@Revelate wrote:
@SouthJamaica wrote:My impression is that recent inquiries are given less, rather than more, weight in FICO 9 than in FICO 8.
Need more data. Assumptions have a well known cliche
.
To date I haven't seen any from anyone else on this topic (not to say it doesn't exist), hopefully others will chime in with datapoints in the future.
Can't say that I have data, it's just an impression. It is not an assumption But here's how I got that impression:
Each of my FICO 9 scores is wildly higher, an average 60 points higher, than each of my FICO 8 scores.... [50 points, 63 points, and 73 points.]
All the FICO-announced FICO 9 differences are totally irrelevant to my particular profile (collections, rental history, etc).
I first heard the theory or hypothesis from Thomas_Thumb that recent inquiries and age of accounts are given less weight.
It resonated with my profile, because the first 3 'negative factors' regularly thrown up at me by FICO 8, regardless of utlization issues, are (1) my one old public record, (2) age of accounts, (3) recent inquiries. So while the public record is the biggest factor, age of credit is the second biggest.
So it appears that there are 3 possibilities:
1. FICO9 would ease up 60 points worth on a single old public record which was my biggest problem in FICO 8
2. FICO9 would ease up 60 points worth on age of credit or
3. FICO9 would ease up 60 points worth -- total -- on (a) a single old public record and (b) age of credit
Under scenario 1, my age of credit issues wouldn't be a factor, but scenario 1 seems by far the least likely of the 3, as I doubt that FICO 9 has suddenly determined that a single old major public record -- my biggest problem in FICO 8 -- is no problem.
So from all of this I get the impression that my second biggest problem in FICO 8 is significantly less of a problem in FICO 9.
@SJ: Yeah don't think there was any data to assert the prior inquiry theory. That happens when we're speculating on this forum with limited information. Like you my history is sorta short (3 years AAOA, 8 year oldest account) and I have an old tax lien and some old lates on EQ/EX. My EQ score matching yours roughly is consistent, what isn't consistent is my EX score does not so that's where this analysis comes in.
Have another not quite datapoint but an observation from the 4/28 pull just looking at the Classic scores:
EQ FICO 9 = 787
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
- You've recently been looking for credit
EX FICO 9 = 735
- You have a public record and/or collection on your credit report.
- You've recently been looking for credit.
- You have a short credit history.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
That actually is a pretty clear indication that inquiry distribution is a non-trivial factor if it's my (for serious with my file??) #2 factor on EX. It is still in the #2 slot on the 5/1 pull too, so the reason codes didn't change even if the score did.
Also and this admittedly isn't fair but it is curious: on the EX FICO 8 models the order is:
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- You've recently been looking for credit.
Certainly suggests that FICO 8 penalizes short credit history more and inquiries less in the relative weights than FICO 9 does.
And actually just throwing this in here because it's a fun comparison:
4/28 TU FICO 9: 736, with a 30 day late from 10/2015 and similar inquiry distribution to EQ
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- You recently missed a payment or had a derogatory indicator reported on your credit report.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
The fact the score between TU / EX is basically equivalent suggests that #2 on EX, recently seeking credit, is a doozy, though why short credit history which doesn't appear to be a deal killer in FICO 9 is ranked above that 30 day late is another kind of headscratcher.
I can't offer any personal data on inquiries. Also, unlike you and Revelate, my Fico 09 Bankcard enhanced scores are less than my Fico 08 Bankcard enhanced scores. I'm sure my scorecard behaves differently so what I see is likely not what you will observe.
For what it's worth I have two full datasets comparing F09/F08/F05 and VS3. They are summarized in the below table. Revelate & SJ you may find it helpful to report data using a TBD standardized format table that includes all revelant information
Note: I have one open installment loan (mortgage) hving a balance to loan ratio between 30% and 40%.
Observations:
1) Reducing # cards reporting (plus AMEX reporting) had a significant positive impact on all Fico 04 scores.
2) All EQ Fico 04 models reacted much more strongly than corresponding TU or EX Fico 04 models.
3) VS 3.0 scores went up across the board with the increase in UT% and # cards reporting. The simulator on CK accurately predicted an increase and simulator testing suggests 4% AG UT is optimal for scoring.
4) Bankcard scores are consistently lower with Foco 09 relative to Fico 08. Not sure why but, I hypothesized more weight given to # cards reporting.
Thanks TT for your data!
It's an interesting bit but to confuse the issue: we know from android01's data that you can have "blemishes" and still be an 850, it just means there's some leeway (better classes of perfect? hard to say) but I don't think that's scorecard dependent: have have tons of data that FICO 8 does react similarly with number of cards with balances with people who aren't at 850. Also we do have a marked difference, you're between 50-100% cards report balances and my data was between <10% and 50%, and the algorithms might well behave differently there. Whenever I get back to a steady state I'll see about getting all the way up to 13/13 and see how things go, would need to adjust some payment dates prbably to really do that.
Anyway can table that discussion for now or break it out later, not really germane to the discussion at hand, again mea culpa for introducing it into the topic.
Yeah, my theory is there is substantial inquiry aging over either a 3 or 6 month period based on my data unlike FICO 8 or earlier. I should be able to get another set of concrete datapoints shortly.
Age of derogatories perhaps not being counted may well be a factor as that would explain both SJ and my scores; that I can't really do much about now, but if I get the obvious and known shift on FICO 8 when my tax lien comes off, and I don't see movement on FICO 9, that would almost prove the case... or in my instance, paid tax liens may not count but that wasn't announced by FICO (nor VS) so I'm hesitant to suggest that.
I use a mix of 3B and 1B reports: the reason I am not including TU is because that file is absolutely different - there is a 30 day late from 10/2015 on there and as such it's not a good comparison for this. While there is some other data around lates and what not which I can tease out from that, I can't really compare this - I expect without the late I'd be nearly the same 780+, one less tradeline, one less month AAOA, and doesn't appear to be the consumer finance account tag on that account (my deep subprime loan isn't on that bureau) on any model but tough to say for certain when I have so many negatives on my reports anyway it could just be pushed out of the top 4.
One more thing to look at with Fico 09 relative to Fico 08.... # accounts
As mentioned, I don't score quite as high on Fico 09 enhanced relative to Fico 08. Also, I don't have a lot of accounts (9 or 10 total with 7 open). Perhaps, Fico 09 is looking for a count threshold. Also, something like the hypothetical below binning example might be in-play with new INQ aging.
@Thomas_Thumb wrote:One more thing to look at with Fico 09 relative to Fico 08.... # accounts
As mentioned, I don't score quite as high on Fico 09 enhanced relative to Fico 08. Also, I don't have a lot of accounts (9 or 10 total with 7 open). Perhaps, like VS 3.0, Fico 09 is looking for a count threshold.
That's entirely possible, though I'm trying to not make other assertions regarding FICO 9 (after making that mistake mixing FICO 04 in there) here. Outside of another perfect 850 (and I don't know credit wise if an 850 is an 850, clearly you can be above that threshold on FICO 8/9 in terms of not maxing out every portion of the scorecard, see your installment utilization ratio for FICO 8 for example) with more accounts it's hard to say. Enhanced scores are laughably difficult to try to analyze though I suppose not any harder than FICO 9 or TU/EX 04/98 other than the different weightings which we have at best SWAG's on. Generally my enhanced scores have been slightly higher than my non-enhanced which given their maximums makes some sense given I have a very "well rounded" file.
One thing I did notice with your data: the absolute difference between FICO 8/9 and FICO 4 isn't entirely meaningful - it wasn't possible at all to hit a 850 under many of the older models, EQ Beacon 5.0 for example maxxed at an 818 so you're still awfully close to max score there too rather than the larger absolute magnitude of the difference.
@Revelate wrote:@Anonymous: Yeah don't think there was any data to assert the prior inquiry theory. That happens when we're speculating on this forum with limited information. Like you my history is sorta short (3 years AAOA, 8 year oldest account) and I have an old tax lien and some old lates on EQ/EX. My EQ score matching yours roughly is consistent, what isn't consistent is my EX score does not so that's where this analysis comes in.
Have another not quite datapoint but an observation from the 4/28 pull just looking at the Classic scores:
EQ FICO 9 = 787
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
- You've recently been looking for credit
EX FICO 9 = 735
- You have a public record and/or collection on your credit report.
- You've recently been looking for credit.
- You have a short credit history.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
That actually is a pretty clear indication that inquiry distribution is a non-trivial factor if it's my (for serious with my file??) #2 factor on EX. It is still in the #2 slot on the 5/1 pull too, so the reason codes didn't change even if the score did.
Also and this admittedly isn't fair but it is curious: on the EX FICO 8 models the order is:
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- You've recently been looking for credit.
Certainly suggests that FICO 8 penalizes short credit history more and inquiries less in the relative weights than FICO 9 does.
And actually just throwing this in here because it's a fun comparison:
4/28 TU FICO 9: 736, with a 30 day late from 10/2015 and similar inquiry distribution to EQ
- You have a public record and/or collection on your credit report.
- You have a short credit history.
- You recently missed a payment or had a derogatory indicator reported on your credit report.
- The remaining balance on your mortgage or non-mortgage installment loans is too high.
The fact the score between TU / EX is basically equivalent suggests that #2 on EX, recently seeking credit, is a doozy, though why short credit history which doesn't appear to be a deal killer in FICO 9 is ranked above that 30 day late is another kind of headscratcher.
My FICO9 negative factors are identical, and in identical order, for all 3 bureaus:
- You have a public record and/or collection on your credit report.
- You've recently been looking for credit.
- You have a short credit history.
My FICO8 negative factors are likewise identical to the above except
that for some reason TU doesn't mention "recently looking for credit"
and EX and EQ have "recently looking" and "short credit history"
inverted. And all 3 FICO8 scores are close to each other.
Bottom line is that FICO9 and FICO8 agree that "inquiry impact"
and "aging" are my second biggest problem, but FICO9 forgives me more
