Do Matchups Matter? (Part 3 – Press Teams)

Posted by on January 15, 2015

The last two seasons I’ve tried to take a look at “opponent compatibility” and how matchups affect game outcomes. My first attempt broadly concluded: “the best way to predict the winner of a game appears to be just picking the better of the two teams“. Last year I put a whole bunch of different ideas to the test with some interesting results.
Today I’m taking a look at teams that press on defense. Take VCU for example. Any pre-game analysis of a VCU game is going to deal with how the opponent is likely to respond to VCU’s havoc. Teams with inexperienced guards or general turnover problems are naturally going to be viewed as extra vulnerable, but do the numbers back that up?
The methodology is exactly the same as my first two installments. First, I identify specific games where the matchup occurs. For this post, I obviously had to define a press team. This is always very arbitrary, but I chose teams from 2010-2014 who press on more than 25% of total defensive plays. This left me with 22 teams (and 705 games) listed below:

Now with a 705 game sample to work with, we can find all the instances where these teams play different types of opponents. I use kenpom adjusted efficiencies to then calculate an expected points per possession regardless of the matchup. Comparing the expectations to the actual results is the final step to attempt to quantify the effects of the matchup.

Are presses just a gimmick?

First let’s start off by taking a look at just how committed these teams really are to the press. When push comes to shove, are they consistently picking up full-court against very good offensive teams? The best way we can check this is by taking a look at turnover percentage:

Obviously we wouldn’t expect defensive TO% to be the same against good and bad offenses. Still, good offenses saw their turnover percentage rise from a season average of 17.3% all the way to 20.1% in games against press teams. It seems pretty clear that in order to qualify as a press team in the first place these teams are not simply using it as a gimmick against inferior opponents.

Do press teams struggle against low turnover teams?

Now that we know that press teams do a pretty good job of turning over good offenses we can look at how this affects overall efficiency. I think the current narrative is that a press team will be exposed against an offense that takes care of the ball. Let’s look at the numbers:

There’s always a trade-off between sample size and a well-defined matchup. Ideally we would like to draw conclusions from more than 140 games, but that would come at the cost of lowering our standards for good/bad TO% teams. With the sample size caveat in mind, notice we see that press teams actually have performed slightly worse than we expect from kenpom adjusted efficiencies alone against both extremely good and bad TO% teams. However, the differences here a very small. Good TO% teams scored .009 points per possession more than we would have expected against press teams. In an average 65 possession game, that only amounts to about a half of a point.
The findings here seem to point to what I’ve found about many different types of opponent compatibility: Just picking the better team is usually the way to go. Still, not finding an effect is much different than an effect not existing. It’s also still possible a highly trained basketball eye could do a better job of identifying what teams are likely to struggle against press than my current methodology.

What are the effects of a press team not forcing turnovers?

This last part looks at the idea backwards. Instead of identifying a potential mismatch¬†before the game (like above), I’m going to look at the post-game turnover percentage. When a press team doesn’t get turnovers (against any type of opponent) in a given game, how badly does that hurt their defense?

It’s no surprise to see press teams performing significantly better in games where they force many turnovers. In those games, opponents scored .073 points per possession below what would have been expected pre-game by kenpom numbers. That amounts to roughly 4.7 points in a game. In games where the press team didn’t force turnovers, opponents scored .055 points per possession above expectations (about 3.5 points over the course of a game).
Turnovers are undoubtedly important for press teams, but it appears to be lazy analysis to just blindly pick press teams when playing against inexperienced or turnover prone guards. Think about the Oklahoma-WVU game earlier this week. The Sooners have a very strong backcourt and don’t struggle turning the ball over. Yet for one reason or another “Press Virginia” was alive and well against OU, turning them over on more than 30% of possessions.
Look for more matchups to be put to the test in the future. If you have an idea you would like me to take a look at leave a comment or let me know on Twitter.

You must be logged in to post a comment.