Every Village Needs An Idiot

“He’s gonna go all the way!”

And I see someone has a “chip” on their shoulder on interception returns again, calling them an “inferior product”. *  I’ll let you draw your own conclusions about just which one of us that idiot might be. 

SOM PRO FOOTBALL LINKS

(This has been going on for fifteen years; I’m not interested in debating him. I haven’t seen as many complaints about Mike’s returns and he’s been doing this calculation now for at least five years.  The sensible thing would be to ask Mike nicely about looking at possible revisions of seasons that you might have disagreement with. But, no it’s easier to be just a hero, and complain.)

I’ll just drop these here (again), so you can see the full history- how the feature was designed, how it should be tested, how well it fits certain situations, it’s limitations, and why seasonal data will vary: 

https://onfredswatch.blogspot.com/2022/09/team-specific-interceptions-for-som-nfl.html?m=0

https://onfredswatch.blogspot.com/2023/01/the-second-of-three-parts-on-team.html?m=0

 https://onfredswatch.blogspot.com/2024/02/the-last-of-three-parts-on-team.html?m=0

https://onfredswatch.blogspot.com/2024/02/one-last-time-with-team-interceptions.html?m=0

On the surface a suggestion of just rolling for interceptions and then returns is stupid. Looking at 2025 Data: 

(Click on it)

Does it make sense, with a typical int return for a TD being about 40 yards or so, for a team like New Orleans which is in the bottom quartile of yards and average to be more threatening than the New York Giants, who had returns of 68 and 56 yards but did not score? Same for Tennessee, a little better average but still half of the number of the average for the New York Football 🏈 team? 

Let’s reorganize the data: 

(Click on it)

Well - now we see teams over 20 yards a return will probably get about 14 pct of those returns taken to the house.  Teams below 10 yards per return will score about 5 percent of the time. 

So the fact that New Orleans scored on 1/10 of their returns is to me not meaningless, but it’s a more or less random act of luck. Do we card to that?  If you look at the TDs for 1997 Barry Sanders (11/335 carries), and 1997 Tommy Vardell (6/32) do you card for Vardell to arbitrarily score from anywhere on the field one fifth of the time? 

No we don’t. We card for the attempts and yards and we presume gameflow or coaches choices will take care of the rest. We card to the threat.  Barry Sanders had 2000 yards that year. 

TD pcts on returns are proportional to the natural log of the average, that is, they are a hockey- stick graph going up at both the team and seasonal level. But that does not mean every season will track exactly to that line. Some will be above, and some below. 

(Note that 1977, The Season Of Much Hatred, had 28 TDs / 562 ints for 5 pct. That would imply an average below 10 yards per return- but 1977 averaged about 12.8 yards per return, which should be closer to 7 pct.  That would be 39 TDs.  Discussion of 1977 ignores the 1961 AFL or 1967 NFL, seasons where the league actually tracked higher than predicted. You could change the average, but Strat didn’t sign up for that.)

And every time SOM gets these complaints, I send them the data and that pretty much ends that. 

Now could Mike - who now works for them- improve on my prior work? Well yeah, maybe! He’s a pretty smart guy! But I’d ask nicely. 

Fred 

* It’s amusing because clearly the above data shows the method to be vastly superior to the original chart. His solution, therefore, is to just call both the feature and the original chart piss-poor.  Well you’ve got 69 (+3) seasons to propose, there, chip, but you might come up with a better idea than the one you have. That dog’s not gonna hunt. Neither is this one: 

-Fred