NCAA announces additions of Torvik, Wins Against Bubble metrics for Tournament seedings

The NCAA announced Thursday that they are diving even deeper into the analytics world when it comes to selecting the perfect 68-team NCAA Tournament bracket.

Their own tool, the NET rankings, has been around for several years now and offers a numbers-driven look at each team. However, the NCAA is also utilizes some third-party analytics sites. ESPN’s BPI (Basketball Power Index) and Ken Pomery’s website, often known as KenPom, are already incorporated. But starting in 2025, so will the numbers from Bart Torvik.

Torvik does some fascinating analytical work in the college hoops world, and his rankings, as well as his Wins Against Bubble stats will be used as a tool by the committee in 2025. The NCAA announced that in a release on Thursday, saying:

“The evaluation of teams will be enhanced by the addition of two metrics as the committee voted to include the Torvik and Wins Against Bubble rankings on the team sheets.”

Dan Gavitt, the NCAA’s senior vice president of basketball, added:

“The committee has always valued different data points and metrics to assist with its evaluation process, and these two metrics have increasingly been referenced by members in recent years.”

Now, they’ll officially be a part of the process.

What are Torvik, Wins Against Bubble metrics?

Bart Torvik has a nice blog post on his website detailing exactly how he ranks teams as part of his T-Rank model, his base rating system. You can view that here.

Torvik admits his rating system is very similar to KenPom in relying on basic points-per-possession data on each side of the ball. There are a few differences, though, and addressing how Torvik implements recency bias.

Years ago, the selection committee made it a point to emphasize the more recent games at the end of the season when it came to evaluating NCAA Tournament resumes. More recently, though, the committee has gone away from that and flattened the value of games to where the wins at the beginning of the year mean as much as the wins at the end of the year.

Torvik’s rankings diverge from that philosophy. Here’s how: “There’s a recency bias,” writes Torvik of his metrics. “All games in the last 40 days count 100%, then degrade 1% per day until they’re 80 days old, after which all games count 60%.”

Whether they realize it or not, by incorporating Torvik’s model into the selection process, more value could be placed on more recent performance, since that’s how Torvik’s numbers slant.

He also notes the other difference is the diminished value of major blowouts. Once a game is well in hand for the much better team, Torvik doesn’t view that “garbage time” as quite as important as the more competitive parts of the game, and therefore discounts the numbers from garbage time.

Wins Above Bubble

Finally, we’ll get to Wins Above Bubble (WAB), a metric Torvik uses throughout the year which constantly projects the NCAA Tournament field and evaluates how many wins above or below the theoretical cutoff line a team is.

For instance, last season, eventual national champion and No. 1 overall seed UConn finished first in his WAB stat with +11.2 wins. No. 2 overall seed Houston was third at +10.0 and No. 3 overall seed and runner-up Purdue was second at +10.6. So, this metric had the top teams figured out.

Going further down to the bubble, a team like Indiana State would have greatly benefitted from Torvik’s inclusion. The Sycamores rated No. 41 overall in his rankings and were the highest-scoring WAB team to not make the dance, posting +1.9 wins above the bubble.

Overall, in 2023-24, Bartb Torvik’s Wins Above Bubble metric was slightly more favorable than reality as eight teams who rated in the positive missed the tournament. Meanwhile, just two teams in the negative made the dance as an at large seed: Texas A&M (-.06 WAB) and Michigan State (-.19 WAB).

That’s pretty darn accurate.