Comparing NCAA Ranking Systems for 2022 NCAA Men’s Basketball Tourney Picks

And we are back one year later to compare NCAA ranking systems for the 2022 NCAA Men’s Basketball Tournament. I’m looking forward to seeing if I learned my lessons from last year’s tournament picks coverage. At minimum, I hope to improve my score of 64% correct in a tie for third place.

And also visit my Dad. Very important. More important. Dad, if you read this, I love you. And, now, back to our regularly scheduled pseudo-analytics.

NCAA Basketball Ranking Systems

For this test, we will add NET (NCAA Evaluation Tool) which is a newer evaluation method released in the past 2-3 years to upgrade how teams are rated. To ensure that RPI is not productive, we will keep that in the ranking competition this year with Sagarin, KenPom, and BPI. Here are the rules for this test:

• In each of these branded rankings, I’ll take the higher ranked choice between two teams for a game.
• For all rounds, I’ll make my picks before referencing any of the respective rankings (though I’m not sure if these rankings will change during the tournament). I’ll use intuition, records, and talking heads to generate my picks. Similar to last year and unfortunately, I’ve seen nearly no basketball this year.
• I’ll pick the same as my family tradition in a round-by-round fashion, summing to 67 games total.
• Finally, COVID-19 could eliminate a team at any point which isn’t a factor in any of these ranking systems. For test neutrality, any COVID-19 eliminations will not be counted toward a win or loss in any of the win-loss counts. Last year, this meant one game was eliminated. With the decline in COVID-19 cases and increase in vaccinations, the likelihood of this happening in 2022 is likely much lower.

And with that, follow me for the next three weeks as March Madness ensues. Let the games begin! (And Roll Tide Roll!)

The First Four

In The First Four, I’m shocked that five of the six ranking systems come to the exact same results. Although, Bryant and Wright State are awfully close to each other in some of the ranking systems. RPI, perhaps as indicated in last year’s effort, picks differently. Wright State does, by virtue of the games being hosted in Dayton, OH, have as close to home court advantage as a team can achieve. We’ll see if Bryant can pull the upset.

After The First Four, five models are tied at 100% correct picks. Earlier, I had anticipated that RPI would perform similarly to last year’s performance. The early result indicates that pattern is in progress. I’m glad I’m performing well so far; it always feels good to have a solid start!

First Round

Several differing picks feature early, making the first three games of the day–Michigan/Colorado State, South Dakota State/Providence, and Memphis/Boise State–important games in the course of this study. Beyond that, a couple of late games where I’m the outlier will also make a difference. I’m hoping that I didn’t hamstring myself with upset picks again. Overall, 10 games out of 32 have differing picks which will make a big difference in the winning rankings.

After Day 1 of the First Round, a rougher and wilder one than normal, I’ve fallen into last place with RPI. The upsets I chose, predominantly, failed. And the upsets I didn’t choose actually happened. It was a nightmarish day for picks, and even the stronger models from last year struggled.

After The First Round, I am performing worse than last year by 5%, sticking me in last place. KenPom continues to pace itself at around 75%. Sagarin is outperforming last year by 11%. BPI is slightly improved by 3%, and RPI has improved dramatically by 13%. The new NET formula is performing comparably to KenPom and Sagarin at a slightly lower pace.

For NCAA March Madness aficionados, we’ve often been advised to select major conference teams when playing against mid-major or low-major conference teams. I believe that pattern is starting to evidence itself across the two years that I’ve been intricately tracking model performance. Hopefully, I do better across the coming days.

Round of 32

For the next slate of games across the first full weekend of play, all models selected the same games with the exception of Saint Mary’s-UCLA. That lone game will make the difference between RPI and I falling further behind or catching up by a game. After that, there will not be much room to jockey for model positioning as a majority of the games will have been completed.

Here are the NCAA Men’s Basketball Tournament Round of 32 picks:

A difficult 50/50 round for me puts me even further behind the leaders. Overall, it was a difficult round for all models with a maximum of 5 out of 8 correct picks. Day 2 will feature no picks differing from the rest, meaning all models will move in tandem with each other.

A better second day with 6 out of 8 picks correct for all models brings us into a close for the second weekend. Most models are in the low 70s with RPI performing second-worst. I continue to perform poorly.

The Sweet 16

Three games will make the difference in the field in The Sweet 16–Texas Tech/Duke, Houston/Arizona, and Iowa State/Miami.

After The Sweet Sixteen, there is a clear separation between me and the rest of the models. (Apparently, I’ve had a terrible year. Don’t trust my picks.) However, this round proved destructive to all models, especially NET which only picked two games accurately. The BPI and normally poor RPI only picked four games accurately to have the most across the entire batch of models.

The Great Eight

For The Great Eight, only one model (RPI) chose one different game. At this point, things are truly becoming academic and mostly about cumulative totals and percentages due to the model separation in rankings and limited top seeds remaining (only three 1- or 2-seeds).

The Final Four

In The Final Four, only one model chose differently from the rest in BPI picking Villanova over Kansas. That match difference will either move BPI into first place or third place.

The National Championship

And with that, we’re down to the final game. All games are picked the same across all models. The standings are set with KenPom and Sagarin in first, BPI in third, NET and RPI in fourth, and me in last.

Lessons Learned

There are two main lessons that I have from this year’s tournament.

• Through the Second Round, high major teams were (by my count) 22-5 against mid-major and low-major teams. Always select high major teams.
• While I came close in a lot of upset picks, I ultimately had losses. Always skew toward picking the higher seeded team.

With those two items in mind, it is my hope that I’ll learn the lessons of 2022 and actually apply them in 2023. Until next year!