APBRmetrics

[Chilltown]

Today I published the results of my aforementioned study on NBA road team attendance. I'm going to link to the full post because its too long to copy/paste, but I wanted to discuss some of the main conclusions here.

Like other work by Dave Berri and Paul Holmes on the subject had found, we found that the main determinants of road attendance were the winning percentage of the team and the number of All-Stars on the team from the previous season. Unlike Holmes and Berri, we found that win-loss record had a great magnitude of effect (although it was a weaker predictor). Because Winning percentage was not linearly related to road attendance, we introduced a (wpct)^(3/2) variable into the regression (the exponent was found using a log-log regression of wpct on road attendance). We also clustered the standard errors at the team level to allow serial correlation between team observations over the years of data.

The more interesting conclusion, in my opinion, is that "marginal fans" in NBA cities appear to come out in larger numbers as the home team's chance of winning decreases. Because we used a fixed-effects regression model, we were able to control for the "brand name effect" that the more well-known/wider fanbase teams might have. Thus the effects represent fans that come to see a team based on its characteristics, not its name. I found this conclusion to be somewhat surprising, and think it merits further study.

[schtevie]

Interesting stuff. Can you offer comment on whether you believe significant competitive inequality is profitable from the standpoint of the league as a whole? From the standpoint of the average team, in a static sense, given a choice between playing two home games against teams of average strength vs. one weak and one strong team, the latter is clearly preferred. Eyeballing the graph shown, and let's say that the visiting teams clock in at 25% and 75%, there is approximately a 2% drop in attendance offset by a 3.8% (?) gain. So, if average attendance is 17,000 and average ticket price is $50, the home team gets an additional $15,300 for the inequality. Multiply that out over the season, and you are looking at not large but non-trivial revenue differences. Enough to pay for a stats guy or two.

Does the supposed positive, inter-temporal relationship between attendance and team strength offset these gains?

Also of interest to me is the 'luxury tax' variable that I didn't see mentioned in the commentary. It is marginally statistically significant, but does it have any economic significance? What does the 0.9 represent? Is it perhaps multiplied by the tax paid by the visiting team in millions? If so, Mark Cuban is quite the rainmaker for the NBA, first paying the tax and then increasing ticket sales.

[greyberger]

I now intend to measure all NBA-related sums of money by the number of stat guys that amount could hire. I.e the Nuggets have offered Carmelo Anthony a 3-year, 812 Winston extension.