Those of you who read EaglesRewind know I occasionally feature guest posts. Following is one such post, that I felt deserves a broader audience, so I'm reposting it here. I'll provide a comment to the model at the very end.
Brent's Comment: Overall, I think this is an interesting exercise (obviously, or I wouldn't have run it). My biggest word of caution would be to not get too hung up on the percentages. Unfortunately, whenever you build a model like this, the results APPER to represent some precision. Due to all of the external variables we can't account for, that's not the case here. Instead, I think it's more useful to just view it as a discrete ranking of which coaches are most likely to get fired and how their relative ranking compares to others.
We could, of course, design it that way, and only show you the ranking, but that wouldn't be as fun. So we've chosen just to trust that you'll read this comment and use the data responsibly. Besides, transparency is important, if for no other reason than as another level of editing (the original version had a slight flaw, pointed out by a commenter, which is now fixed)
In any case, I think there are some additional steps we could take from here, and build things like Tenure, injuries, or maybe even Market Size into the model.
The value lies in the pre-season projections. We could potentially be able to project just how many games each coach has to win in order to keep his job. That would certainly make some teams more interesting to follow.