Wins Above Bubble and the At-Large Case for Stephen F. Austin
Why blame SFA for their schedule when they’ve dominated it?
** stats accurate as of Wednesday morning (3/4) **
What I find most interesting about the NCAA’s switch from the RPI to the NET as the selection committee’s primary tool last season is just how different the two metrics are.
For example, Auburn currently ranks #3 in the RPI but falls to just 28th in the NET. A few spots below them at #33 in the NET is a 16–14 Purdue team that can be found all the way down at #77 in the RPI. Here we have two teams that are viewed similarly in one metric but are so far separated in another that one would be a projected 1-seed and the other wouldn’t even be close to the bubble.
This isn’t just a matter of two teams, either. Northern Iowa is #19 in the RPI and #37 in the NET, Texas Tech is #22 in the NET and #68 in the RPI, and Stanford is #25 in the NET and #56 in the RPI.
Now, even when it was still in use, the RPI wasn’t the only determining factor of a team’s NCAA Tournament fate. The NET isn’t today, either (it‘s mainly used to determine which teams fit where in the quadrant system). That being said, the startling truth is that had this same season taken place just a few years ago, the tournament field would likely look much different than it will on Selection Sunday.
The NET may be a better tool than the RPI, but that doesn’t mean aspects of it aren’t still arbitrary. There could have been any number of quadrants, but the NCAA went with four. Where the lines are drawn between quadrants and the differences between home, road, and neutral games are also somewhat random. Yet, all of these factors heavily influence our perception of teams.
Zooming back out to the selection process at large, the members of the committee each have their own beliefs, so each year will bring slight differences in philosophies. Changes to the team sheet mean the committee works from a different set of information. And listing numerical rankings for metrics rather than the rating itself disregards the varying gaps between teams.
The selection committee follows a specific procedure, but there are no rules about how they justify ranking a team above another nor a best metric or formula for doing so. Their task is simply to do the best job they can with the information at their disposal. Over time, as our values change, we continue to refine the process.
Right now, the process values strength of schedule and wins in quadrants 1 and 2. Naturally, this favors larger schools, as they get more opportunities for quality wins by virtue of playing in stronger conferences and having the prestige to play in nonconference tournaments and schedule stronger opponents for home games.
Smaller schools don’t receive these opportunities. Instead, their strength of schedule and predictive metrics get crushed by playing a conference schedule of weak opponents where they are unable to prove their strength and any loss is seen as disqualifying. But surely, some of those small schools are actually deserving of spots in the tournament, right?
Enter the Stephen F. Austin Lumberjacks. SFA is 27–3 overall and 18–1 in their conference. There’s only one problem. That conference is the Southland, which ranks 30th out of the 32 conferences in Warren Nolan’s conference NET rankings and went just 18–82 in nonconference play against Division-I opponents.
Conventional wisdom suggests that teams in conferences like the Southland are undeserving of at-large bids. Despite having the fourth-best winning percentage in D-I (behind San Diego State, Gonzaga, and Dayton), the Jacks need to win their conference tournament to make the Big Dance and avoid ending up like their 2012–13 team that went 27–4 before being sent to the NIT.
They currently rank #80 in the NET and have just one victory against a team in Q1 or Q2. Of the five additional advanced metrics listed on team sheets, all but one has them in the low 100s. (More on that “one” later.)
But let’s look a little deeper at their resume. SFA may only have one quality win, but that came on the road at Duke. The Blue Devils were at that point the #1 team in the country and still project to receive around a 3-seed. This is a huge win.
The Jacks played two other Q1 games, losing at Rutgers and Alabama. Rutgers is 17–1 at home this season, so it’s tough to fault that performance. Alabama is a team likely to end up on the wrong end of the bubble, but the fact remains — it’s not easy to beat a good team on the road.
Outside of these three contests, SFA is 3–0 in Q3 games and a whopping 20–1 in Q4 games. That’s one loss for the entire rest of the season, coming by a single point against an admittedly awful Texas A&M-Corpus Christi team which comes in at #314 in the NET.
To some, that loss is proof that SFA doesn’t deserve an at-large spot in the 68-team field. However, using one game to discredit an entire season’s worth of performances doesn’t seem fair. After all, Kentucky lost to an Evansville team that went 0–18 in Missouri Valley play, but no one seems to be worried about the Wildcats’ March Madness chances.
Given 24 games against Q3 and Q4 opponents, is it realistic to expect an at-large team to run the table? Do we really think a team like Purdue, which despite being 16–14 overall is in a much better (although still not great) at-large position than SFA, would have won all of those games? Cincinnati and Providence are both on the bubble despite each on the bubble despite having four Q3/4 losses apiece. They have more quality wins than SFA, but of course, that comes with many times more opportunities.
A better question would be this: what would it take for this SFA team to get an at-large bid? It’s something Ken Pomeroy mentioned in another article about the Lumberbacks a few weeks back. He guesses that they would need one or fewer losses to be in position for an at-large, which he correctly deems an unreasonable standard.
Even in that situation, I don’t think the numbers would suggest they make it. Flipping the one-point loss to Corpus Christi to a one-point win would change virtually nothing in terms of predictive metrics, and a single win over Rutgers or Alabama wouldn’t be able to make up the difference between SFA and other bubble teams. The committee might be compelled to award the Jacks an at-large bid based purely off of their overall record, but the numbers wouldn’t agree, and that’s a problem.
Teams with a dozen or more losses routinely make the NCAA Tournament, using the luxury of a good strength of schedule as a buffer to their overall performance. Shouldn’t we be trying to measure how teams play instead of who they play?
Wins Above Bubble (WAB) is a different way of evaluating the strength of college basketball teams. Popularized by Seth Burn, WAB calculates the expected winning percentage for an average bubble team in each game of a team’s schedule and then subtracts that total from the team’s actual number of wins.
For example, if an average bubble team would have won 18 games against Team X’s schedule, but Team X only won 17 games, they would have a WAB of -1. By ranking teams by WAB, you can theoretically see what teams are most deserving of making the tournament based on their proven results.
Remember when I said that of all the advanced metrics found on team sheets outside of the NET, all but one had SFA in the low 100s? That “one” is ESPN’s Strength of Record, which, like WAB, attempts to rank teams based on how they performed against their schedule. In Strength of Record, SFA ranks 43rd, which would place them just inside the tournament field.
This is consistent with other WAB formulas. Burn tracks WAB based on KenPom, Massey, and Sagarin ratings, and in a blended average has SFA 42nd with a WAB of 1.04, meaning they’ve won roughly one more game than the average bubble team playing their schedule would be expected to.
Bart Torvik sees them slightly more favorably, ranking SFA 40th in the nation with a WAB of 1.4. All of these WAB formulas place the Lumberjacks ahead of Purdue, Cincinnati, Providence, and other bubble teams. Even a team like Texas Tech thought to be in a safe position by most ranks below SFA, hovering around a WAB of 0.
On the other hand, bubble teams that WAB views as more safely in the field include Wichita State (Burn average of 2.07), East Tennessee State (2.06) and Texas (1.91).
Now, I don’t think WAB is perfect. It’s a relatively simple metric in that it only takes into consideration who a team played, where they played, and whether or not they won. There are questions to be asked on how much focus we should place on things like margin of victory, recent performances in conference games, and how injuries and suspensions impact how we view specific games. Additionally, win probabilities depend on the rating system used to judge teams, which is why different WAB formulas produce different results.
Still, while the importance of other factors may change over time, WAB is consistent and apolitical, and I would like to see it used more heavily by the selection committee. Does that mean Stephen F. Austin is worthy of an at-large bid? While it’s debatable, I would say yes. Either way, they deserve to be considered, and at present, they likely won’t be if disaster strikes in the Southland Conference Tournament.
Connor Groel is a writer who studies sport management at the University of Texas at Austin. He also serves as editor of the Top Level Sports publication on Medium, and the host of the Connor Groel Sports podcast. His book, “Sports, Technology, and Madness,” is available now. You can follow Connor on Medium, Facebook, and Twitter, and view his archives at toplevelsports.net.