These 3 High Variance Teams Could Be March Madness Sleeper Picks

February 8, 2012 - by John Ezekowitz

Imagine it’s the third week of March, and you are staring at your blank bracket.

You want to correctly pick some first round upsets, maybe a couple 12 seeds over 5 seeds, or 11’s over 6’s. Or maybe you think a 3 or 4 seed will lose. Upsets happen in every tournament, but how will you know which ones to pick?

Of course it is impossible to know with certainty which teams will pull upsets, but there are certainly factors that can help predict which teams are more likely to pull upsets.

One set of factors that I believe does not get nearly enough attention is the variance of a team’s performance during the regular season. Teams with higher variance are capable of playing extremely well, but are also capable of terrible games. Lower variance teams play at roughly the same level throughout most of the season.

Why Is Variance Important?

Think about it this way: Team A and Team B both face a First Round opponent where they are expected to lose by 10 points. Team A is much more consistent: their variance is lower and most of the time their performance does not deviate more than eight points from their expectation. Team B, however, has a higher variance: their “confidence interval” is much larger and they can deviate around 12 points from their expectation.

	Expectation	Lower Bound	Upper Bound
Team A	-10	-18	-2
Team B	-10	-22	+2

In this concocted example, Team B has a greater chance to pull the upset, but also a greater chance to lose by a larger margin. Since margin of victory does not matter in NCAA pools, if you were picking an upset, you want the team with the greater chance to actually win the game. Thus, you’d pick Team B.

Determining Which Teams Are Most And Least Consistent

I calculated the offensive and defensive variances for every team in college basketball by looking at each game’s Points per 100 Possessions scored and allowed — in other words their offensive efficiency and defensive efficiency. Those numbers were adjusted for the strength of the opponent by dividing by the opponent’s average performance over the course of the season. The Offensive and Defensive SD numbers below represent the standard deviation in Points per 100 Possessions for each team.

Although we still have weeks to go until Selection Sunday, it’s not too early to look at teams that potentially might be in the bracket on the 11-14 seed lines with an eye towards their variance. Using The Bracket Project’s collection of over 50 online projected brackets, I examined these teams and found three that have the variance profile to be able to pull upsets, and two that might not be able to elevate their games enough against tougher competition.

Potential Sleepers (High Variance Teams)

Cincinnati Bearcats (TR Predictive Power Rating Rank: 57, Offensive SD: 15.5, Defensive SD: 12.4): The Bearcats have been one of the surprises of the conference season so far, rebounding well from the brawl with Xavier. They have been extremely variable on offense, putting up great performances against good teams like Villanova, Georgetown, and UConn, but also throwing out stinkers against Presbyterian, St. John’s, and Rutgers (all under 0.9 PPP). Cincinnati has clearly shown that they are capable of beating better teams. They might be an attractive 11-6 upset pick come March.

Cleveland State Vikings (TR Rank: 58, Off. SD: 12, Def. SD: 13.6): Cleveland State has quietly become the best team in the Horizon League, and they’ve done it with defense. The Vikings force a huge number of turnovers, and have held six opponents to 0.80 PPP or lower. The Vikings have the ability to be an extremely good defensive team, and that skill can win a game against a better opponent in the NCAA tournament.

Oral Roberts Eagles (TR Rank: 55, Off. SD: 14.7, Def. SD: 12.4): The Golden Eagles are one of the nation’s best-kept secrets. Dominique Morrison is a fantastic player, and the ORU offense has been unstoppable on some nights. Oral Roberts has put up more than 1.3 PPP on offense five times this season. While their defense is not great, they have shown the ability to simply blow teams away offensively. That ability is rare in a team you would expect to see on around the 13-seed line.

Probably Not Cinderella Material (Low Variance Teams)

Minnesota Golden Gophers (TR Rank: 52, Off. SD: 8.5, Def. SD: 11.3): The Golden Gophers have been very consistent on offense this year. They have only scored more than 1.2 PPP once, and that was with star forward Trevor Mbakwe, who is now out for the season. In 11 of their games, they have scored between 1.05 and 1.15 PPP. That might work well in the regular season, but against the better competition they will face in the NCAA tournament, that limits their known performance ceiling quite a bit.

Nevada Wolf Pack (TR Rank: 88, Off. SD: 12.8, Def. SD: 9.5): Nevada has run up a gaudy record (19-4), but their best win is probably Washington, who has been extremely disappointing this year. The Wolfpack play solid defense, but have not shown the ability to lock anyone down. They have not allowed fewer than 0.9 PPP to any of their top 150 opponents. While Nevada has shown that they do not allow a huge number of points (their worst defensive efficiency is 1.14 PPP), they are also unable to raise their defensive game against better opponents. This tight band of performance limits their ability to pull an upset over a 4 or 5 seed in March Madness.