You keep saying that, but like I said, there are entire industries that make boatloads of money doing exactly what you say is impossible. (I would know, I work in one.)
Again, do any kind of analysis to show it was a bad decision. We're still waiting for that, since you and others are the ones making the claim that it was such a bad decision. Prove it (or don't and just keep arguing for no reason).
I work heavily with data at a large tech company and externally deal with Senior Execs at F100 companies who do nothing but look at data.
A lot of people who collect data don't fully understand how to analyze it and a lot of those who analyze it don't know how to collect it. There are numerous ways to collect and analyze data depending on what you're end goal is. If you work with data, I'm sure I don't need to go down this rabbit hole. My point is, it can be misleading or misunderstood very easily, and often is.
When looking to respond to the questions you have posed to Bankerguy, I went on a quick hunt for data. Interestingly enough, ESPN published an article on this just 4 days ago, it was the first result I found.
Do NFL analytics models always recommend going for it on fourth down? No. Here's how the decisions actually are made.
www.espn.com
That's the article. You can actually dive deeper on it into this particular game here:
https://espnanalytics.com/decision
A few assumptions have to be made here, and I won't just grant these assumptions bc
it's ESPN so it has to be right.
Anyways, if you were to simply analyze the data as a whole in the graph from the article, it says
GO on any 4th & 4 or shorter from your 30 and out. That graph or something likewise is prob used by NFL coaches and their analytics guys across the board. And to my earlier point, I guarantee there are some using the data improperly. You need very specific data for specific situations (and it can be compiled, it's just challenging) to always make the right call based on data. There are plenty of situations where going for it on 4th and 4 on your 31 yard line would be wildly incorrect, even if the data says
GO, according to ESPN.
And FTR, I can manipulate the data sets I'm collecting (based on collection methods and chosen inputs) to justify most decisions I make.
How many teams did X when Y was Z?
Okay, that doesn't get me my desired outcome.
What about How many teams did Y when Z was X,
okay this data looks good. Just a simple example.