Monday, March 29, 2021

Can We Measure Fan Loyalty By Looking at Disconnect Between NFL On-Field and Financial Performance?

One day in the late aughts, I walked into a grocery store the day after the San Jose Sharks had lost in the Stanley Cup finals. I passed a guy walking out with a Sharks jersey on, who reacted well to my statement that today, we can tell who the real Sharks fans are. After our short exchange I thought: presumably a fan wants their team to win right? In economic terms, fans identity with teams to the extent that they win and increase prestige. Could it be that we can define "fanbase loyalty" as the irrational "price inelasticity" shown by some fans or entire fanbases, as this fellow was doing? (More here on loyalty as irrationality, vs the psychopathy of casual fandom. If you're not offended by this post, you're likely a fellow psychopath. Welcome.)

Of course I recognize how naive it is to get upset that money influences professional sports. Indeed it's really irritating when people cry about their favorite sports stars signing with another team for (gasp!) more money. But if you count yourself as a sports fan, should it NOT be about better games? You as a sports fan want to see good games, and don't care about the league making money, right? (Seinfeld said that really, sports fans are cheering for clothing; I think really they're cheering for market sizes and marketing executives. Inspiring!)

METHODS: To measure the impact of NFL team performance on their financial performance, I took as my inputs the CHANGE in performance in the 2019 and 2020 seasons, both in terms of win record (ties count as half a win) and Elo rating per 538. I compared that to CHANGE in team valuation and total revenues in both 5/2020 (which reflects the 2019 season) and 12/2020 (which reflects the 2020 season.) Notes on methodology - using the CHANGE in quantities has the benefit of providing an internal control for market size, stadium size, local sports susceptibility, etc. and things like inflation and COVID apply to all teams. (Perhaps not equally - the more athletically successful the team, the more revenue they get from stadium sales - more on this later.) Sources for this data below. (Side note, I found three articles in Forbes and Yahoo claiming to list franchise value, but oddly they all gave the same numbers in August 2019, September 2020, and December 2020 - so I used the 18 to 19 changes assuming that reflected performance during the 2018 season.

RESULTS: Comparing team performance measured by Elo with valuation, there was a very weak positive correspondence (improving Elo means improving revenue) with R^2 of 0.0827; no clear outliers.

The same could be said for Elo and revenues, with an even weaker R^2 of 0.029. Taking out 3 apparent outliers did not improve this.

Using the lower granularity, more context (league) dependent measure of win record, the absolute change in number of games won from 2018-2019 was a better fit at R^2=0.1959 (more absolute wins, better valuation) but there was essentially no relationship with revenues - which is a better reflection of fan engagement. There was a similar but slightly weaker result using the input of improvement in record compared to last year's number of wins* against valuation improvement (R^2 of 0.116), and essentially no relationship between this record-improvement statistic and revenues. Again, fans don't care whether the teams win - at least not in a way that affects their spending the following year.

Of note, the Chargers and Rams are new teams so we might expect them to be off-trend, that is with revenues increasing regardless of record, and that's the case - but Indianapolis has no such story and is the next-most off-trend in terms of "better revenues despite getting worse."

*(Reasoning for trying this statistic instead of absolute increase in wins - is the fan impact of a team going from 8 to 10 wins more like going from 6 to 8 wins, or from 4 to 5?)

(I also wanted to compare performance to Nielsen ratings but couldn't find a good data source - and even if I had raw viewers per game, there's the methodological question of how to credit the teams.)

A loyalty index for each team would require tracking the relationship betweeen "worsening athletic performance" against revenue improvement over some period of years. Although I stated above that revenues is a more immediately sensitive indicator of fan engagement, over time one assumes that a constantly losing team MUST lose value - so as a proxy, I looked at current valuation against the 2010-2018 W/L %. Again, the better you play, the higher tha valuation, with an R^2 of 0.0739. Dallas is a major outlier here, though removing them does not make a major differencein the fit. I think Dallas's "inelasticity" (the irrational exuberance and/or broadness of its fanbase) is especially impressive when you consider it's not exactly in a population center (see NFL loyalty map here) - it's not just benefiting from a big market like the Giants.


NFL teams vary in the amount of their income they get from stadium revenues. Interestingly the graph of percentage of revenues coming from stadium events against win-loss % from 2010-2018 looks very much like the graph for absolute valuation. Again, the Cowboys are way off trend, getting almost 2/3 of their revenue from game days. One imagines NFL executives use people showing up in person as a proxy for team sustainability. One shudders to think of the price of a beer.


Profit, and sustainability over time. Fairness and exciting games are merely a means to that end. That's obvious - businesses focus on profit - but fans are often loathe to think of their tribe as a business in this way. Yet it's obviously true. The coefficient of variation for Elo rating is 31 and 49 times more than valuation change and revenue change, respectively. And profit sharing, a draft system based on performance, and lack of free agency are all things trying to keep this balance. (In much the same way that college football's bizarre insistence on equally bizarre preseason rankings is really just a way of deceptively seeding high-ticket-sales legacy teams into bowl games. Always think about the money. The games are just a vehicle for making money.


There is little connection from season to season between a team's performance and a fan's loyalty, as measured by annual revenues.

Over the longer-term, there is some connection betweem a team's performance, judging by stadium revenues and (more distantly overall franchise valuation) against performance for almost a whole decade. This does appear to differ between franchises, but measuring this is beyond the scope of this post.

Darwin wrote a whole book called On the Origin of Species and of many questions he did answer in that book, that was one he did not. So don't judge.


Revenue by franchise reflecting 2019 season (Aug 2019) here Revenue by franchise reflecting 2020 season (mid-Dec 2019) here Franchise valuations reflecting changes after 2019 season (Sept 2020) here 2010-2018 win-loss average here Elo ratings at