(Probabilities are generated using metrics-based stats.)This year will certainly go down in the record books as one of the Maddest March Madnesses. Sports Illustrated recently reported that the 2021 Sweet 16 field features the highest seed average ever (source). With all this being said, there are a bunch of upsets that caused major shifts in calculated win probabilities this week—much moreso than in Round 1. The graph below shows just how impactful the Round 2 games were in comparison to Round 1:
But perhaps the largest storyline of the tournament so far has been the Big 10. With 9 teams originally qualifying, per 538’s data, the Big 10 had a 0.158% chance of not having any teams reach the Sweet 16. Just prior to the Michigan game, that same probability had risen to 25.6% with 8 Big 10 teams eliminated. Michigan is now the only Big 10 team left standing.
#1 Illinois vs. #8 Loyola Chicago
The first major upset of the Second Round happened to happen in the first game, with Loyola Chicago a 13-point loss. Going into the game, the probability model I use gave Illinois a 64.66% chance of winning, with an estimated point margin of 4.14. With the loss, Illinois became the 21st 1 seed to not make the Sweet 16. Whatever questions existed about how Loyola Chicago’s defense would hold up against more athletic teams than they typically see in the Missouri Valley Conference have surely been erased, as Illinois was held to under 60 points for the first time this season. Loyola’s defense held Illinois to 28.6% 3pt. shooting (9% below Illinois’s season average) and created 17 turnovers (4 more than Illinois’s season average). All this was done while giving 13 total free throws to Illinois, compared to the 22 total Loyola Chicago received.
With the result, Loyola Chicago and Illinois metrics-based chances of making the Final Four essentially flipped. Loyola Chicago’s chances soared from 10.69% to 30.27%: the largest total gain in the tournament up until this point. Unsurprisingly, Houston’s chances of making the Final Four went up ~5% as well, largely due to the best metrics-based team in the region now out of contention.
Interestingly, this switch reflects a trend I first learned about in this video from Secret Base. Before the tournament kicked off, I looked at the calculated average win percentage by seed in the Sweet 16. The 6-10 seeds all have an easier path if they manage to get past the winner of the 1 vs. 16 game in the Second Round. This has certainly held true for Loyola Chicago, whose chances of reaching the Final 4/winning the National Championship originally faced a roadblock in Illinois.
This is certainly applicable in Loyola Chicago’s case, who now faces 12 seed Oregon State in their next matchup. The same model gives Loyola Chicago a 73.5% chance of winning, with an estimated point margin of 7.
#3 West Virginia vs. #11 Syracuse
The Midwest Region led the other regions in total number of upsets during the Round of 32 (with 3), and the second of the day happened when Syracuse beat West Virginia 75-72. This caused a similar pattern that occurred with the Loyola Chicago game to occur, except with Syracuse overtakes West Virginia’s previous chances of reaching the Final 4 after the results of this game were factored in.
#7 Florida vs. #15 Oral Roberts
I am pretty sure it is safe to say that Oral Roberts is the biggest Cinderella story in a tournament with many Cinderellas. Coming off their upset win over #2 Ohio State, they beat #7 Florida 81-78, improving their chances at winning the championship another 487% over where they were prior to the Round of 32. Their chances of winning have now gone from 0.000168% at the start of the tournament to 0.00644% at the end of the second round. For a visual representation of just how much the Florida win impacted their chances compared to the first round, check out this graph:
#4 Oklahoma St. vs. #12 Oregon St.
If anything, the PAC-12 has surprised many in just how underrated some of its teams are, while the Big 10 and Big 12—despite the number of teams admitted from each conference—have been largely underwhelming. This matchup between Oklahoma State and Oregon State was another example of this. I will point out that Oklahoma State sits at 33rd in the KenPom ratings at time of writing, with Oregon State not sitting too far behind at 48th, so the seed differential of 8 between the two teams is a bit deceiving. Nonetheless, the PAC-12 tournament champions Oregon State came out with a 10-point win over the Big 12 runner-up. Oregon State held Oklahoma State to under 30% shooting (both overall and 3pt), 19% below their season average FG% and 7% below their season average 3PT%. With potential tough matchups against Loyola Chicago and the winner of Syracuse/Houston facing them, the Beavers sit with about a 6% chance of making the Final 4, up 4.3% total since the start of the round.
And that’s just Day One.
#2 Iowa vs. #7 Oregon
This one shocked a lot of people. In what was essentially an offensive shootout, Oregon took control of the game toward the end of the first half and never looked back in yet another example of the PAC-12 outperforming projections while the Big 10 and Big 12 underperform (I’ll be doing a deeper analysis of this at some point, I’m sure). Oregon was a pretty sizable underdog coming into this game, and the 8% total jump in their chances of making the Final 4 shows that (about a 261% jump).
I think the most interesting thing about this game wasn’t that Iowa necessarily played badly—they pretty much were a touch below their season averages in most statistical categories. Rather Oregon’s offense was able to catch fire, allowing Oregon to put up 95 points (20 above their season average, and with ~10% better than average FG%) and the Iowa defense was unable to stop it.
#3 Kansas vs. #6 USC
A lot of people had doubts about USC and their potential to make a deep run. USC seems to just keep proving those doubts wrong, putting up a 34-point win over the #3 seed. In yet another example of the PAC-12 beating lower-seeded Big 10 and Big 12 teams, USC managed to hold Kansas to their lowest score of the season, and their third worst loss in program history. USC’s chances of making the Final 4 nearly doubled as a result of the game, up to 17.38% from 9.48%. Metrics-wise, USC was actually favored to win the game by 1.5 points, translating to a win probability of 55% favoring USC.
Notes on Methodology:
These probabilities all assume that metrics in this pandemic season are a measure of ability. They are and they, as we’ve seen, are not. It’s all part of why we watch and what leads to the Madness of March.
To calculate win probability for an individual match, KenPom’s Adjusted Efficiency Margin and ESPN’s BPI metrics are first used to determine an estimated point margin. Factors such as notable injuries and rest days are taken into account when generating the point margin. This point margin is then applied to a normalized distribution curve with a standard deviation of 11 (what KenPom uses to calculate its win probabilities). To come up with a team’s odds of winning the tournament, weighted averages of win probabilities against all possible teams in each round of the tournament are used. This yields similar results to FiveThirtyEight and other groups predicting tournament/round win probabilities.
