Maximizing the final 25% of the pie to 850?

[Thomas_Thumb]

---

@iv wrote:

---

@Anonymous wrote:

IV, outside of the couple of new accounts you've opened since 2015, what else do you chalk your 780's mid mortgage score to when comparing it to your 850 FICO 08's?

---

Beyond the inquiries and new accounts, there's the 95% B/L on the installment loans (2x auto, 1x mortgage), and that I'm not bothering (at the moment) to optimize utilization/reporting for cards at all. (Paying in full before due date, but NOT before statement cut/reporting.)

As such, depending on the time of month, all or most cards are reporting balances at the same time.  Total utilization is usually staying only around 1% (mostly due to high limits...), but % of cards reporting balances is high.

---

It is the number of cards reporting balances that dropped the Fico mortgage scores. I have tested that many times. My Fico 8 scores held at 850 for all the below data points. I suspect IV's lowest mortgage score is from Equifax. It would be interesting to know how all three mortgage scores compare.

Here is another example of data that contradicts the generalization: "must have an open installment loan with total B/L below 9% for best Fico 8 score". It is just not true - as a rule- even though B/L ratio strongly affects scoring of share secure loans and "new" auto loans. A low B/L requirement is NOT necessary for other loan types, as shown yet again, by IV's data. IV has a 95% B/L ratio on installment loans. My-Own-Fico reported 850 with a B/L above 90% as well. Inverse reported 850 scores with aggregate B/L above 65% (2 open mortgages and 2 car loans). I experienced similar results with 850 scores and mortgage B/L above 50%. Interestingly, a 95% threshold for a mortgage was cited by one of our long term posters (JDK???) last year.

As mentioned in other threads, there are strong indications that scoring of installment B/L may conditioned based on what loans are in the mix (particularly a mortgage) AND length of time active for the oldest open installment loan - payment history. Once oldest open installment loan age passes one or more critical thresholds - generally thought to be two years but possibly one - the impact of "high" B/L becomes muted. A portion of points associated with B/L are reallocated to payment history.

[Anonymous]

---

@iv wrote:

---

---

Beyond the inquiries and new accounts, there's the 95% B/L on the installment loans (2x auto, 1x mortgage), and that I'm not bothering (at the moment) to optimize utilization/reporting for cards at all. (Paying in full before due date, but NOT before statement cut/reporting.)

As such, depending on the time of month, all or most cards are reporting balances at the same time.  Total utilization is usually staying only around 1% (mostly due to high limits...), but % of cards reporting balances is high.

---

I would have thought the installment loan utilization was the biggest factor in this when considering the differences between the two models, but TT references the number of cards reporting balances which is very good to know!

[iv]

---

@Thomas_Thumb wrote:
I suspect IV's lowest mortgage score is from Equifax. It would be interesting to know how all three mortgage scores compare.

---

Indeed, I see the same spread as in your graphs - EQ lowest, TU middle, EX highest.

They all bounce around a bit based on reporting dates/balances, but I haven't recently seen more than a 27-point delta between EQ at the low end and EX at the high end.

---

@Thomas_Thumb wrote:

Here is another example of data that contradicts the generalization: "must have an open installment loan with total B/L below 9% for best Fico 8 score". It is just not true - as a rule- even though B/L ratio strongly affects scoring of share secure loans and "new" auto loans. A low B/L requirement is NOT necessary for other loan types, as shown yet again, by IV's data. IV has a 95% B/L ratio on installment loans. My-Own-Fico reported 850 with a B/L above 90% as well. Inverse reported 850 scores with aggregate B/L above 65% (2 open mortgages and 2 car loans). I experienced similar results with 850 scores and mortgage B/L above 50%. Interestingly, a 95% threshold for a mortgage was cited by one of our long term posters (JDK???) last year. 

---

Amusingly, by sheer chance the mortgage happens to be at exactly 95.03% right now, and all three loans combined are dead-on at 94.99%. Hasn't seemed to make a noticeable difference on the mortgage scores over time. (Although I did get an alert today that TU finally reported this month's mortgage payment, and TU FICO8 jumped to 842...)

---

@Anonymous wrote:

I would have thought the installment loan utilization was the biggest factor in this when considering the differences between the two models, but TT references the number of cards reporting balances which is very good to know!

---

Yup, number (or percentage) of cards with non-zero balances is most likely the main driver here.

I should have also mentioned - some of those cards have limits between $35k and $50k, and are likely being counted on FICO8 but not the mortgage scores.  That effect can go either way, of course - balances on all the sub-35k cards but not the others will hold down the older models, while balances only on the 35k+ cards may/will be ignored...  Optimal usage (for mortgage scores) is likely all but one sub-35k card reporting zero, and balances on the 35k+ cards are partially or completely irrelevant.

Maybe I'll play around with this next month, optimizing the card balance reporting... but it's just a theoretical curiosity at this point, no real practical concern.

[Thomas_Thumb]

---

@iv wrote:

---

@Thomas_Thumb wrote:
I suspect IV's lowest mortgage score is from Equifax. It would be interesting to know how all three mortgage scores compare.

---

Indeed, I see the same spread as in your graphs - EQ lowest, TU middle, EX highest.

They all bounce around a bit based on reporting dates/balances, but I haven't recently seen more than a 27-point delta between EQ at the low end and EX at the high end.

---

@Thomas_Thumb wrote:

Here is another example of data that contradicts the generalization: "must have an open installment loan with total B/L below 9% for best Fico 8 score". It is just not true - as a rule- even though B/L ratio strongly affects scoring of share secure loans and "new" auto loans. A low B/L requirement is NOT necessary for other loan types, as shown yet again, by IV's data. IV has a 95% B/L ratio on installment loans. My-Own-Fico reported 850 with a B/L above 90% as well. Inverse reported 850 scores with aggregate B/L above 65% (2 open mortgages and 2 car loans). I experienced similar results with 850 scores and mortgage B/L above 50%. Interestingly, a 95% threshold for a mortgage was cited by one of our long term posters (JDK???) last year. 

---

Amusingly, by sheer chance the mortgage happens to be at exactly 95.03% right now, and all three loans combined are dead-on at 94.99%. Hasn't seemed to make a noticeable difference on the mortgage scores over time. (Although I did get an alert today that TU finally reported this month's mortgage payment, and TU FICO8 jumped to 842...)

---

@Anonymous wrote:

I would have thought the installment loan utilization was the biggest factor in this when considering the differences between the two models, but TT references the number of cards reporting balances which is very good to know!

---

Yup, number (or percentage) of cards with non-zero balances is most likely the main driver here.

I should have also mentioned - some of those cards have limits between $35k and $50k, and are likely being counted on FICO8 but not the mortgage scores.  That effect can go either way, of course - balances on all the sub-35k cards but not the others will hold down the older models, while balances only on the 35k+ cards may/will be ignored...  Optimal usage (for mortgage scores) is likely all but one sub-35k card reporting zero, and balances on the 35k+ cards are partially or completely irrelevant.

Maybe I'll play around with this next month, optimizing the card balance reporting... but it's just a theoretical curiosity at this point, no real practical concern.

---

Frankly, it would be much more enlightening to get a bona fide "all cards reporting some balance" data point from another 850 profile. aWe already have countless data points for one card only reporting. It would be very interesting to see a side by side data for 3B Fico 8 scores and 3B Fico mortgage scores.

Just a guess here but, I would anticipate your EQ Fico 04 (score 5) to drop to the 760 to 765 range with all open accounts reporting balances.Optimization data points for high scoring profiles are rather boring. It's capturing the significant score drops relating to non derog activity that is noteworthy - IMO.

[iv]

---

@Thomas_Thumb wrote:

Frankly, it would be much more enlightening to get a bona fide "all cards reporting some balance" data point from another 850 profile. aWe already have countless data points for one card only reporting. It would be very interesting to see a side by side data for 3B Fico 8 scores and 3B Fico mortgage scores.

Just a guess here but, I would anticipate your EQ Fico 04 (score 5) to drop to the 760 to 765 range with all open accounts reporting balances.Optimization data points for high scoring profiles are rather boring. It's capturing the significant score drops relating to non derog activity that is noteworthy - IMO.

---

Perhaps AFTER the next mortgage, I'll play with that.

Even with a very solid file, actively TRYING to push mortgage scores down just before a loan (refi in this case)? Unwise...