Measuring Luck to Find Greatest All-Time NCAA Tourney Upset
Editor’s note: This article was co-authored by father-son duo, Wayne and Greg Winston. Two things they absolutely have in common is a passion for basketball and analytics. We hope you don’t let the numbers upset you too much.
It doesn’t take long for conversation in the Winston household to derail straight toward sports analytics. It’s exciting stuff. One thing that has always been interesting to the both of us is luck, particularly in basketball. One thing led to another and we ended up applying analytics inspired by Shot Plus Save Percent (SPSV%), used by the NHL, to basketball in the hopes of measuring luck.
This means that we used a team’s season field goal percent (FG%) and their opponent’s to determine what each team would be expected to shoot.
Once we did that, we adjusted for the strength of schedule to create an even more precise estimate of what both teams would shoot. With those adjustments, we were able to apply the difference between what each team was supposed to shoot and what they actually shot. This creates a decently precise estimate of how lucky each team was.
What does this prove? It’s in an effort to determine the biggest NCAA tournament upset of all time.
The pros of this method are clear. You get to see a deviation from the norm for both teams’ shooting percentage. The strength of schedule is based around teams taking roughly fifty shots per game, which doesn’t always ring true, but is still pretty accurate. In an ideal world, we would be able to use effective field goal percentage, since three pointers are a major part of upsets, and basketball in general.
Unfortunately, we couldn’t get our hands on that data for every team as easily. However, the only other flaw we must note in this model is the score of the game is increased based on shooting percentage. Meaning, a game like UMBC-Virginia would be amplified because of it being a blowout.
As you can see the biggest upset according to the formula is Villanova-Georgetown. Villanova had to hit over 75% of their shots to win that game. It’s only natural our formula makes that upset the winner.
Our data implies that the Villanova-Georgetown game and the UMBC-Virginia game are by far the most shocking upsets in the history of the NCAA tournament. Villanova because of how well they had to play to win, and UMBC because of how lopsided the game was supposed to be in the other direction.
The next closest game was a fifteen seed beating a two (Santa Clara-Arizona). For the people wondering why the NC State-Houston game wasn’t more of a historic upset, it was because NC State had played a significantly harder schedule that year.
The main takeaway: while the NCAA Tournament is always going to produce amazing and historic upsets, there are a couple outliers that far and away more amazing than others. The beauty of the NCAA Tournament is that it’s unpredictable. Yet, through the application of our theory, we can start to predict the likelihood of upsets before they happen and how devastating they would be.