Monday: I think that's the problem.

If function A is basically a flat plane, and function B is basically a possibly different flat plane, then attempting to find the solution B = \alpha A + \beta is effectively a degenerate system.  As a result of this, if A does have a deviation from the flat plane model, then the minimization process fitting \alpha and \beta are very likely to set \alpha to a very small/large value to effectively smooth that deviation out, using \beta to compensate for the introduced DC offset.

I'm 90% sure that this is the cause of the bad fits I'm seeing, and why I can't find the deviations I expect to exist.

Random red panda.

• Dog wants tater tot.
• Bob's Burgers is better than all the other shows on Fox Sunday.
• How Mississippi screwed themselves to ensure people didn't look at them as a success.