Early Season Luck

Posted by on December 27, 2013

With 11 or 12 games under most teams’ belt, efficiency numbers are starting to stabilize. However, random variation still plays a role in team performance. Defensively, we know that opponent three-point percentage is especially “lucky”. Ohio State, the nation’s number one defense to date, is allowing opponent’s to shoot just 25% from three. Some of that is certainly the Buckeyes perimeter defense, but data tells us that this number will regress as the season continues.

For this post, I wanted to look at how defenses are performing regardless of opponent 3P%. To do this, I ran a regression between AdjD and opponent 3P%. The r-squared value is .35. In other words, 35% of AdjD (defensive points per 100 possessions) can be explained by changes in opponent three-point percentage. The regression allowed me to get a predicted AdjD (xAdjD) based solely on 3P%. The next step is to subtract xAdjD from actual AdjD to get a 3P%-neutral defensive rating (the residual values, in stats terms).

Of course, this method has some limitations. First, opponent 3P% isn’t completely luck. Zone heavy teams like Syracuse have consistently been able to force their opponents into bad threes. Second, opponent 3P% isn’t adjusted for strength of schedule (unlike AdjD). Still, I wonder if regular AdjD or this new method is more predictive of future defensive success. Take a look at the top 10 3pt-adjusted defensive teams:



Cincinnati is currently ranked seventh in AdjD, but take over the number one spot when adjusting for threes. Villanova and Virginia both stay at number two and three. Texas takes a huge jump to just crack the top 10. Of the top 50 teams in AdjD, here are the biggest gainers and losers:




Syracuse has even been unlucky defensively by a regular team’s standard, and are probably due for some good luck. Iowa’s length makes them a pretty good candidate to continue holding opponent’s to sub par shooting nights. It’s also interesting to see Stanford’s defense ranked so highly here. Their offense hasn’t been as good as I would have expected before the season, but it looks like their defensive ceiling may be fairly high.

I’ll save this post for later in the season and see how the lucky/unlucky teams have changed.



One Response to “Early Season Luck”

  1. RK Anderson

    Good work. I’m trying to remember from Ken’s opus’ on 3 point defense what if any link there was between 3FGA and AdjD? Not sure how that would, or if it should, fit into this analysis…For a team like Syracuse, it probably matters a great deal more since they concede a lot of 3′s.

You must be logged in to post a comment.