AAoA thresholds? UPDATE 5/2/19

[Anonymous]

---

@SouthJamaica wrote:

---

I wonder what testing you did that makes you so confident that the results were attributable, or not attributable, to this single factor.

---

A pretty standard test:  Went from AZ (then crossing AZEO) to all accounts with balances and then back to AZ again, pulling my 28 FICO scores at the several significant points in time and comparing them.  All reported balances were small, so utilization-related score changes is ruled out.

TT theorized at the time that I posted my results that perhaps because I hadn't had an inquiry and/or new account recently (both > 12 months) that perhaps that's why I saw the results I did.  Maybe the majority of people that are looking at their mortgage scores relative to their FICO 8 scores have had inquiries in the last 12 months and/or have opened accounts, which may impact the signal strength on number of accounts with balances.  Who knows.

[Anonymous]

---

@Revelate wrote:

Also Birdman’s data has shown on EQ FICO 5 not only accounts with balances matter but also revolvers with balances matter... double whammy.

Every file is different but I agree with the assertion I believe TT made first, in general mortgage scores are more strongly affected by AZEO than FICO 8, and I am on 3 entirely different scorecards between the 3 bureaus.

---

Agreed that EQ is the most sensitive.  That's the only mortgage score that I saw take a big hit when moving to all acounts with balances from AZEO.

That being said though, one mortgag score taking a drop when the middle score is used is sort of a non-factor for me.  My middle mortgage score moves 5 points in going from AZEO to all accounts with balances where my average or "middle" FICO 8 drop is 11 points.  As you said, every profile is different.

[Anonymous]

BBS, which do you posit was impacted by the lack of inquiries/new accounts? Are you saying TU FICO 08 was impacted and therefore more reactive to # of accounts with balance OR that TU FICO 04 was impacted and was therefore less reactive to # of accounts with balance? Or both?

[Anonymous]

Not sure the circumstances or factors surrounding it.  Just to provide the numbers, on TU going from 2 of 8 accounts (1 revolver, 1 loan) to 8 of 8 accounts with a balance saw my FICO 8 go from 850 --> 835 where my FICO 4 went from 819 --> 811.  So, 15 points verses 8 points.  Also, with potential top end buffer on that 850, it's possible I'm looking at a few more than 15 points, but that's impossible to tell really.  0 scoreable inquiries on TU, actually 0 anywhere on my TU file (none within 24 months) and AoYA was 21-22 months or so at the time of the test.

[Anonymous]

Interesting, indeed.

[Revelate]

Actually and I hate to say this: but quite likely 2/8 isn’t optimal on TU 04 to begin with, which may be why yours didn’t move much whereas mine moved over 20 points and I didn’t even get to all accounts with balance.

As an aside in both my and Birdman’s data accounts with balances is just a lame implementation... I know TU 4 can be optimized at least because my reason code is totally gone in my derog scorecard there but EQ FICO 5 is total SMH on that. Also SJ has a large number of accounts too.

Actually do you have reason codes for those datapoints for TU FICO 4 BBS?

[Anonymous]

I think 2 of 8 is optimal on TU 4 if you're talking a loan and a revolver, but like many other scoring models it's going to generate the too many accounts with a balance reason code regardless.  The only option there is to go to AZ on revolvers to get that ratio down to 1 of 8, but I think the general concensus is that the penalty associated with no revolving credit use exceeds the penalty associated with too many accounts with a balance in that case.

With my TU 4 score I'm only provided ever with 2 negative reason statements.  They are both the same, regardless of if I'm at 2 of 8 or 8 of 8 accounts with a balance reported:  length of time accounts have been established and length of time revolving accounts have been established.  My guess is that number of accounts with a balance would be at 3, but who knows if it's only impacting my score 8 points you know? 

[Anonymous]

Nevermind, Received an answer

[Revelate]

---

@Anonymous wrote:

I think 2 of 8 is optimal on TU 4 if you're talking a loan and a revolver, but like many other scoring models it's going to generate the too many accounts with a balance reason code regardless.  The only option there is to go to AZ on revolvers to get that ratio down to 1 of 8, but I think the general concensus is that the penalty associated with no revolving credit use exceeds the penalty associated with too many accounts with a balance in that case.

With my TU 4 score I'm only provided ever with 2 negative reason statements.  They are both the same, regardless of if I'm at 2 of 8 or 8 of 8 accounts with a balance reported:  length of time accounts have been established and length of time revolving accounts have been established.  My guess is that number of accounts with a balance would be at 3, but who knows if it's only impacting my score 8 points you know? 

---

Not what I meant, my point is that if you were at 2/8 there's a good chance you would have that reason code on TU FICO 4, you weren't at optimal for that algorithm, you already had a penalty baked in.

Like I said before accounts with balances is pretty absurd, apparently you need a much thicker than average file to optimize it.

Unfortunately this is where having access to #3 and #4 matter.  I know for example both loan utilization and accounts with balances can be completely eliminated from the reason codes which means there's optimal thresholds, but given they sit usually in the 3/4 spots even on my file, bleck.

[Anonymous]

2/8 is definitely NOT optimal on TU FICO 04. Might be optimal for YOUR profile but that's only because you don't have enough revolvers on your profile to fully optimize it. I mean the negative reason code is explicit, BBS.

No, it's not gonna generate it regardless. It's generating it for a VERY good reason. You are not able to fully optimize for TU FICO 04 with 8 accounts. Simple as that. Mathematically impossible. You have to chose between your penalties because you don't have enough accounts to optimize. No revolving penalty or too many accounts penalty. That's all you can do with 8 accounts. 1/8=12.5%; 2/8=25%. The line falls in between there. Yeah, you can bet #3 is number of accounts with a balance.