Bracketeering Sweet 16 update: The Infallible Braculator agrees to never speak of this past weekend again

By: Richard W. Sharp. Graphics: Patrick W. Zimmerman.

Remember, thou art mortal

The 2018 NCAA Men’s Basketball Tournament began on the Ides of March, and by now everyone who put together a bracket based in any way on reality is feeling like they got stabbed in the back. Upsets have laid waste to the South and West. In the Midwest, Syracuse has toppled Michigan State, delivering the coup de grâce to Obama’s bracket. Closer to home, I kissed my 10 bucks goodbye when Cincinnati bit the dust.

Damn the torpedoes, full speed ahead! How is the strategy (chalk + winning streaks + a dash of random noise) holding up under pressure? Did the Infallible BraculatorTM produce a significant number of brackets that can withstand the onslaught?

Definitely maybe.

Scoring the Brackets

The good news is that we still might end up with several decent brackets.

We scored each bracket according to a standard scheme in which a correct pick receives 1 point in round 1, 2 points in round 2, 4 points in round 3, and so on, up to 32 points for the championship. We also calculated the maximum number of points that each bracket could possibly achieve once the incorrect picks are eliminated.

Mouseover for details.

Mouseover for details.

The box plots above show the range of scores in each of the scenarios we introduced at the beginning of the tournament. Only the scenarios with a random noise component show a spread of outcomes. The deterministic scenarios, such as the Poll of Polls (aka chalk), correspond to only a single point.

The Braculator’s numbers aren’t great, but they are looking a lot better (median hovering a little above 100) than the Poll of Polls (max 76).

The Upsets: Et tu, Buffalo?

The greatest upset of all time was the Washington Generals’ 100-99 win over the Harlem Globetrotters in 1971….until last Friday night, when #1 Virginia fell to 16th seeded UMBC.

By the end of the weekend, two #1 seeds, two #2 seeds, two #3 seeds and three #4s were packing their bags and heading home. All nine of these teams should have made it to the Sweet 16.

What impact did these losses have on bracket performance? We put the upsets in perspective by listing each upset along with the difference in pre-tournament team ranks (Massey rank), the margin of defeat, and the difference between seeds. We also took a look at the number of points each upset cost the Poll of Polls. Virginia/UMBC leads the way across the board, and UNC and Michigan State piled it on. Interestingly, Xavier doesn’t factor heavily despite being a #1 seed. Also note that the Poll of Polls didn’t lose any points when Florida St. beat Missouri because the model didn’t consider this an upset and called the game for Florida St.

Mouseover for details.

The histograms below start to reveal the effect of each upset on each family of randomized brackets. By selecting the losing team in an upset with the filter at the bottom of the chart, you can see the distribution of points lost for each scenario (the default setting is Virginia).

Mouseover for details.


Sic transit gloria mundi.

About The Author

Richard is a Seattle area data scientist who builds predictive models and the services that deliver them. He earned a PhD in Applied and Computational Math from Princeton University, and left academia for the dark side of science (industry) in 2010, following his wife to the land of flannel. Fan of coffee, beer, backpacking and puns. Enjoys a day on the lake fishing, and, better, cooking up the catch for a crowd.

No Comments on "Bracketeering Sweet 16 update: The Infallible Braculator agrees to never speak of this past weekend again"

Leave a Comment