NCAA Tournament Teams You Should Favor Given the Public's Picks
Look, I don't want to put unnecessary pressure on the numberFire algorithms. And I don't want to seem like a giant back patter who only focuses on predictions that went right.
But, folks, this article has historically been kind of awesome.
The idea here is pretty simple: numberFire's algorithm spits out stage odds for every team in the NCAA Tournament. For example, it's currently giving Gonzaga a 42.3% chance to make it to the finals. Illinois, meanwhile, is second at just under 30%.
You could fill out your bracket by just picking those teams -- the teams with the best odds to make it to Final Four and championship round. That doesn't necessarily give you a leg up on your competition, though. Because, in the end, if you're just picking the favorites, you're not taking advantage of variance and the boost you'd get if one of those teams loses. And as we know, nothing's guaranteed in March.
That's where ESPN's "Who Picked Whom" data comes into play. By utilizing the public's picks and matching it with numberFire's math, we can optimize our bracket's expected value.
That's a hugely important exercise to go through if you're submitting your bracket to any sort of pool. The process has worked well historically, too, as I alluded to at the top. During the 2019 NCAA Tournament -- the last one we had since last year's was cancelled -- the numbers pointed to Virginia as the most optimal selection to win it all.
And they won it all.
The year before that, Villanova was the team that the numberFire algorithm loved versus the public's picks.
And they won it all.
In 2017, North Carolina won the national title. The expected value process wasn't on them, unfortunately, but it was on the loser of the National Championship game, Gonzaga.
What about this year?
The Final Four
According to numberFire's Stage Odds, the teams with the best odds to make it to the Final Four are Gonzaga (1 seed), Illinois (1 seed), Michigan (1 seed) and Ohio State (2 seed). That doesn't make them the best selections in medium- to large-scale pools, though.
| Team | Public's Pick | nF Odds | Difference |
|---|---|---|---|
| Gonzaga | 66.90% | 57.49% | -9.41% |
| Illinois | 53.50% | 45.30% | -8.20% |
| Michigan | 41.20% | 38.41% | -2.79% |
| Ohio State | 22.80% | 27.71% | +4.91% |
The numberFire odds give the Zags a 57.5% chance to make the Final Four, but the public is picking them at a rate that's over 9 percentage points higher. They may be an ideal choice if you're filling out your own bracket for pride, but if you're really trying to find an edge against multiple brackets, they're not necessarily the best choice.
Instead, you may want to focus on these teams:
| Team | Public's Pick | nF Odds | Difference |
|---|---|---|---|
| Iowa | 14.10% | 22.44% | +8.34% |
| Wisconsin | 1.30% | 8.08% | +6.78% |
| Purdue | 3.80% | 10.21% | +6.41% |
| Colorado | 2.10% | 8.17% | +6.07% |
| Houston | 15.80% | 20.98% | +5.18% |
| Ohio State | 22.80% | 27.71% | +4.91% |
| USC | 1.50% | 5.73% | +4.23% |
| Creighton | 1.40% | 5.46% | +4.06% |
| Colgate | 0.50% | 4.51% | +4.01% |
| Tennessee | 3.40% | 6.48% | +3.08% |
When looking at it with this lens, Iowa becomes an optimal contrarian pick. That intuitively makes sense, since they're in the same region as the top-seeded Gonzaga: if everyone's picking Gonzaga, then picking the second-best team in the region would provide some leverage.
You'll notice Colgate, a 14 seed, is listed above. That's mostly because the numberFire algorithm sees them as an underrated team. Their nERD score -- which measures the number of points we'd expect a team to win by against an average team on a neutral court -- is actually 24th-best in the country. Their liklihood of reaching the Final Four isn't great, but they're a pretty damn good 14 seed.
National Championship Game
The "Who Picked Whom" data clearly favors Gonzaga in the National Championship game -- 53% of brackets have them there. Like the Final Four exercise above, the numberFire algorithm sees that as the second-biggest mistake you can make when entering a relatively decent sized pool. With National Championship odds of 42.3%, that's a downgrade of about 11 percentage points when comparing the public's picks to numberFire's. Only Baylor (-15.8%) is worse.
| Team | Public's Picks | nF Odds | Difference |
|---|---|---|---|
| Iowa | 7.70% | 13.56% | +5.86% |
| Ohio State | 9.30% | 13.48% | +4.18% |
| Houston | 7.30% | 11.34% | +4.04% |
| Wisconsin | 0.50% | 3.58% | +3.08% |
| Purdue | 1.30% | 4.15% | +2.85% |
| Michigan | 14.10% | 16.72% | +2.62% |
| Alabama | 6.10% | 8.28% | +2.18% |
| USC | 0.60% | 2.78% | +2.18% |
| Creighton | 0.60% | 2.63% | +2.03% |
| Colorado | 0.50% | 2.39% | +1.89% |
Sensing a theme?
Iowa, once again, tops the list. One thing to point out here, though, is that the odd differentials aren't all that great. In the last NCAA Tournament, that "difference" number for the top team was over 17%. This tournament season, the public seems to be sharper. That means it may be less necessary to stray from numberFire's raw stage odds data.
National Champion
The last time we did this exercise, over 40% of the public had picked Duke to win the title. That made them an easy fade in most pools. The same can generally be said about Gonzaga this year. No team has a greater delta (on the negative side of things) between the public's picks and numberFire's odds than the Bulldogs. The team that's second-worst is Baylor, and then comes Oklahoma State, Kansas, and Texas.
On the positive end, these are the five teams that numberFire's odds like most compared to how the public is currently picking:
| Team | Public's Picks | nF Odds | Difference |
|---|---|---|---|
| Iowa | 3.50% | 7.91% | 4.41% |
| Ohio State | 3.40% | 5.47% | 2.07% |
| Houston | 2.70% | 4.50% | 1.80% |
| Michigan | 8.10% | 9.69% | 1.59% |
| Alabama | 2.80% | 4.29% | 1.49% |
To reiterate what was said above, things are much closer when compared to what we've seen historically. In 2019, Virginia's National Champion "difference" was almost 12%. When Villanova won in 2018, numberFire's algorithm gave them a 27.6% chance to take home the title, when the public was hovering the 14% to 15% range.
That's what makes this tournament so difficult. Iowa keeps popping up at the top of these lists, but you're not actually gaining that much of an edge by taking them. It's at least not as glaring.
In medium and large pools, there seems to be one clear and obvious expected value-related strategy this season: fade Gonzaga.
