March Madness: Using Trends to Narrow Down the Eventual NCAA Champion for 2021

Sure, your natural instincts and lucky jerseys may play a part in which teams win the NCAA Tournament each year, but at numberFire, we're more inclined to be looking at the statistical trends that the champs have in common to help determine winners.

That and our detailed game-by-game projections and bracket tools, of course.

Do defenses really lead to championships? Or is an elite offense actually a better indicator of championship success after all?

Who knows? We don't -- not yet at least. But we will after we run through some historical data to see common threads between champions.

And then, of course, we'll figure out which teams in the tournament have the common traits of winners.

Seed Data

Since 2000, the lowest (i.e. "worst") seed for a champion belonged to the Connecticut Huskies back in 2014 when they were a 7 seed. Other than that, the other 18 winners between 2001 and 2019 were a 3 seed or better.

We won't rule out the lower seeds just yet, but teams who aren't a 3 seed or better don't historically win this thing.

And while we're at it, the 2014 Huskies were underrated. The average 7 seed has had a nERD score (or expected point differential against an average opponent on a neutral court) of 11.64 points since 2000. Those Huskies were a 14.06, 2.42 points better than your average 7 seed.

The average 4 seed, historically, has had a nERD score of 14.22, just 0.12 points better than Connecticut. So, really, UConn was a 4 seed listed as a 7 seed.

Either way, there's a soft cutoff for a 3 seed.

nERD Data

Because nERD actually indicates team strength rather than simply trying to seed teams based on whatever criterion the committee is using to do so, we can get a better gauge of where winners come from.

Of course, we'd expect teams seeded 1 through 4 to have high nERDs (and they virtually always do), so it makes sense that the lowest nERD by a champion did belong to the 7 seeded Huskies back in 2014.

The next lowest nERD by a winner was by -- coincidentally -- UConn in 2011 (with a nERD of 14.86).

What this means, then, is that while a 7 seed has won a tournament in the past 20 years, no team has won with a nERD lower than 14.00. That'll cut off a lot of squads this year, even with parity dragging down a lot of the nERD score across the country.

Perhaps a more practical way to look at it: no team has won the NCAA Tournament by ranking outside the top-20 in nERD, and only the 2014 Huskies were worse than 13th. The 2011 Huskies were 13th. The other 17 winners over the past 19 years were top-10 in nERD.

Entering the tournament, only eight teams have a nERD of at least 14.00. Due to the overall balance we're seeing this year, we'll use the top-20 as our initial cutoff.

Offense and Defense

We've seen elite defenses win the NCAA Tournament, which shouldn't surprise us. Good teams win this thing.

Of the 19 winners since 2001, 8 of them had a defensive rating in the 95th percentile or better in the country, and 11 of them were in the 90th percentile or better. Just two squads -- the 2018 Villanova Wildcats (80th) and 2011 Huskies (70th) had defenses below the 84th percentile.

However, those numbers don't really compare to the offensive side of things.

Of the past 19 champs, 16 of them had offensive ratings in the 95th percentile or better, and only one team (again, the 2014 Huskies) had an offensive rating below the 91st percentile. They were in the 81st percentile.

So, the low-water marks for these numbers become an 81st-percentile offensive rating and a 70th-percentile defensive rating (though it should really probably be closer to 91st- and 84th-percentiles, respectively, if we remove the outliers).

We can also use raw numbers if it's easier to think about first: the lowest offensive rating for any champion over the past 10 years has been 108.9 for the 2014 Huskies. The worst defensive rating over the past 10 years has been 99.2.

Teams That Fit

If I filter our database for teams who meet certain criteria, here is who fits.

Top 20 in nERD
Here are the top 20 teams by nERD entering the tournament.

If we stuck with the 7-seed-or-better criteria, as well, that would rule out the Wisconsin Badgers (8) and Loyola (IL) Ramblers (9).

Offensive Rating of 108.9 or Better
These teams have an offensive rating of at least 108.9, via Sports-Reference.

Defensive Rating of 99.2 or Better
I'm not sure how to handle this one because almost every team (51 of them) in the big dance has a non-terrible defensive rating, and it's not as important as offensive rating for deciding champions. Via Sports-Reference, these are the top 8 seeds with a defensive rating worse than 99.2: Ohio State Buckeyes, LSU, Iowa Hawkeyes, Villanova Wildcats, West Virginia Mountaineers, and Oregon Ducks.

That leaves most of the teams in play from a defensive standpoint.

Teams That Fit
These teams are inside our top 20 and fit those offensive and defensive cutoffs.

If we wanted to lean on the top-3-seed narrative (again, 18 of the past 19 were 3 seeds or better) and the 14.00 nERD cutoff (all of the past 19 winners had a nERD of at least 14.00, then we're down to these squads:

It should be no surprise that it's the chalk.

But if we want to go back to the percentile ranks for offensive and defensive rating that we uncovered, too, only one team fits the following marks: 3 seed or better, top-20 in nERD, nERD of at least 14.00, 81st-percentile-or-better offensive rating, and 70th-percentile-or-better defensive rating.

That's the Houston Cougars.