33 of 56
3 years ago
Wed Feb 02, 2022 12:06 am
Justaguy
Seasoned Rebel
Justaguy has been gardening for over 2 years.
Birdman wrote: ↑Tue Feb 01, 2022 8:43 pmYou lost me. I will try to work through the earlier posts and respond where and when I can. But I don't follow this one. Sorry. If your theory depends on a strict application of the 35%, 30%, 15%, 10%, 10% category weights, I think that's a weakness. I have always understood those to be approximations, aids to consumers in understanding the concept of unequal weighting. Since the people who designed the FICO algorithms were trying to predict borrowers' risk, it seems to me very likely that they would depart from strict adherence to those percentages if (really, when) it suited their quest for predictive accuracy. Can you point to something that says they always stick to those precise percentages? Or am I misunderstanding how important that is to your theory?BrutalBodyShots wrote: ↑Tue Feb 01, 2022 7:53 pm Cool. So if the percentages vary by scorecard, the points then vary by scorecard. This would mean then that the point at which buffer / surplus points occur would possibly vary by scorecard. All interesting stuff to think about.Correct except for the possibly, imho, it definitely varies by scorecard. That’s not to say more than one scorecard could not have the same threshold on a particular category, but we definitely know the percentages vary by scorecard. Fico documentation establishes this. How else could it possibly be implemented other than to put a cap on each category’s contribution of awarded points? And if that occurs, the remaining surplus points are buffered/withheld, imho. Additionally, signal strength of scoring metrics varies by scorecard, but varying signal strength of particular scoring Metrics does not change the category’s top end award that we know of, therefore category buffers to maintain the assigned ratio between category contributions are necessary. The only possible variation I can think of would be if the difference between the scorecards was strictly the change in metrics’ signal strength, which then causes the differences in percentages for categories, if there are no category buffers. So either individual category buffers or the change in signal strength causes the difference in percentages, I believe. If it’s the former there are category buffers; if it’s the latter, there could be one overall buffer. My money is on the former. Can anybody shoot any holes in any of this?
- Score data EQ8: 806 TU8: 841 EX8: 810
- Classic 8 Scorecard CLEAN/THICK/MATURE/NEW-REVOLVER
- Mortgage Scorecard CLEAN/THICK/MATURE/NEW-ACCOUNT
- AoOA 28y+
- AoORA 28y+
- Date of Last Inquiry and/or New Account Opening November 16th, 2021