score in that game than SJ did against UConn in BE semi (.051 Sam vs. -0.000 SJ).

[Tom Stein @Tom326]

The Big10 was going to suspend James Franklin, in order to punish Michigan for sign stealing, but their lawyers talked them out of it.

#BIG10

[Tom Stein @Tom326]

When it debuted, the Magnum XL 200 was the tallest, fastest, steepest roller coaster in the world.

For a fearless six-year old, taking a picture from the exit line with that first drop she just conquered in the background, it's no big whoop.

#nextgeneration #cedarpoint

[Tom Stein @Tom326]

I am biased. But damn straight from my old friend @MikeWiseguy

[Mike Wise @MikeWiseguy]

I had no dog in Peach Bowl fight, so I can say without bias Javon Bullard should have been called for targeting against Marvin Harrison, Jr., who never returned to the game. Ohio St. settles for FG instead of 1st and goal at the 1. Bullard becomes game’s defensive MVP. So wrong.

[Mike Wise @MikeWiseguy]

I had no dog in Peach Bowl fight, so I can say without bias Javon Bullard should have been called for targeting against Marvin Harrison, Jr., who never returned to the game. Ohio St. settles for FG instead of 1st and goal at the 1. Bullard becomes game’s defensive MVP. So wrong.