This is not directed towards you Peoria, but it always bothers me when football analysis borrow from statistical concepts to shore up their articles and arguments. I just finished reading an article at "Football Outsiders" which read so much like a confused half-truth + half-falacy myth. In my opinion, those writers don't know enough about statistics to write about it, let alone justify some of their conclusions.

I live in a profession that requires some degree of stats. I've studied it enough to know that it is an world unto itself when applied to an industry. I've dealt with some real half-cocked people in this profession, who knew enough about stats to talk the talk but couldn't apply it to real world situations, where process stability is required.

Last year, there was a short discussion about statistics and "trends" and some argued that the Rams were definitely trending upwards and that Fisher and the Rams had taken some strides forward. I think it was early in the season and we were 3/1 or something like that. Ya'll know how that trend ended.

But that discussion got me thinking, there are so many variables in the game of football, which has a horrible stable sample size, that trying to predict team Wins/Losses, based only on stats, for a season is unreliable. Change one or two or three key factors and W/L records can get turned on their heads.

I once conducted an expensive DoE for a PhD over a manufacturing process. I gathered tons of data and crunched it. When I evaluated the results I was disheartened because I knew I had failed in my work. A lot of time and money was spent on this effort, and some big bosses expected strong results from me. Even though I knew I screwed up, I had to make the presentation, show the results, apologize, and admit that I screwed something up badly. At the end of my presentation and apology for the results, she looked at me and said, "actually what you presented makes perfect sense and it is 100% accurate." I said, "no way, how can it be the data says otherwise? something is wrong with my results." She then explained a key processing factor that I was not aware of and did not take into account. Once she explained it, I said to myself, "oh crap, this makes perfect sense now." I felt damn good about myself because then I realized that my work was really a touchdown, not a fumble.

(Sorry if I am rambling, I have a little extra time on my hands).

All of that to say that I don't think you can use "Regression towards the mean" to predict a team's W/L record confidently. There are far to many changing variables and far too small sample sizes.

There are certainly a lot of potential ways for the Rams to return to a 4 or 6 win season, regression towards the mean isn't one of them.