Ken Pomeroy explains why his stats model likes Gonzaga as No. 1

GLENDALE, Ariz. – There exist folks who are fans of college basketball but are not nearly so fond of advanced statistical models.

These people may wonder why their arguments about top teams and who has really earned what NCAA Tournament seeding are frequently interrupted and ended by self-satisfied references to something called “Kenpom.”

If they should happen to be Gonzaga fans, or disbelievers, this has been happening more frequently in 2017.

“Kenpom” refers to a website, www.kenpom.com, named for its founder, Ken Pomeroy. The website, which requires a mild annual fee to be fully enjoyed, contains various advanced statistics tracked by Pomeroy and predictive models.

It also contains blog posts that seek to find meaning in the madness of college basketball and answer questions such as, “Who is the best in-game coach?” (Steve Fisher); “Are some players more clutch than others?” (No); and “Should teams foul when tied?” (It’s complicated).

At its core is a ranking of all 351 NCAA Division-I basketball teams. At the top of the ranking is Gonzaga. According to Pomeroy’s model, Gonzaga has the No. 1 defense in the country, the No. 14 offense and ranks No. 115 in “luck,” a category too complicated to explain here.

Pomeroy, who is in Phoenix this week for the Final Four, took some time to explain why his model is bullish on the Bulldogs.

“They’re a tough team to rate,” Pomeroy noted. “Any team that kind of dominates a lower-level league is difficult to rate because what we know is that they’re much better than those teams in the league, but how they compare to top level teams is more difficult to ascertain.”

The schedule structure of college basketball also makes high-performing midmajors like GU difficult for statistical models to profile. GU began the year ranked No. 22 in Pomeroy’s rankings, based on its returning players, what his models knew about transfers like Nigel Williams-Goss and Johnathan Williams’ past performances, and the expectations for recruits like Zach Collins.

“I don’t go in and tweak the system at all,” Pomeroy said. “It’s based on what similar situations have happened previously. It assesses the value of a transfer based on where they played before, and recruits based on the impact they had coming in. All those evaluations are based on history.”

Despite a 23-point win in the opener, the Zags actually fell to No. 28 in Pomeroy’s model after their first game because the model expected a larger win against Utah Valley.

The Zags quickly rose in Pomeroy’s model by beating highly regarded teams like Arizona, Florida and Saint Mary’s, attaining the top spot after pasting Santa Clara on Jan. 19.

The trouble for Pomeroy’s is that those teams have improved – Arizona was without star player Allonzo Trier – but the level of GU’s competition waned significantly once the Zags rose to the top.

“The way my system works and I think most analytical systems work is they assume Arizona was the same team the whole season,” Pomeroy said. “Obviously, they weren’t and I think that’s an area for further refinement in systems. It’s an issue when you’re dealing with quality wins from a long time ago, and my system probably doesn’t do a great job of handling that, so that’s an area for potential errors.”

Pomeroy’s website does more than rank teams. It ranks players, too. Based on Pomeroy’s numbers, GU point guard Nigel Williams-Goss was the fifth-best player in college basketball this season. Kansas’ Frank Mason, the Associated Press Player of the Year, finished No. 6 in Pomeroy’s rankings.

Pomeroy’s model values Williams-Goss as a player who is proficient in many aspects, if not dominant in any. For example, Williams-Goss plays 78.8 percent of the possible minutes he could play. That figure ranks No. 383 among all Division-I players.

His steal rate of 3.1-percent is No. 143 nationally, while he commits just 1.9 fouls per 40 minutes, which ranks 86th.

“He’s been very efficient. He’s a very efficient point guard on one of the best teams in basketball, so all those factors come into play,” Pomeroy said. “The ranking is determined on his efficiency and his role in the offense. He also has a high steal rate so those numbers grade out, and GU having the No. 1 defense in the country also helps with that.”

Looking at Pomeroy’s numbers can provide insight into GU’s Saturday matchup against South Carolina in the Final Four. The Gamecocks hold opponents to a lower field-goal percentage than all but a few other teams. Pomeroy has written in the past, however, that 3-point field-goal percentage defense may not be a predictive statistic.

That is not to say that some teams do not defend 3-pointers better than others, as Pomeroy explains.

“It’s more based on style of play. When you have a really good defense, like both (Gonzaga and South Carolina) have, opponents are going to take lower-quality 3-point shots for various reasons,” Pomeroy said. “They can’t take quality shots at the rim, so they take more 3s as a result, or they run up against the shot clock. I don’t know that 3-point percentage is a mirage for either team, but it has less to do with what people focus on: people contesting the 3s.”