When I was younger, long before I divided my points by possessions, I remember a particular Sports Illustrated for Kids article about the NCAA tournament catching my eye. The article was something along the lines of: “10 things to remember when filling out your bracket”. The article was clearly effective, I remember even saving it for the following March. Today, I can only recall one of the author’s bracket tips. The general idea was:
“National champions tend to wear blue uniforms.”
We all understand that national champs wearing blue is a case of correlation and not causation. Give Albany some brand new blue uniforms (seriously, they could use some new uniforms) and they are not any more likely to beat Florida. There’s nothing special about blue, it just so happens Kentucky, Kansas, UNC, and Duke all have it in their uniforms. Those schools win a lot because they are really good.
Obviously the article was mostly just a fun look at the tournament for the young sports fan. However, the irony here is that most NCAA tournament analysis doesn’t go far beyond telling you to pick blue uniforms. For example, think about the 5-12 upset. Emphasizing the historical stats of 5-12 upsets is not much different than emphasizing blue uniforms. Sure, it’s useful to know that a 5-12 upset happens a lot before the bracket comes up. Once the bracket comes out, we can do much better than picking based off historical seeding (correlation). We can, you know, actually look at the two teams playing in the game.
There’s a reason for this poor level of NCAA tournament analysis: Picking winners in a single elimination basketball tournament is hard. Picking exciting (non-chalk) winners in a single elimination basketball tournament is really hard.
With all this in mind, the next logical step in NCAA tournament analysis would probably be to break down individual matchups. For example, how does New Mexico State’s size matchup against San Diego State’s athleticism? While I certainly enjoy this type of basketball analysis, the jury is still out on how useful it really is. When I’ve tried to look at opponent compatibility, the answer is usually to simply pick better of the two teams.
I wouldn’t be writing this introduction if my NCAA tournament picking solution was to simply go to KenPom and pick the higher rated team in each game. That’s certainly not a bad strategy (adjusted efficiency stats get you most of the way there), but we can probably do better.
John Ezekowitz (and Luke Winn) put together an extremely interesting NCAA tourney analysis by assuming there are factors (experience, consistency, etc.) beyond just KenPom efficiency differential that are specific to tournament success. Factors like these are often very popular tournament narratives. Some of these narratives are worthwhile, and probably more are not.
I took a look at four general March Madness narratives and four specific teams in 2014 that would potentially be affected by each. The idea here is to bring something to the table regarding team strength beyond simply efficiency differential. The four narratives and corresponding links are below:
Narrative 1: Regular season results in Feb/Mar are more important than Nov/Dec.
Narrative 2: Regular season results on the road are more important than at home.
Narrative 3: Defense wins championships.
Narrative 4: Regular season results against quality competition are more important than poor competition.