Positive score buckets a matrix?

[vanillabean]

Could it be that positive score buckets are not simply tiers of either oldest TL or AAoA, but rather a matrix of both? If so, what might the top combo be?

[Anonymous]

---

@Anonymous-own-fico wrote:

Could it be that positive score buckets are not simply tiers of either oldest TL or AAoA, but rather a matrix of both? If so, what might the top combo be?

---

It's fun to guess - but the coolest part of speculating is learning whether your assumption was accurate - or not.

Since FICO's not sharing the bucketing info - it's simply an exercise.  It"s entertaining - just important that folks don't start taking guesstimates and conjecture as FICO fact. 

[haulingthescoreup]

---

@Anonymous-own-fico wrote:

Could it be that positive score buckets are not simply tiers of either oldest TL or AAoA, but rather a matrix of both? If so, what might the top combo be?

---

You may be saying the same thing I've been guessing about. * note to self: "matrix" sounds cool; consider using in the future *

I think there is some level of expected consistency between AAoA and oldest account. For instance, with 20-something years but 4 years AAoA, I got snarky remarks about short history, which went away (or at least dropped lower down on the list) when I hit 5. But plenty of other people with 7 years longest, or 10 years longest, don't get "short history" with their 4 years AAoA. I'm guessing it's because that's a reasonable combo --oldest account not that much older than AAoA.

As for the "top combo", I lean toward the High Achievers profile of oldest account of 19 and AAoA between 6 and 12 years. I'm pretty skeptical about other things on the list, such as the extremely limited number of accounts, both open and closed, as these stats were gathered before the explosion of rewards cards. But the age part seems pretty believable. (Now I wait for all the replies from those with 19/ 6-12 showing "short history. )

Of course, if this wasn't what you were talking about at all, never mind. I had to learn matrix math in 10th grade pre-calculus, and it permanently cured me of higher-level mathematics.

[RobertEG]

No one really knows the specific system of grouping credit files into categories, or “buckets,” prior to determining the specific scoring algorithm it will use in the actual scoring.   However, there is a published document by First USA Bank that describes their own in-house categorization of credit files.  It is U.S. Patent No. 6,202,053

In their methodology, it is a branching tree.  The first cut is made based on whether the credit file is “thick” or “thin.”  In general, a thin file is, under their system,  less than three trade lines, or no credit card.

A second major branch is based on delinquencies.  Files are classified as “clean” or “dirty”, based on the absence or presence of a major delinquency in the file, which is defined as a late of 60+, or a CO, CA, or PR.

A third major branch is %util on revolving accounts, with various sub-branches thereunder, depending upon the %util.

A fourth major branch is length of credit history.  They use less than 10 years, or 10+ years as a primary breakdown.

There are sub-branches off of these major branches, which in all, match approximately the 12 major buckets that are reportedly used by FICO.  So it at least gives some objective informtion as to how this "bucketing" game works.

This description is, of course, not published by Fair Isaac, and is not the FICO methodology, but it is one of the few published descriptions of how a major credit scoring system uses categories, or “buckets.”

Take a quick read of U.S. Patent No. 6,202,053.  It shows their branching tree evaluation of scoring categories, or "buckets."  At the very least, it arises above mere speculation on how "bucketing" is done by major players in the game.