2019 March Madness Bracket Model: Why You Should Fade Duke, UNC & Buy Virginia
Photo credit: Geoff Burke-USA TODAY Sports. Pictured: Virginia basketball
- A perfect March Madness bracket has the best combination of probability to win plus contrarian picks.
- Using power ratings and ESPN bracket data, I devised a metric to help identify the best leverage spots to help you win your 2019 NCAA Tournament bracket pool.
- We'll start with some takeaways, then get to the full list of best value picks in order.
We’ve written a lot this week about how to use our power ratings to help you fill out your bracket. We not only have win probabilities for every first-round game, but probabilities for every team to make each round.
We’ve also written a lot about how to create contrarian brackets that will stand out over your peers. Being contrarian is important: If everyone in your work pool takes Duke to win the title, you don’t gain anything by being right. If everyone but you takes the Blue Devils and they lose, you gain a lot.
Also, your level of contrarian-ness should depend on the size of the pool. In the ESPN challenge, for example, you need to be much more contrarian because you’re competing with so many people. In a small family league, you can go more favorites and just pick a few select contrarian spots.
But remember that contrarianism for its own sake doesn’t help: Taking Iona to beat UNC would be very contrarian, for example, but it’s also so unlikely that it’s still an unwise pick.
Thus, the best way to build a bracket is to blend value (win probability) AND contrarian picks. Thankfully, we have all that data available to us since ESPN releases pick percentages for all teams and all rounds, so let’s create a metric to narrow down all of the possibilities.
The following “Leverage Score” is just a combination of probability to make X round plus how many people picked them to do so. It’s on a 0-10 scale with 10 being the highest leverage pick in this year’s bracket currently.
The results? Take the chalk, just not the super-popular chalk.
Let’s start with three main takeaways before getting to the entire table at the bottom.
3 March Madness Bracket Model Takeaways
1. Fade Duke and UNC
The thing that stands out the most is that Duke and UNC are way overvalued in brackets. Yes, they have some of the best odds to win the championship, but our simulations have those numbers as way lower than the percentage of brackets in which people are picking them to do so.
As such, the best way to gain an edge on the field is simply to not select Duke or UNC to make the Final Four. And again, that doesn’t mean you have to select a longshot to make it. Since Duke and UNC are overvalued, some of the best teams are also reaping the benefits of that valuation.
For example, the highest Leverage Score to win the championship is with Virginia, which literally ranks first in KenPom’s adjusted efficiency margin metric. The data suggests the Cavaliers are the best team in the nation — and they’re the most undervalued. That’s quite an opportunity. Take advantage of people worrying about them after last year’s embarrassment to UMBC and take Virginia to make a serious run.
Again, you can still have a Final Four that includes some of the best teams in the field. Using our data, this is the current most valuable iteration (including likelihood plus pick data from ESPN):
Two 1-seeds and two 2-seeds!
2. Fade Trendy Picks Like Villanova
There’s always a few trendy picks each year, either because analysts hype up mid-majors or people rely on name recognition. That seems to be the case with a team like Nova, which obviously has huge name recognition after winning two of the last three titles.
Those teams are not similar to this year’s, however, and people are picking the Wildcats at a much higher rate than their odds to advance. For example, we’re giving them a 60.7% chance to win their first game; they’re being selected in 76.6% of brackets. We’re giving them a 22.0% chance of making the Sweet 16; they’re being selected to do so in 39.5% of brackets.
Other trendy Sweet 16 picks you should fade this year are Houston, Virginia Tech and LSU, among others.
3. People Overreact to 12-5 Upset Chances
Some of the best first-round betting values are with 5-seeds like Wisconsin and Marquette, as the 12-5 upset narrative hit the mainstream long ago. It’s true that a 12-seed often wins (all but three years since 2005), but the odds of that happening in any single game are much lower than the ESPN percentages.
Murray State, a trendy 12-seed with a future NBA stud in Ja Morant, is absolutely one of the trendiest double-digit seed in the tournament. And while they might have the best chance among all the 12-seeds, it’s not 42.2% like the bracket percentages suggest.
All-in-all, the best way to create a value bracket with contrarian picks is not to fade all the good teams. Rather, it’s to fade the big narratives and still go chalk around those select choices. Fade Duke, UNC and trendy cinderellas, and instead take really good teams like Virginia, Kentucky, Gonzaga and Wisconsin.
The Best Value Picks for Your 2019 March Madness Brackets
Note: The table is interactive, so you can sort or search for teams/situations. For example, to see the highest Leverage Scores to win the championship, type “win championship” in the search bar and it will auto-sort.