@SouthJamaica wrote:
To test it meaningfully for the mortgage scores would require this hypothetical person with the huge limit to have either (1) a very, very quiet profile, with absolutely nothing going on, and nothing passing any aging benchmark, for a period of months, or (2) the willingness to plunk down 2 or 3 extra $45 payments to MyFICO, over and above the cost of the monitoring service.
I think those are fair points, but I don't think they are unique to this question. They apply to anyone trying to produce test results for any conjecture about how the back-end algorithm works. People are constantly claiming that they have "data" that show this or that, and I have to remain privately skeptical, since there are lots of alternative explanations. Even in that monster thread about CLs I was the most consistently (if gently) doubtful. BBS will remember in particular how people were trying to draw inferences from more than one person's test "data", which frequently considered of badly remembered details (little documentation) of an "experiment" that strected out for many months, with a score produced by a credit card in Oct being compared with something else in February. It nearly drove me to Bellvue, since there were simple clean easily replicable ways to run the tests that I kept voicing.
So I definitely get you here, SouthJ! Just that I'd extend what you say to cover a lot of stuff.
One advantage here is that the test is easily replicable, even by the same person. He lets two report, then one, then two, then one, etc. That rules out a lot of the problem of confounding factors. If he keeps seeing the same behavior and especially if he sees the relevant negative reason code appear each time, that tells us a lot.
So all we need to test the mortgage scores is someone with a big CL and who has myFICO Ultimate. I'd like more than one such person, which reduces the chance of error further.
