Upsets make March Madness special.

In 2018, Virginia was the top overall seed heading into the tournament. As a 21.5 point favorite, they lost to UMBC. This was the first and only time a 16 seed beat a 1 seed in the men’s NCAA tournament.

The only thing better than watching upsets is predicting upsets. How can we do that?

The easy answer is looking at predictive analytics. My member numbers have the following double digit seeds as favorites in the Round of 64:

• Michigan (11) over Colorado State (6) by 3.2 points.
• San Francisco (10) over Murray State (7) by 2.0 points.

In addition, my predictions also suggest that two 10 seeds with excellent shooting statistics on offense will keep the game close.

• Michigan State (7) will beat Davidson (10) by 2.2. points.
• Ohio State (7) will beat Loyola Chicago (10) by 2.1 points.

But that seems unsatisfying. How does one pick games in which a big underdog wins in the NCAA tournament?

To explore upsets, I looked at games in which an underdog of six or more points won. This gives a sample of over 400 games this season.

To get a feeling for this data, you can ask some simple questions. First, underdogs that win slow down the tempo. This strategy maximizes the variance, which is beneficial to the dog.

However, underdogs of six points or more do not slow down the game. This season, they average about a half possession more than their season average. Over the past three season, there is no evidence that the tempo slows down in these upset games.

A second hypothesis is that underdogs shoot more three pointers. However, this also turns out to be false. Underdogs of six or more points take 1.7% fewer of their field goal attempts from 3. I saw this decrease in each of the last three seasons as well.

So what happens in these upsets?

As I discussed on an episode of Bracket Wisdom, the biggest reason for upsets is three point shooting. Underdogs of six or more points that win make 5.5% more of their three point shots than their season average.

Let’s estimate how this impacts the scoreboard. This season, college basketball teams averaged 21.8 three point shots per game. An increase in 5.5% shooting implies 3.6 points more per game.

In addition, favorites of six or more points that lose shoot 5.5% worse from three than their season average. There is about a seven point swing from three point shooting in these games.

To predict big upsets, you need to predict three point shooting, an almost impossible task. As I discussed in an episode of Bracket Wisdom, a team’s three point field goal percentage on offense has almost no ability to predict the future.

While three point shooting is a skill, randomness plays a big role in three point shooting percentage. This makes predicting big upsets as difficult as keeping Tom Brady off the football field at 45 years old.

I’ve always preached going strictly by the numbers in the early rounds and spending your time analyzing which team to pick as champion. This research suggests you double down on that advice.

Data driven betting information

This article was sent to The Power Rank’s email newsletter. This is a free service that strives to be:

• Valuable
• Concise
• Entertaining

You also get The Power Rank’s March Madness cheat sheet that makes it drop dead easy to fill out your bracket.

To get this free service delivered straight into your inbox, enter your best email address and click on “Sign up now!”