# Nylon Calculus: What stats best correlates to fan attendance?

Fans want wins. How do I know that exactly? As any man would do, I referred to the smartest person I know: my wife.

She touches on two aspects, the latter being more important than the former as you’ll soon see. Fan experience is important, but it’s generally safe to assume that most stadiums and teams have figured out how to entertain its customer at some fundamental baseline. Fans can buy beers, participate in shooting contests, or watch a lady on a unicycle balancing dishes. Unless the people managing the NBA game event really push halftime shows to the ultimate edge with a bear fight, there’s not much else to be done to improve the experience that can bring more fans into stadiums.

Gimmicks are fun, but — as Jackie Moon eventually figured out during his Tropics tenure — winning is everything. Fans pay and work to help get their team pumped, and want to be repaid with a win — and corndogs if their team scores a ton.

I also noticed what my wife didn’t say as well. She only mentioned winning and not necessarily big names or budding stars. Call her and others fairweather, but she’s still a potential customer that an NBA team would want to buy a ticket. She and all fans demand wins.

So, I wanted to explore attendance more closely. The following piece expands a bit upon this research paper while attempting to bring in 19 seasons’ worth of data instead of just two. However, without getting too bogged down in academic largess, I tried to distill the most important findings in visualizations.

## When do fans like to attend games?

The research paper above looks over just two seasons, so I expanded the following view to 19 seasons. There’s a clear drop-off on games across the NBA held on Mondays, Tuesdays, and Wednesdays. A five percent difference might not be saying much but it’s a hefty load of cash.

For example, take the Capitol One Arena that can host 20,356 Washington Wizard fans. An 88 percent fill percentage means that only about 17,913 people are attending the game. Using the chart above as a general guideline, the same game played on Saturday assumes that exactly 1,000 more fans attend the game.

A rough gauge using 2013-14 prices suggests that a thousand fans in D.C. amount to \$64,000 in ticket sales. Multiply that number over the course of a season’s worth of Wednesday and Saturday games, and that’s a hefty difference between the two days. Should every one of those fans also purchase a hotdog and a beer, the Wizards would stand to lose at least \$74,830 from missed ticket and food sales.

As far as specific time slots are concerned, it seems as if the 2:00:00 PM game time should probably be ended. These games are usually never on national TV and apparently don’t provide that great of assurance on profitability. I’m sure these games could be stacked better as a doubleheader but that kind of talk starts to but up against TV politics, which is a whole other discussion.

## Score a bunch? Or dominate? Both.

At this point, you might be wondering which teams have been noting their attendance numbers to be

over 100 percent full

.  It’s disingenuous for me to speculate, but most teams of those dots represent the Dallas Mavericks.

This is still happening now, but I digress.

SRS was one of the stronger indicators of fan attendance. If a team looks like it’s winning, the fans come, and it really is as simple as that.

The plot above covers the attendance of every game from 2001 to the end of the 2018 season. It tells us that fans love points! While there isn’t a huge difference in games with combined scores of 250 over 175, there remains a slight difference. The biggest takeaway with information like this is to suggest to teams that are hemorrhaging for attendance should just start scoring. At the very least, a bad team that scores a bunch can bring in fans and make money while ever exciting “development” occurs.

## Player classification

I attempted to use Jacob Goldstein’s Wins Added scale to classify players in every game since the 2001 NBA season. I did this because the scale classifies players based on statistical merit instead of a classification based off of fan voting. We saw above that fans tend to show up to games when a team’s SRS is strong or when scoring is up. Generalized All-Star votes give great insight into who’s popular globally, but those votes are largely done online and beget outside influence. Specifically, a lot of the votes can come from China or from cities without NBA teams and thus aren’t representative of the ticket buyers at these games. I could be totally wrong.

To review the Wins Added scale:

What I want to see is if there are more Best Players and MVP Candidates in a game, more fans will tend to be in attendance. Such an observation should correlate more strongly than fan influenced All-Star voting. I compiled the following correlation coefficients based on a number of Pearson tests done in R to try and get the best overall picture of the data. All p-values fell well below 0.05.

• Simple Rating System: 0.48
• All-Stars from fan voting: 0.34
• *Games with Best Players: 0.13
• Games with Best + MVP: 0.20
• Games with Best + MVP + All NBA: 0.26
• Games with Best + MVP + All NBA + All-Star: 0.21
• Games with Best + MVP + All NBA + All-Star + Starter: 0.25

*The more that there are Best, MVP, etc. players, the more games there are that are close to capacity.

Alas, a wrong hypothesis. Turns out, fan-based voting is a better indicator of NBA game attendance than a metric indicative of in-game play.  This could be that fans choose All-Stars of teams that are already really good in general and have a strong SRS.