Simulating 2023 March Madness
An objective look at the 2023 NCAA Men’s Tournament Bracket
As far as analytics go, March Madness is perhaps the favorite time of year for those who love sports and data. It always feels like a long wait, but it’s finally here and in full swing. The play-ins have begun (We see you, Texas A&M CC and Pitt). The field is almost completely set.
It’s been interesting to see analysts slowly begin to lean into advanced statistics. Voices I haven’t always agreed with like Jay Bilas are now summoning the monikers of KenPom.com and Adjusted Efficiency. Ten years ago, I wouldn’t have thought it possible but good, objective data is starting to ply the minds of the basketball world.
In 2020, I began running a simulation to replace the tournament that never was. In each year since, it has helped me evaluate the field to win some pretty deep bracket pools and land consistently in the top 15% of all brackets on ESPN. Last year, one of those brackets trended as high the top 250 overall.
I’ve also tinkered with creating an automated process for Selection Sunday and seeding all of the teams. I’m currently running with an equally balanced aggregation of NET rankings, KenPom’s AdjEM, and the Sagarin rankings. It can help immensely to see who the committee over and undervalued. This year I have not assembled a visual bracket (Becoming a father can seriously dig into your free time), but will reveal the least deserving teams in the field, and those who were absolutely robbed. In ascending order from least to most:
The Fraudsters
N.C. State, Mississippi State, Nevada, Arizona State, Pitt
The Rejected and Worthy
Michigan, Oregon, North Carolina, Oklahoma State, Rutgers
According to the hard number aggregate, 23 teams had a better case than Pitt to join the dance. It’s a perfect example of how much the committee overvalued the ACC in a year where at times they ranked below the Mountain West and West Coast Conference in overall competitive value.
On the other, less desirable side of the equation we have Rutgers. The numbers had Rutgers deserving of a 9 seed; instead as of this writing they have already been eliminated from the NIT. Ouch. Interestingly enough, the Tar Heels had a better stat rank than the two above ACC teams who got in, and would have been on the 11-line. All in all, I’ll give credit to the committee for only being 5 teams off, even if Pitt was a bit of a wild inclusion. Knowing how March can go, I wouldn’t be surprised to see the Panthers go on a fortuitous tear.
Who’s got the best shot?
Now, let’s look at the teams you should be looking at to make deep runs in your brackets.
We’ll focus mostly on the top 8-seeds who have full-percentage point chances to win (No seed below 8 has ever won the tournament). This year is a lot more competitive than last year, where Gonzaga blew one of the biggest statistical advantages we’ve ever seen. After running 500 simulations, here are the likelihoods of each team taking the title of 2023 National Champion.
1. Houston (15.4%)
The true overall #1 seed, Houston pulled down the top spot in the NET and KenPom ratings, even after suffering the loss to Memphis in the AAC Tournament final. The Cougars’ path to the final certainly isn’t the easiest with Big XII champ Texas in their region eager to defend a state title, but the Cougars still managed to grab the most simulated wins in the field with a sizeable lead over...
1. Alabama (11.8%)
2. UCLA (11%)
Jumping in front of the other 1-seeds, Mick Cronin’s squad has the experience and talent to beat anyone on the right day and in these simulations, grabbed the lion’s share of regional victories over Kansas.
1. Purdue (7.6%)
Zach Edey is the likely player of the year and has the largest lead in individual rank on KenPom in 8 years. The last player with that gap? Frank Kaminsky, who took Wisconsin all the way to the Championship game.
3. Gonzaga (6.6%)
Take it with a grain of salt after the disappointment of last year’s team. This year’s edition might not have quite the same firepower and depth, but is still well-balanced and has the benefit of playing in what looks to be the most chaotic region.
1. Kansas (4.8%)
The Jayhawks may not be deserving of the favoritism they’re seeing this year among some analysts and fans (They’re a 3-seed if ranked solely by AdjEM), but they were overshadowed by longer odds last year, when they came out on top.
4. UConn (4.6%)
The first of two 4-seeds crashing the party, which may signal that it’s not a great year to ride a 3-seed, unless you’re over last year and ready to believe in Gonzaga again.
4. Tennessee (4.4%)
Conventional wisdom says Tennessee remained highly ranked because of how dominant they were in non-conference and early conference play. The last few weeks have been a bit of a mess. Could some of the year’s earlier magic be captured again?
2. Marquette (4.4%)
The remaining 2-seeds didn’t fare as well against their counterparts this year, but all still remain in the range of possibility for victory, especially in the earlier matchups. Vermont is no St. Peter’s, but does have the best chance (14.8%) of 15-seeds for pulling the upset.
2. Arizona (3.2%)
Arizona is streaky, but they showed up against UCLA to steal the PAC-12. This feels like one of those impossible to predict teams that could drop out early, or make a long run. Risk takers could be well-rewarded.
6. Creighton (3.2%)
Could this be the most under-seeded team in the tournament? They very well may be, with nearly as much likelihood to advance from the South to the Final Four as Wildcats team above. AdjEM would’ve had the Bluejays on the 4-seed line.
2. Texas (3.0%)
Perhaps the most surprising result in this field was the Longhorns failing to win more than just 3% of the time. Maybe the buzzsaw of having to potentially take down both Xavier and Houston to escape the region is too tall of a task.
5. San Diego St. (2.0%)
The Mountain West was a lot tougher this year than most people realize, coming in above the ACC from wire-to-wire in overall value from the very start of the season. The Aztecs have the 2nd best chance of a sub-4 seed to reach the Sweet 16. More wins from there wouldn’t be the strangest of outcomes.
3. Xavier, Baylor (1.8%)
The aforementioned 3’s just don’t seem to be the safest bets this year.
5. St. Mary’s, 10. Utah St. (1.2%)
St. Mary’s success against Gonzaga during the regular season made headlines, but can they handle other top opponents they’re less familiar with? Utah St.’s odds to go beyond the Sweet 16 are an anomaly and might be the best chance of a true upset run, albeit for an under-seeded team.
Long shots (1% chance to win or less)
Kansas State, Indiana, Virginia, Duke, TCU, Kentucky, Iowa State, Texas A&M, Arkansas, Memphis, Illinois, Maryland, Florida Atlantic, Auburn, West Virginia, Boise State, USC, Penn State, Providence, Drake, Michigan St., Iowa, Pittsburgh, Iona, Kent St. I wouldn’t bet the farm on any of these, but all of these schools were statistically possible winners. Those last 5 schools won precisely one title in 500 tries. Unlike the following…
The Zero Chance Bunch
Miami, Northwestern, Missouri, NC State, Arizona State, Nevada, Mississippi State, VCU, Oral Roberts, Charleston, Louisiana, Furman, Grand Canyon, Montana State, Kennesaw State, UC Santa Barbara, UNC Asheville, Vermont, Colgate, Princeton, Howard, Northern Kentucky, Texas Southern, Fairleigh Dickinson, Texas A&M CC. Look, at least a couple of these teams will get a win. Or even make the Sweet 16. The best odds here of making the Elite 8 are at 4.2% for Missouri. If you’re a fan of any of these schools, temper expectations.
What do you think of the field? Are there different rankings or formulas you’d love to see used? Comment with your suggestions for the chance to be featured in the tournament follow-up.
