Last year, Navigate’s SVP of Analytics & Innovation, Matt Balvanz, published his predictions for the most likely D1 Men’s Basketball teams to make the Sweet Sixteen, using a model he created that factored in certain performance metrics.

This year, rather than using Matt’s model to simply predict the Sweet Sixteen, we’re putting it to the test against one of our team’s biggest college basketball fans – SVP of Client Strategy, Jordan Bloem. Matt’s model filled out an entire bracket and will compete head-to-head with Jordan’s predictions in a winner-take-all showdown.

At Navigate, we love data, but we know that data is only as good as the interpretation. We’re curious to see if Matt’s scientific approach can triumph over Jordan’s intuition and years of experience playing and watching basketball.

The Madness Behind The Scientific Approach

The model Matt created last season analyzes 12 factors for determining a school’s likelihood of advancing in the NCAA Tournament:

  1. Strength of Schedule
  2. Points per Game
  3. Opponents Points per Game
  4. Field Goal %
  5. 3-Point %
  6. Free Throw %
  7. Offensive Rebounds per Game
  8. Total Rebounds per Game
  9. Assists per Game
  10. Blocks per Game
  11. Turnovers per Game
  12. Personal Fouls per Game

One major difference this season is that Matt applied the model’s findings to the entire field of tournament teams AFTER the bracket was set, while last season’s predictions were made prior to Selection Sunday. Based on the way teams were seeded, it would have been impossible for all 16 predictions to be correct. Two of the teams selected by the model (Michigan & LSU) ended up playing in the Round of 32, which meant that rather than 16 possible correct predictions, he was left with only 15. Of the 15 possible teams, Matt’s model correctly predicted eight (53%) of them.

It’s also important to remember that last year’s tournament was a particularly exciting one, filled with upsets and unforeseen victories. The average seed of the teams in the Sweet 16 was 5.88, compared to an average year’s 4.45, making 2021’s tournament 32% crazier than normal. We can’t be sure exactly how much that extra madness impacted the model’s success or if we’ll see a return to normalcy in 2022.  If you want to read more about how he did, check out last year’s recap here.

This year, Matt went through the bracket game-by-game, and chose the team that his model preferred. He did not make alterations to the model, nor did he let his own personal fandom or biases impact his final bracket. There were some picks that Matt found personally surprising, such as Illinois losing to Chattanooga in the first round, but the Mocs had a strong statistical performance this season, so the model picked them to win that round.

Navigate March Madness - Matt Balvaz's Bracket
The Science – Matt’s Picks
The Art of Experience 

Jordan is an avid fan of Michigan Wolverines college basketball and a casual fan of college basketball more generally across the country, so he typically approaches March Madness with a baseline set of biases regarding who he sees as good or bad, as well as overrated or underrated. His impressions are typically formed by the one or two times he catches a team play during the regular season, so it’s not the most scientific approach. With this in mind, he’ll usually fill out a first draft bracket based solely on intuition.

Once the first draft is done, he likes to dig into some research to make any necessary changes. Jordan points out that he tries to include one or two 12/5 upsets, or perhaps some 13/4 upsets or 11/6 upsets.  Generally, these upsets are picked after exploring various advanced metrics systems like KenPom.com or BartTorvik.com. Most years there are a few of these 12/5 or 11/6 games where the higher seed is actually the better ranked team in the advanced metrics (or at least it’s a toss up), so Jordan will use those to help guide some of his upset picks.

Similarly, he likes to look for a few 3, 4 or 5-seeds with the potential to make a deep run to the Final Four and he tries to identify those using some of the same advanced metrics. Is there at team that KenPom thinks is 7th or 8th in the country but somehow ended up with a 5-seed? That might by Jordan’s Cinderella.

Finally, if it’s a large bracket pool, he tries to go against the conventional wisdom to pick his champion. For example, with a lot of people picking Gonzaga this year, Jordan made sure to pick another highly ranked team (Arizona) to win it all even if he thinks Gonzaga really does have the best chance to win. This ensures that if  Jordan’s pick is matched up with the conventional wisdom pick in the Final, it’s more likely that he’ll win the whole pool if his pick wins that one game because he’ll have fewer competitors that also picked his champion.

This season, Jordan employed a combination of strategy and intuition. He chose Arizona as his winner – partly because they were ranked favorably by KenPom – but mostly because he watched them beat his Michigan Wolverines by 18 (in a game that felt like a 30+ point blowout to him at the time).

He picked Iowa to make the Final Four for 3 main reasons:

  1. With the momentum of winning the Big Ten tournament, he thought they were under-seeded as a 5-seed.
  2. He felt they were in the region with the weakest 1-seed to contend with in Kansas.
  3. He wanted to troll his rival in this competition, since Matt is a huge Iowa fan.
Navigate March Madness - Jordan Bloem's Bracket
The Art – Jordan’s Picks
The Competition

Now that the brackets have been filled out, we’ll track their progress through a standard points scoring system that awards 1 point for a correct pick in the first round, with point values doubling in the following rounds. It’s the age-old battle between art & science, and at Navigate, we’re excited to see which side wins.