Is Wins Above Bubble the future, or merely a committee one-off obsession?

An open question in the collective quest to schedule better

Back in the summer of 2024 - truly the Before Times - I saw this press release come out.

Well, I saw it from CBS Sports, not the official NCAA website, because I do admittedly read most of what Norlander puts out. But! Point still stands. Two metrics got added to the pile of consideration for what the Selection Committee reviews every spring to make up the eventual field of 68: Wins Above Bubble, or WAB, and Bart Torvik’s predictive metrics. One new resume-based piece to the puzzle, and one new predictive piece.

That gives a lot for an average committee member to consider. You now have the following items on teamsheets: NET ranking (the NCAA’s KenPom), actual KenPom, Torvik, ESPN’s BPI, KPI (resume metric), Strength of Record (ESPN’s resume metric), and Wins Above Bubble (actual resume metric). Oh! And don’t forget your beloved Quadrants 1-4, with a heavy emphasis on what you did against Quadrants 1 and 2.

I think that, in the average season, this ends up being a bit of paralysis by over-analysis. The committee members are almost given too many data points to consider now. Tally up all the possibilities on teamsheets - the resume metrics, the quality metrics, quadrant records, and NET - and it’s 13 different items to track before you get into averages for each. This is also without introducing anything else to the puzzle: key injuries, away-from-home performance, etc. It’s a lot to take into account for the average committee member who, well, isn’t watching all the games because such a feat is impossible.

As such I noticed circa mid-March a push for Wins Above Bubble to be the determining factor in seeding, a team being in or out of the field, or hair-splitting. Possibly all three at once. As well, I found it notable that designers of metrics the committee supposedly uses to determine team quality thought they shouldn’t use their metrics in selecting or seeding the field.

After a run from 2022-2024 where most metrics had pretty similar correlations and it was usually quadrant records/performance that won out, including a 2024 Tournament where KenPom actually correlated better with seeding curve than any other metric used, 2025 saw some changes. Sure, quadrant records still performed quite well in terms of seeding correlation, with pure number of Q1+Q2 wins leading the pack as it has for some time. (People focus way, way too much on Quadrant 1. The first two together have correlated better with the field than either Q1 or Q2 have individually since their inception.) But we have a new champion among the individual metrics, one so far out in front that it set a post-COVID record.

Personal note here: I don’t have it as 7% higher correlation than every other stat on record, but rather about 3%. Still obviously quite impressive!

In 2021-22, our friends at Hoop-Explorer measured the NCAA Tournament’s seeding curve as being most closely correlated with a construction of 85% resume, 15% quality metrics. Quietly, this actually swung back around for 2023-24, to the point that quality metrics were used to seed the field almost exactly as much as resume metrics, which led to a thoroughly inconsistent seeding curve everyone loved. Just kidding. (A fun one to recall: Saint Mary’s being 41st in resume metrics, yet 17th by quality. They got a 5 seed. Meanwhile, Auburn - 4th by quality metrics, 10th by resume - got a 4. 🤷‍♂️)

I wondered if the introduction of new metrics would lead to a bit of shiny object syndrome within the committee. We have new and improved metrics available, therefore they must be superior to the old ones. This happened to an extent in 2019, where the introduction of the NET led it to immediately be a top two well-correlated metric with overall seeding in its initial run. (It was beaten out by ESPN’s Strength of Record, which is probably why a Kansas State team 24th in NET, 23rd in KenPom, but 16th in SOR got a 4 seed.)

Obviously, this doesn’t last forever. In the first normal-ish season post-COVID (2021-22), NET went from the second-most correlated metric and nearly dead even with Strength of Record to a distant third. The committee removed Sagarin from its rolls in 2023-24 yet placed a greater emphasis on KenPom than ever, leading to its status as the most well-correlated metric (by a hair) to the seeding curve overall. NET has consistently sat either third or fourth out of the seven (six in ‘23-24) metrics used in terms of correlation, which I guess would be fine if it seemed to be of any importance at all beyond “less-clear KenPom.”

The point is less about NET and more about an open question going forward: is WAB going to continue to be The One? If it is, there’s a serious chance it could change basketball scheduling for the better going forward. Here’s my theory.

---

1. Wins Above Bubble gives actual appropriate weighting to road/neutral games, or anything that’s away from home.

Here is one item to get my point across.

Here is a second item to get my point across. You are the 36th-best team in college basketball - better than the bubble average of 45th or so, but certainly no Houston, Florida, or Duke. This past year, that would’ve made you North Carolina, whose inclusion into the field certainly provided no controversy whatsoever. However, we’re not focusing on their March. We’re focusing on what they did in November and December.

Note that in terms of Wins Above Bubble, exactly one Best Win on the list (Clemson over #1 Duke) was the work of a home team. Of the nine non-conference battles on there, every single one was either in a road environment or on a neutral court. Now, I don’t get paid millions of dollars to win or lose, so as someone who makes a good-but-not-earthshattering amount of money from a paid newsletter, I can say “schedule good teams.” But, of course, if you’re the guy whose job depends on winning and losing games, you can’t and wouldn’t go out and schedule every top 10 team that’s not in your conference.

Instead, you can focus on marginal gains by sacrificing a home game or two to play a testy-but-not-terrifying team away from home. I like using Alabama as the example here because they’ve played all of North Dakota, Liberty, South Alabama, Memphis, and Davidson away from home (not counting MTEs/preseason tournaments) in the last four years alone. Minus this past year’s North Dakota team, none of those teams are going to be truly terrible. None will be amazing, either; by playing South Alabama on the road, for example, Alabama turned a Quadrant 3 home game into a Quadrant 2 road affair and didn’t have to sacrifice a ton of crowd power for it.

Or, hey, remember how the SEC went bonkers in terms of getting teams in the field this year? This is not a one-off; this is the evolution of a scheduling strategy that’s worked wonders for them for years now. Over the last three seasons, 12 of the 79 Power Five teams have played 16 or more non-conference games away from home. Seven of those 12 have been from the SEC. Only the Big 12 (Arizona State and Oklahoma State) have even offered two.

Even this year, North Carolina played seven games away from home in November and December, tied for the national lead among P5s. They went 3-4. They still ended up with a WAB of +0.39. Now, if UNC had flipped three of these to home games and still gone 3-4, their WAB would’ve been around ~0.4 lower or so…a drop-off that would’ve represented having an equivalent WAB to Indiana, the team they beat out for First Four inclusion. Want to know how many non-home games Indiana played in the non-con? Three. WAB makes this matter in a way pure W-L, quadrants, or your average understanding of rankings can’t do.

2. By making road/neutral environments more important, it should lead to better Novembers and Decembers.

In theory. Look, I like an MTE just like you do, though I don’t really love Feast Week the way you do. It’s not you; it’s me. I get overwhelmed.

Point being this: MTEs are just three games. Even Big East teams, who have 20 conference games to fill out and can play a maximum of three games in these MTEs, still have to find eight games on their schedule. Obviously, every home game equals greater overall revenue, more tickets sold, more hot dogs in hands, and more beer sales. I am, of course, not going to deny anyone the satisfaction of making it to conference play before playing their first road game.

And yet: not playing away from home, as evidenced in bubble discussions and in the committee’s new favored metric, is a huge own goal. Do you want to know what happens to bubble teams that don’t play road games? Well, they don’t make the NCAA Tournament. 19 teams - NINETEEN - of 79 Power Five schools did not play a single road non-conference game last year. That’s a quarter of your sport’s most prominent teams that aren’t touching someone else’s home court until possibly January.

Some schools (Alabama) are pretty good (Alabama) at sucking it up and playing somewhere hostile (okay, Marquette has also been really good at this) prior to January. The problem is two-fold: this leads to teams being untested and usually a bit shell-shocked in their first road game two months into the season, and it feeds into the sport’s reputation as essentially using non-conference play as unserious testing grounds that more casual fans don’t have to tune into.

The WAB effect is this: you’re #45 Indiana, and this year, you want to play #132 Illinois State. Now, almost always, you would look at this as a home game. Given your status as #45, based on last year’s KenPom numbers, you’re the exact median bubble team. Now, on average, you’d have about an 89% chance of winning this game at home. That represents a potential Wins Above Bubble of +0.11, which equals the delta between the projected win percentage and the actual win number of 1.0. However, a surprise home loss would be a potential Wins Above Bubble loss of -0.89, which could have the capacity to move you down 10 spots in just one night. A win functionally has no effect and maybe moves you up one spot, if that.

Now, try this on for size: you have instead opted to play this game at Illinois State. Applying a fairly standard home-court advantage of three points (also, guessing a good number of Indiana fans would make a 3.5 hour drive), your odds of winning this game are now 73.2%. That’s a real dropoff; you go from about a 1-in-9 chance of losing to 1-in-4. But: you now have a potential WAB of +0.27. That difference of +0.16 would’ve single-handedly moved you from 47th to 44th, ahead of San Diego State, and a loss of -0.73 would have been harmful but only drops you 5-6 spots as opposed to the 10 mentioned earlier.

You can beat Illinois State by 1 or by 100 at either place; it counts the same in WAB as a result, not a predictive. I think it’s the path forward for creating better, more exciting scheduling in college basketball, and it could deepen some lighter nights of the non-conference slate.

3. You can track the metric over the course of the season, which adds value to games beyond “Quadrant 1 opportunity.”

Okay, look, I don’t expect anyone to ever get on the microphone and say “by winning this game, VCU adds 0.8 WAB to their number.” That sounds really goofy. But…well, can we just call the wins good? Could you just say they beat a top-whatever team on the road? Can we stop obsessing over it being a Quadrant 1 Win™ if the road opponent is ranked 49th instead of 52nd? Can they just be good wins? Can winning on the road, regardless of if the opponent is ranked, FINALLY mean something in media coverage?

I may be asking too much.

4. It would help mid-majors in several ways.

Refer to #2, but also, it rewards pure accumulation of wins. Beating #166 Appalachian State, in a vacuum, is not that impressive. Beating teams at the level of App State 18 times in 20 tries when the average bubble expectation would be 17.2 in 20 is reasonably impressive. That should matter to the committee. It probably has no effect on the 2025 field that was but it probably makes Xavier/Texas a 12 seed game instead of an 11 so UCSD/Colorado State can get the 11s they deserved. Using North Carolina as the 2025 WAB poster child is fine if it means a Saint Louis makes it in 2026.

5. It removes paralysis by over-analysis.

I mean, it’s the best singular metric at measuring how you did versus the schedule you had. And I don’t have to hear about KPI or, God forbid, “Strength of Record” again.

---

The problem with this is obvious. What if the committee decides next March that WAB is old hat, and it’s just one of the metrics, not the metric? We’ve already done that before with other metrics. You’ll recall a time when the committee was hot and heavy with ESPN’s Strength of Record tool; it was beaten out by WAB this year and by KenPom last. 2023-24 looked like The Future in terms of using metrics to seed the field as much as resume; when Memphis gets a 5 seed, I think you’re aware the metrics took a step back in 2025.

The flipside of this, where the committee continues to lean on WAB to seed the field, is the better and more hopeful outcome. It can calculate that going 20-13 against a top-15 schedule and going 28-6 against the 125th-best one are pretty much the same thing. It has no idea what a Quadrant is, and thank God. It would make November/December scheduling a bit smarter, as long as coaches can get over their fear of anything risky. (Again: easy for me to say!) And, frankly, it would make analyzing the eventual field of 68 a lot easier: either you won games or you didn’t.

Anyway, we only have to wait 11 months to find out. What’s this ‘portal’ you all keep talking about?

It’s not the 2000s video game? Oh. My bad.