The following is part of a series of NCAA tournament posts. You can find the intro to the series here.
Narrative: Results vs. good teams are more important than bad teams.
I first took a look at this idea back in early January. At that time, Louisville had put up historically good numbers against bad teams while losing both games against quality competition (Kentucky and UNC). Louisville had also just dismissed Chane Behanan. Many national writers were actually removing Louisville from the list of Final Four contenders.
In order to win an NCAA tournament you usually have to beat at least four or five really good teams. It makes some sense that results against quality competition would be more predictive than results against weaker competition. However, this idea is generally taken to far. Crushing bad teams is at least somewhat predictive of tournament success (more on that below). In Louisville’s case, they had only played two games against quality competition back in January. When you only have two games to go by, it’s a no-brainer to instead look at the full resume.
I took a look at pre-tournament efficiency differential over the last 10 years. I defined “weak” competition as teams ranked outside the KenPom top 100 and “strong” competition as teams ranked in the KenPom top 50. Take a look at the graphs below:
The sample sizes here for each type of game are roughly the same. On average, top-5 seeds play about 12 pre-tournament games a year against both “weak” and “strong” opponents. However, I do wonder if there’s more variability in the amount of games played against strong teams. This might be a reason for why results against strong competition was only marginally more predictive than weak competition.
It has been consistently proven that margin of victory is more predictive than simply win percentage. Margin of victory allows to actually take some meaning out of non-conference games against Final Four contenders and mid majors. The competition narrative appears to be very overemphasized.
Since my initial post, Louisville has continued to demolish inferior opponents. The most extreme example was a 61-point win over Rutgers in the AAC conference tournament. Against stronger AAC teams, the Cardinals have played much better than they did against UNC and Kentucky. Apparently well enough for public perception to almost completely change about Louisville.
The 4-seed is now a very popular Final Four pick. Still, much of the Cardinals high KenPom rating is due to crushing bad teams. Let’s take a look at the 10 teams in recent history similar to Louisville:
Based on seeding along, we would expect the 10 teams above to win about 24 games. They actually won 27 games. However, the most recent example of a team like Louisville is Indiana from last year. The Hoosiers struggled against Temple and then lost to Syracuse in the Sweet 16.
We are at the point where stats and eyes have somehow come to an agreement on Louisville. They still haven’t proven all that much against top tier teams, but an AAC tournament championship was apparently enough to change the narrative surrounding Louisville.
Russ Smith and company weren’t given any favors with their tournament draw, but there’s little reason to think right now (or at any time this season, in reality) that Louisville isn’t one of the top five teams in the country.