Did a flaw in predictive metrics hide UCLA’s Cinderella potential?

This year’s tournament had plenty of “madness”. Per the Associated Press, the cumulative seed numbers of the Sweet 16 teams was 94, the highest total since the tournament expanded to 64 teams in 1985.

To a large extent, the NCAA Tournament is always going to be unpredictable, but was there any way to see any of this coming? The predictive metrics (KenPom, T-Rank, etc.) are generally a decent place to turn to for a slight edge. Most were quite very high on Loyola Chicago for example, who made them look prescient with their upset victory over 1-seed Illinois.

These metrics obviously aren’t perfect, however, and they’re even less so due to an imperfection most — if not all of them — have (FiveThirtyEight might be the exception). While they account for factors like margin of victory and strength of opponent, they don’t account for game-to-game personnel changes — particularly injuries. If a team’s best player get injured for example, it won’t fall in the rankings until its performance warrants it.

For the record, I imagine there’s good reason for this. While a star player going out for the season would seemingly have a significant impact, it’s difficult to assess just how much. These metrics aim to remove subjective factors from the rankings equation, and manually dropping a team due to an injury would be antithetical to everything they “stand for.” That said, this doesn’t mean college basketball fans shouldn’t take injuries into account when filling out their brackets. It’s just important to keep in mind that the predictive metrics won’t be doing so.

Which teams were most impacted by injuries this season — and potentially underrated by the predictive metrics as a result? From the 68 teams in the NCAA Tournament, there were 13 players that met the following criteria:

  • >= 30 minutes per game
  • >= 10 points per game
  • Entering the NCAA Tournament, had played in less than 90 percent of their team’s games

Ayo Dosumnu: Illinois went 3-0 in his absence and looked extremely impressive as they did so.

Were the Illini underrated in predictive metrics due to Dosunmu’s absense? Probably “No.”

Marcus Sasser: Houston went 3-0 without him this past season, winning each game comfortably (77-67, 88-55, 112-46).

Were Cougars underrated in predictive metrics due to Sasser’s absense? Probably “No.” Despite Houston’s impressive Final Four run, there’s not much reason to believe his missed games led to the team being underrated by the metrics.

Justin Smith: Arkansas went 1-3 in the four straight games he missed in January. Although the Razorbacks had a blowout victory over Georgia, they fell to Missouri, Tennessee and Alabama (81-68, 79-74 & 92-76). Smith played limited minutes (18) in his first game back vs. Alabama, and Arkansas dropped this game as well. The forward was the 6th-most valuable player in the SEC per evanmiya.com, and his absence was clearly felt.

Were Razorbacks underrated in predictive metrics due to Smith’s absence? Probably “Yes.” Even prior to the NCAA Tournament, the Razorbacks were clearly a different team with him on the floor (21-3 with him, 1-3 without). When one takes this into account, Arkansas’s Elite 8 run is a little less surprising.

Cade Cunningham/Avery Anderson: Oklahoma State went 2-1 without Cunningham, offsetting an 81-66 home loss to Baylor with an impressive 85-80 road victory over West Virginia. The Cowboys went 3-0 without Anderson, with none of the results standing out as unexpected.

Were the Cowboys underrated in predictive metrics due to Cunningham and/or Anderson’s absences? Probably “No”.

Kevin McCullar: Texas Tech went 7-2 in the nine games he missed to start the season, losing to the two best opponents it played (Houston 64-53 & Kansas 58-57). McCullar would have helped the cause, but it’s not clear how much. While the guard averaged 10.4 points this past season, he only shot 46.5% on two’s and 28.3% on threes. Additionally, it’s worth emphasizing that he missed the first nine games of the season. Given that most predictive metrics put more weight on recent games, early season absences have a not negligible, but lesser impact on a team’s metrics. KenPom handles it the following way: “In a 30 game schedule, Game 1 will weigh about 40% as much as Game 30, assuming equal significance.” 

Were the Red Raiders underrated in predictive metrics due to McCullar’s absence? Probably “No.”

Will Richardson: Oregon went 9-3 in the 12 games he missed to start the season, falling to Missouri (84-75), Colorado (79-72) and Oregon State (75-64). The Ducks played A LOT better when Richardson returned, and the guard’s contributions in a hefty 35.5 minutes per game (11.3 points, 40.3% from three) definitely contributed to this. Like McCullar, he missed games in the first half of the season, so Oregon’s ranking should have “caught up” during the team’s hot play down the stretch. Still, not only is Richardson likely a better player than McCullar, he only played in 53.8% of the Ducks’ pre-tournament games. None of the other 13 appeared in less than 60%, so regardless of when it happened, Richardson’s absence was quite significant.

Were the Ducks underrated in predictive metrics due to Richardson’s absence? Probably “Yes.” Much of the college basketball community seemed to feel the same way, as Oregon was a popular pick to upset Iowa and reach the Sweet 16. Well, this is how it played out.

James Bouknight: The guard missed eight games spanning January and February, during which the Huskies went only 4-4. Uconn was clearly a much better team with its best player healthy

Were the Huskies underrated in predictive metrics due to Bouknight’s absence? Probably “Yes”. It didn’t matter in the NCAA Tournament, however, as UCONN fell in the 1st Round to Maryland.

Geo Baker: Rutgers went 3-0 in the three games Baker didn’t play early in the season. The team performed quite well, recording a 79-69 home win over Syracuse.

Were the Scarlet Knights underrated in predictive metrics due to Baker’s absence? Probably “No.”

Tyrece Radford: The Hokies went 3-1 in four games Radford missed in January and February. A disappointing 83-72 road loss at Pittsburgh was offset by an impressive 65-51 home win over Virginia.

Were the Hokies underrated in predictive metrics due to Radford’s absence? Probably “No.”

Johnny Juzang: UCLA’s NCAA Tournament star missed four games to start the season, during which the Bruins went 3-1. Along with two comfortable victories, the team lost 73-58 at San Diego State and needed three overtimes to beat Pepperdine. UCLA didn’t play great without Juzang, but given he missed games early, this shouldn’t have had THAT big of an impact on the Bruins’ metrics. It’s worth noting that Juzang did miss one other game, however — a 64-63 home loss to USC in March.

Were the Bruins underrated in predictive metrics due to Juzang’s absence? Probably “Yes.” Big picture, Juzang’s missed games probably didn’t have THAT big of an impact. He did miss five games, however, more than any of the 13 players except for McCullar (tied with Jason Preston). The answer should probably be “Probably “No””, but given how amazing Juzang was during UCLA’s Final Four run, it’s difficult to ignore the possibility that his absence led to the metrics underrating the Bruins.

Buddy Boeheim: He missed three games in early December. The Orange went 2-1, including a 79-69 loss at Rutgers. As Syracuse’s best player, Boeheim definitely would have made a difference in this game…but it’s just one game.

Were the Orange underrated in predictive metrics due to Boeheim’s absence? Probably “No.” Syracuse made a surprise Sweet 16 run, so it’s tempting to say Boeheim’s missed games caused the metrics to underrate the Orange. While it’s definitely interesting to consider, I can’t quite get there.

Jason Preston: Ohio went 3-2 in the five games Preston missed in the middle of the season. Given the guard’s value to the team, it’s definitely possible Ohio would have been a bit higher in the metrics if he hadn’t missed any games — the fact the Bobcats managed to upset Virginia makes this an especially tempting conclusion. Ohio’s 13-5 record with Preston in the lineup wasn’t overly impressive, however, so I don’t think the circumstances quite make the cut.

Were the Bobcats underrated in predictive metrics due to Preston’s absence? Probably “No.”


Whether or not these players’ absences led to their teams being underrated by predictive metrics is definitely subjective. That said, it’s interesting how many of their squads outperformed expectations in the NCAA Tournament. Coincidence? Maybe. It’s interesting to analyze nevertheless.

(Credit to KenPom and sports-reference for statistics)

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.

As you found this post useful...

Follow us on social media!