FICO scores have become so unreliable

[Revelate]

---

@Peter1142 wrote:

---

@Anonymous wrote:

---

@Peter1142 wrote:
No one is a top tier auto risk and a lukewarm mortgage risk.

---

Why not? Being able to afford a $500 car payment on $30-40K loan is small potatoes compared to a Mortgage payment for $400-700K Loan. 

---

FICO scores do not evaluate what you can afford to pay. This is a whole separate and detailed analysis that is done by mortgage lenders (as well as card credit lenders, or any lender.). It is a single score that is supposed to evaluate your overall creditworthiness. Not your finances. In fact, I was applying for a relatively small loan that would reduce the mortgage payment I have not been late on once in over 5 years.

I do not have any collections, but I do have an authorized user account with lates on it, I think is ignored by current models. I thought this might be the issue, however my Transunion score which included it was much higher than the Experian score that did not.

---

Get rid of those authorized user accounts.

AU's count on the older models (FICO 04 and earlier), they may or may not count on the FICO 8 and later models... but the simple fact of the matter is if you have lates on it get rid of that AU immediately.

This thread /sigh.

[Peter1142]

---

@Anonymous wrote:

---

@Peter1142 wrote:

---

I understand you are saying someone's creditworthiness can vary wildly between auto loans and mortgage loans, and we are just going to have to agree to disagree, I don't think this makes any sense at all. Maybe a minor difference sure, but not over 100 points and several credit tiers.

---

Alright, we'll move past the analogies and go back to facts.  One major one that you failed to comment on provided by CGID that almost completely gives you a "reason" for your 100 point variance is that the ceiling of the auto model is 900 and for the mortgage models is far less, as low as 818.  That's an 82 point variance right there between the "top" score of each, so a 100 point variance like you're seeing is many times not "several credit tiers" different and often isn't any tiers different.

---

Oh come on this is just getting silly. A 695 ia verifiably two tiers below top rates. You can verify this right on this site.  https://www.myfico.com/loan-center/home-mortgage-rate-comparison/default.aspx

An 800 Auto FICO is god tier. You can get anything available. I got best rates with scores around 710 without any difficulty. 

And while an auto FICO has a higher theoretical maximum, it also has a higher theoretical bottom, and credit scores aren't linear anyway, it's a bell curve thing. Sure there may be some difference but you are probably no more likely to get a 900 Auto FICO than an 850 FICO.

[Anonymous]

Hey Peter.  When you started this thread I'm guessing you thought two things:

(1)  That the FICO mortgage scores were scores designed by FICO to assess mortgage risk

(2)  That the Auto score you were seeing and the "mortgage" score were both pretty much contemporaries, scoring models created by FICO pretty close to the same time. 

Both are natural assumptions.  If we grant them, then your initial question makes a lot of sense: why does FICO seem to believe that I am much much riskier for mortgages than I would be for a car loan.

But both assumtions turn out to be untrue.  I know you heard me explain this earlier, but I feel like I should take another shot at it. 

Regarding #1... the scoring models that are currently used by the mortgage industry were never designed specifically to assess mortgage risk.  They are just very old FICO models that were created to assess a kind of generic risk, without any special focus on mortgages vs. auto loans vs. credit cards.  The Auto flavors of FICO, in contrast were designed very specifically to assess a person's specific risk to an auto lender.  My guess is that being late an auto loan (or a re-po, etc.) is very heavily weighted in the Auto models, more so than in others. 

If the mortgage models were never designed specifically to assess mortgage risk, then why are they called mortgage scores?  That's because they are the scores that the mortgage industry uses, and they do so because Fannie Mae forces them to. 

Regarding #2... the "mortgage" models were made a very long time ago.  The EX mortgage model, for example, was released over 20 years ago and hasn't been changed since then.  FICO 8 Auto was made much more recently.  So even if we forget about point #1, that score represents a risk assessment from a very long time ago, based on what a team of statisticians and developers thought decades ago, whereas the Auto score you are likely looking at is based what a different team of people thought based on consumer data that is much more recent, and with the hindshight of seeing what was not ideal in at least two major prior models and therefore changing it.

So to recap this general point, your initial question is a good one if the mortgage model was created by FICO to assess mortgages, and if the mortgage vs. auto models were created around the same time.  But neither of those things is true.

That's the general more theoretical point.

As far as the narrow practical issue of what might be going on your report to cause the two scores to be different, you'd need produce the entire report line by line and then the folks here might be able to guess.  We wouldn't be telling you why we think you are at much greater risk to a mortgage lender than an Auto lender.  We'd simply be explaining why the two models would give different values -- models that were created many years apart by different groups of fallible people.  The folks here do know a little bit the models from a descriptive angle -- which is different from us commenting on them prescriptively.

If you have a chance to reply again, can you confirm for us that you actually don't have any practical interest in the mortgage scores themselves?  By practical I mean you are not planning to refinance your current mortgage nor do you think you might buy a home in the next few years.  I asked you that early on and you said no, but some other folks may have missed that, and are thinking that your chief interest is in improving your mortgage score.

[Peter1142]

Hey thanks for the reply, what you are saying makes sense. I guess it is more the old scoring metrics only the mortgage industry is sticking with being crappy than the new ones.

I posted before I gave up on the idea of refinance. But I will try to raise the score anyway and see where the rates go next month.

I know some of the reasons my score is lower:

1. Of course, there was a new car loan after that 800 FICO (but an old one was paid off too, unclear the effect on old FICO for the latter.)

2. I got a new card and transfered a balance to it from a card with a higher limit.

So that's 2 new accounts in the last 12 months or so, and one of them is at 88% usage (the rest are all at 0-5% though, overall low utilization.) I have plenty of accounts, AAoA is 4.5 years according to Experian.

Also:

3. I have 3 30 day lates on authorized user accounts I put in a dispute to remove today. Transunion is already done. I believe FICO 8 ignores them.

And

4. I have a bunch of cards with a 1-5% balance and the old scoring model doesn't like this.
This I can also correct and will do so.

[Anonymous]

Yep, implementing AZEO will help your mortgage scores a lot (given that you have many cards with a small balance).  Just make sure that the card that does report the positive balance is....

A true credit card (not a charge card)

A card in your name (not an AU card)

A card with a credit limit of < 34.9k

Getting rid of all derogs would be great.  The mortgage models count all AU cards but FICO 8 might be ignoring them (FICO 8 sometimes counts AUs and sometimes doesn't).

You are also being hurt by having a card with a high individual utilization.  Individual Util is a different scoring factor from  Total Util.  Is it financially doable to pay all cards to under 48.9% (individual)?  If not, what about 68%?

[Peter1142]

I can't afford to pay off the big balance, and I am getting 0% interest on a balance transfer, I am not going to transfer it off and pay interest just for a refinance I am lukewarm about, it is something I am stuck with. I understand the high util on the single card still hurts me even though overall is low.

I have a single 30 day late on a Best Buy card from about 3 years ago that I forgot I had used, it is the only derog in my name. I have previously asked them to remove it and did not have luck. The mortgage guy brought it up like I committed a crime or something in trying to explain why my scores are not good - I can't imagine a single 30 day late that is 3 years old on a file overflowing with on-time payments can have any significant effect, even on the old models?

Do you know how long I should expect Equifax to take to remove that AU account I disputed? Transunion had it gone in hours.

[Anonymous]

---

@Peter1142 wrote:

---

---

Oh come on this is just getting silly. A 695 ia verifiably two tiers below top rates.

---

And you've already provided reasons (and have been told) why your score is lower than it could be.  If you were to get rid of your "many" small balances and take your high utilization card down to an ideal place, your mortgage score(s) would jump up a good amount.  Considerably if you're talking EQ.  Just because the auto version isn't as sensitive to these factors doesn't mean that there's anything "wrong" or that FICO scores are "unreliable" as you put it in your thread title.  Again, different scoring models view/weigh CR data in different ways.

[Peter1142]

---

@Anonymous wrote:

---

@Peter1142 wrote:

---

---

Oh come on this is just getting silly. A 695 ia verifiably two tiers below top rates.

---

And you've already provided reasons (and have been told) why your score is lower than it could be.  If you were to get rid of your "many" small balances and take your high utilization card down to an ideal place, your mortgage score(s) would jump up a good amount.  Considerably if you're talking EQ.  Just because the auto version isn't as sensitive to these factors doesn't mean that there's anything "wrong" or that FICO scores are "unreliable" as you put it in your thread title.  Again, different scoring models view/weigh CR data in different ways.

---

You are jumping between arguments, to a different one now.

And I was not arguing there are no reasons that my score could be higher, I just listed a bunch of ways I can improve it myself. I was arguing that there is no good reason absent any derogatories or something like a massive utilization spike (my overall utilization went down not up), that a rating of my creditworthiness should drop entire tiers as it has. That's my opinion, and it is a valid one. I get it, you don't agree. That's fine. But I would indeed argue there is a great deal "wrong" with the scores that were pulled on my mortgage, as there is zero chance I would ever default on the loan (absent the zombie apocalypse or something), and I believe my credit file clearly shows that to anyone looking at it with an objective eye. But yes, I realize it is maybe not that the new scores are bad, the old ones are obviously not so good. If they consider a bunch of cards with 1% utilization as a reason that someone is a bad risk for a mortgage. That's "wrong", if you ask me. And hey, that's not just my opinion - it's obviously FICO's too, as they have moved away from that. I am sure would welcome selling the latest scores for mortgage loans if lenders would accept them, or even making a mortgage enhanced version.

What I meant by "unreliable" in the topic title, is that one score is an unreliable indicator of another, or what any lender will pull. It used to be just this way with FAKO's, now it is every score. They can all vary so much, and they obviously can't all be right. A person is a good credit risk or they aren't (with some minor differences for credit types.) A credit score is a measurement, it doesn't change facts.

[KJinNC]

Although it's easy to give specific reasons to refute the OP's point, I wonder the same thing.

In a nutshell, can you really be a great credit risk for a card, a great credit risk for an auto loan, and a mediocre credit risk for a house? Or, shuffle the order, it's the same question.

I understand how the numbers end up different. I just wonder if there is, in reality, a valid reason to have different numbers for different types of credit. FICO 9 vs FICO 8, I understand. They are trying to improve or perfect the formula. VantageScore vs FICO, I understand. They are trying to score people who are not scorable yet on FICO. What I am talking about is all of these very specific FICO scores for different types of loans, as if you can be a safe bet for one type but a risky bet for another.

[Anonymous]

---

@Peter1142 wrote:

---

What I meant by "unreliable" in the topic title, is that one score is an unreliable indicator of another, or what any lender will pull.

---

A score from one algorithm isn't supposed to be an indicator or at all proportional to a score from a different algorithm.  I guess that's where your gripe resides.  For me, I don't take issue with any of that.  The only score that matters is the score that a potential lender may pull if you're thinking of apping with them.  Also watch saying that there's "zero chance" that you'd ever default potentially.  I'm quite sure that the vast majority of those that have defaulted on mortgages had no intention at all of ever defaulting and at the time the applied felt there was zero chance they'd default in the future as well.