086: Bracket luck
A look at trying to use WAB to assess bracket luck, plus some thoughts on how the times are changing.
In order to win a 68 team single-elimination tournament, a team likely needs to catch a few breaks. A favorable whistle, game location, or even opponent can be the catalyst to a championship.
The 2024 men’s NCAA Tournament has produced a chalky Sweet 16. The 16 teams remaining in this iteration of the big dance have an average seed of 3.3. A normal Sweet 16 from 1985 to 2023 has an average seed of 4.5. The only other seasons in that span with a lower average seed came in 2009 and 2019 (3.1 average seed).
It’s not easy to win two games in the NCAA Tournament. The paths or road to the Sweet 16 are not created equal. There are built-in advantages based on a team’s overall seed, and advantages that are created by the madness of a single-elimination team event.
Which of the remaining 16 teams traveled the hardest path so far?
Ryan Campbell (@5th_Factor) and myself continue to beat the WAB drum1. In this instance, we’re using wins-above-baseline to assess bracket luck2.
The idea is to determine exactly how tough each team's opponents were by:
calculating the percent chance that a relatively strong team3, the baseline, would win each NCAA game
determine each team's path as the probability that this baseline team won both games (probability of game one x probability of game two)
sort the results by the remaining 16 teams
This give us an idea of the percent chances each team is still around, based on how tough a road they’ve faced. The two games are weighted equally, and there are multiple ways to arrive at a single number.
For instance, playing one easy game (95 percent chance to win) and one hard game (40 percent chance to win) is about as tough a path as playing two medium (60 percent chance to win) games.
The majority of this data can confirm assumptions, but it also proves how the seeding process can be tricky. The committee appears to use both art and science to seed teams, which is why you sometimes have a worse seeded team as the favorite against a better seeded team.
The idea is a team should have similar luck to other comparable seeds, but that’s not always the case. Here is how the paths worked out when grouping by seeds.
1 Seeds
Team Opp Seeds % Chance
North Carolina 16, 9 0.517
Houston 16, 9 0.573
Connecticut 16, 9 0.602
Purdue 16, 8 0.676All four 1 seeds remain in the tournament. As the weakest 1 seed or the number four overall seed, North Carolina’s path was more difficult than UConn, Houston, and Purdue. The baseline team would have only about a 52 percent chance to beat the teams North Carolina beat (Wagner and Michigan State) on a neutral floor.
Purdue, the number 3 overall seed, had the easiest path amongst all 1 seeds with the baseline team having a ~68 percent chance to beat both Grambling State and Utah State.
2 Seeds
Team Opp Seeds % Chance
Marquette 15, 10 0.492
Tennessee 15, 7 0.513
Iowa St. 15, 7 0.561
Arizona 15, 7 0.607All four 2 seeds also advanced to the second weekend of the tournament. Marquette was the 7th overall seed and the Golden Eagles had the toughest path. The baseline team would have about a 49 percent chance to beat both Western Kentucky and Colorado at a neutral site.
The 6th overall seed, Arizona, had an easier path than fellow 2 seeds - Marquette, Tennessee, and Iowa St. - and an easier path than two of the 1 seeds (North Carolina, Houston).
3 seeds / 4 seeds / 5 seeds
# 3 seeds
Team Opp Seeds % Chance
Creighton 14, 11 0.603
Illinois 14, 11 0.678
# 4 seeds
Alabama 13, 12 0.601
Duke 13, 12 0.644
# 5 seeds
Gonzaga 12, 4 0.431
SD State 12, 13 0.671
Three pairs of 3 seeds, 4 seeds, and 5 seeds earned a trip to the Sweet 16. Only Gonzaga played a lower or better seeded team in the first weekend. The baseline team would have about a 43 percent chance to beat both McNeese St. and Kansas on a neutral floor. It’s by far the hardest path amongst the 3, 4, and 5 seeds that advanced.
Consider that number 1 overall seed UConn had a harder path to advance than Duke, San Diego State, and Illinois. Dan Hurley uses irrational angst to motivate his team, but he might have a point in this instance.
6 seed and 11 seed (ACC)
Team Opp Seeds % Chance
Clemson 11, 3 0.309
NC State 6, 14 0.536The worst seeds to advance to the second weekend were the 22nd overall seed (Clemson) and the 45th overall seed (NC State).
Clemson’s path has been the most difficult of all of the remaining 16 seeds under this methodology. The baseline team would only have a ~31 percent chance to beat both New Mexico and Baylor at a neutral site.
NC State’s path was easiest amongst any of the 11 seeds drawing Oakland in a second round matchup. The baseline team would have about a 54 percent chance to beat Texas Tech and Oakland on a neutral floor. The Wolfpack’s path was easier than the following overall seeds: North Carolina (4), Tennessee (5), Marquette (7), Gonzaga (21), and Clemson (22).
But that’s March Madness.
Should we have seen this coming?
In hindsight yes, or at least, in hindsight maybe. Prior to the conference tournaments, there was a fairly clear set of teams headed for the 1-4 seeds when you looked at team strength and resume.
This plot (from March 15th) shows that those 16 teams lack any weak members, and you can see that once you stepped off the cliff from 4-seed to 5-seed things immediately got dicey.
Looking back this was portending the specific type of chaos we've seen with seeds 1-4 relatively safe and seeds 5 and above very much at risk. In fact, the current Sweet 16 includes 12 of those top 16 teams from March 15th - swapping in Gonzaga, San Diego State, Clemson, and NC State.
More math will need to be done to compare with previous seasons, but maybe the signs of a calm round of 32 were there all along.
What does this mean going forward?
The beauty of the tournament is that - what - often matters more than - how. It doesn’t matter how any of the remaining 16 teams advanced in the tournament. All that matters is what happens moving forward.
Given every 1 seed and 2 seed advanced, we should expect more competitive games over this second weekend. Yes, some paths might be more favorable than others, but all the teams left in the tournament can get beat.
15 of these teams will get beat too.
A note on conference loyalty
The conference loyalty conversation is approaching a race to the bottom. You can make an argument a couple ACC teams caught breaks (Duke, NC State) and a couple ACC teams (Clemson, North Carolina) had difficult paths to get to the Sweet 16. It doesn’t matter, the league accounting for 25 percent of the remaining field is notable or at least is a step towards #AccomplishGreatness4.
The Big Ten produced two teams in the Sweet 16 and both benefited from the easiest paths (Purdue, Illinois). The Big Ten has earned 23 bids to three of the last NCAA Tournaments while producing only four total Sweet 16 trips.
The ACC pointing its finger towards the Big 12 always felt a little bit in the wrong direction. There is a stronger argument if it points it towards the leaders and legends.
Times are changing
About a decade ago, the NCAA was handing out serious violations and punishments for players sleeping on couches. Illegal benefits.
This weekend you saw multiple commercials with current players. Armando Bacot for Turbo Tax. Elliot Cadeau for Marriott.
You even saw former college basketball legend Tyler Hansbrough promoting a gambling app (FanDuel) and providing color commentary for the Carolina radio crew. Hansbrough’s profit boost may have even won you some money this weekend.
The entire college sports landscape has changed. I hope we can stop using the term “student athletes” and move towards treating the players more like employees.
And with that move, I hope we can all realize how silly it is that we give coaches lifetime contracts with buyouts the size of a small country’s GDP. Think of how else that money could be used to improve the lives of the people that work around the sport (the players, support staff like athletic trainers and sports information directors, and even the officials).
The sport lacks familiarity from year to year, and we find it in the coaches that stay on the sidelines. In the future, my hope is that more of the attention and money shift away from the head coaches and towards the players and the support staffs.
Anyhow, end of rant.
Thanks for reading this far, enjoy the Sweet 16, and please subscribe below if you haven’t already. And big thanks to Ryan (@5th_Factor) for helping with this idea. You can find the code to generate this data and the table here.
Seth Burn had a great analogy for WAB on this podcast. Every team’s schedule is a golf course, and WAB helps you determine what par is on that course.
PASE or performance against seed expectation is popular way to determine bracket luck too. This is an attempt to try something a little bit different.
The baseline is defined as the #16 overall team, the weakest team expected to make the Sweet 16 (Alabama). The overall seed list can be found here.
I’m over the conference loyalty or pride conversation. I hope every team, except for the one I cheer for, loses in heartbreaking fashion. I’m ready for the horseshoe to fall out of NC State. I’m ready for Duke to play tougher teams. I want them both to lose. Sorry if this offends!
Awesome stuff! Really great to contextualize some of the runs we’ve seen so far. Also, keep beating that #WABdrum