NCAA Tournament: 5 Potential Sleepers Based on Consistency

Inconsistency is often frowned upon in sports, but high-variance strategies can actually be very useful for March Madness spoilers.

If your bracket pools are anything like mine, you’ve noticed every year there seems to be “that guy.”

It’s the person who does not pick any upsets, rolling with chalk from the first round to the Final Four.

While it may not be the worst strategy, given how seeding tends to correlate with tournament wins, it is also a boring one.

A big part of the magic of this time of the year is the underdog success stories, and there is no feeling like nailing an upset pick.

Sometimes, though, we need to resist the urge to have a bracket that is too upset heavy. The 12 seed-5 seed upset can be a popular one, but picking four 12s is probably ill-advised, considering 12 seeds are 35-69 against 5 seeds since the tournament expanded to 64 teams in 1985.

11 seeds have actually fared a bit worse in the first round during this span, going 33-71. 13 seeds, meanwhile, have gone 22-104 and 14 seeds are 16-88.

10 seeds fare a bit better, but their 42-62 first-round record still only translates to a 40.4% winning percentage.

So here we have the conflict between our desire to pick lower seeded teams and the fact that such teams struggle (they’re upsets for a reason).

We can try to reconcile this by looking at some factors that can make an upset more likely.

Though we do not know what the individual matchups will be, for now, we can look at teams that shoot a lot of three-pointers, generate turnovers, play a slow pace, and do other things that allow randomness to play a larger role in the outcome.

Over at our team rankings, we have a Consistency rating that, as you might expect, looks at the predictability of each team’s performance.

Generally speaking, you’d like the favorite you pick to be consistent and for the underdog selections to have a wider range of outcomes.

Consider a hypothetical underdog that is very consistent. That team's average level of play probably won’t be good enough to beat a better team.

However, take another fictional team with a wide range of outcomes. That team will probably still lose and has a better chance of losing big, but while their floor is lower, their ceiling is also higher.

This is the team that, other things being equal, has the better chance of pulling off an upset.

With that in mind, here are some real-life teams that fit this mold.

Butler Bulldogs

Inconsistency Rank: 1st (Most inconsistent)
Projected Seed: 10

The Bulldogs’ inconsistency derives mostly from inconsistent three-point defense, and this is most evident in their best win and worst loss.

In December, Butler defeated Purdue, who is ranked 12th in the country in our power ratings, and held the previously unbeaten Boilermakers to 7-for-22 (31.8%) from beyond the arc (their season average is 36.8%).

Against Marquette, though, who beat them by six in January, the 102nd-ranked Golden Eagles were 10-for-17 (58.8%) from three-point range (their season average is 34.2%).

For the season, Butler has allowed opponents to shoot 33.6% from deep, which is actually better than the national average. The problem is they allow a three-pointer on 36.9% of field goal attempts allowed, which ranks 245th nationally.

As evidenced by the Purdue and Butler games, this can be a double-edged sword. Three-point shooting percentage is incredibly random on both sides of the ball, and if defenses want to prevent three-pointers, the best way to do so is prevent them from being attempted in the first place.

Butler does not do this, choosing instead to play what Ken Pomeroy appropriately calls the “lottery.” 

When the Bulldogs play, it is almost a given they will give up a ton of three-pointers. A hot opponent can rout them, while a cold one could put themselves in danger of an upset.

Stephen F. Austin Lumberjacks

Inconsistency Percentile Rank: 99th
Projected Seed: 14

The Lumberjacks have been a trendy upset pick in their last two tournament appearances and rewarded their backers in 2014 with a win over 5 seed VCU.

Their aggressive defense is a big reason why, and this is the third season they have ranked in the top-five nationally in Turnover Rate. SFA is actually number-one in the country this season, as opponents have coughed up the ball on 23.7% of their possessions.

This is one of the ways the Lumberjacks shorten games, which is good for an underdog. The fewer possessions an opponent has, the fewer opportunities they have to show they are better.

The popular example is you, the average reader, playing a one-on-one game against Stephen Curry. Your chances of winning are better if you play first point wins, where you could randomly make a single shot, than if you play to 21.

SFA does this in other ways, too. Aside from stealing possessions from opponents, they excel at extending their own, ranking in the 86th percentile nationally with a 34.5% Offensive Rebounding Rate.

Finally, despite the high-pressure defense, the Lumberjacks play at an ultra-slow pace, ranking 279th in the country in average length of possession, according to KenPom.

Going for turnovers and crashing the boards can both backfire if they don’t work, but if they do, they are great ways to spring an upset, something SFA fans can attest to.

Vanderbilt Commodores

Inconsistency Percentile Rank: 95th
Projected Seed: OUT

We currently have the Commodores outside the Field of 64, but other models are more optimistic.

At the Bracket Matrix, which is a composite of various tournament projections, they are a 10 seed. If they do make it in, they would be a solid bet to pull off an upset.

The Commodores are ranked 19th in our power ratings, coming in at 14.27 points above average. Other statistical rankings are high on them as well, despite their 19-12 record, as they are 23rd at KenPom, 19th in the Sagarin Ratings, and 23rd in ESPN’s BPI.

They are also tied for 29th nationally in raw average margin of victory, outscoring opponents by 9.8 points per game.

So how did they go 19-12? A dreadful record in close games is the culprit, as Vanderbilt went 1-5 in games decided by two buckets or fewer. Only three of their losses were by a double-digit margin, coming in road games against Kentucky (9th in nERD), Purdue (12th), and Texas (31st).

By contrast, 16 of their 19 wins came by double digits, and a one-point squeaker against Florida (36th in nERD) was their only victory in which the margin was smaller than seven points.

By our model, there are only 18 teams in the country that have been better than Vanderbilt, making them a good “upset” pick for the simple fact that a first-round win probably wouldn’t actually be an upset.

In terms of consistency, there is little in terms of their style of play that would seem to explain it. They make 39.0% of their three-pointers and allow opponents to shoot only 29.1% from deep but are not particularly reliant on three-point shooting, nor do they allow a high rate of such shots (110th nationally in Three-Point Attempt Rate on offense, 10th on defense).

They don’t force turnovers on defense (346th in Turnover Rate), and they don't crash the boards on offense (263rd in Offensive Rebound Rate). They rank in the 54th percentile in Pace.

Still, for Vanderbilt, the standard deviation in terms of their average scoring margin was 17.5 points. Given they average out to one of the better teams in the country and have a high ceiling, they could be the double-digit seed that is most likely to make a deep tournament run.

They just have to make the field first.

Yale Bulldogs

Inconsistency Percentile Rank: 93rd
Projected Seed: 12

The Bulldogs are going dancing for the first time since the Kennedy administration.

If they want to build off their one-and-done performance in the 1962 tournament, it will likely be the product of aggressive offensive rebounding.

The Ivy League champions are seventh in the country with a 39.3% Offensive Rebounding Rate, despite only ranking 154th in average height and 228th in effective height (the average of the center and power forward position), according to KenPom.

This means they are forced to take risks in terms of crashing the offensive glass (a tactic which was more than worth the gamble in the slow tempo Ivy League, as only 19.7% of the shots Yale allowed came in transition, ranking 240th in the country, according to Hoop Math).

This slow tempo itself -- Yale ranks in the 23rd percentile nationally in Pace -- also adds to the variance that could make the Bulldogs an attractive upset pick.

Connecticut Huskies

Inconsistency Percentile Rank: 90th
Projected Seed: 12

The Huskies are not unlike Butler in that they give up a ton of three-pointers, which can lead to a wide range of results.

They rank 264th in terms of three-point attempts per field goal attempt, and as you might expect, this has led to some good and bad outcomes.

In their 10 losses, they have allowed 19.3 three-point attempts, and opponents shot 37.3% on them. In their 21 wins, opponents took 21.6 three-pointers but made just 31.1%.

The one guarantee with UConn’s defense seems to be a high volume of threes, and the rate at which they go through the nets is anyone’s guess.