@JLK93 wrote:
@Revelate wrote:Stupid inquiry math time: the expected way first.
Pre mortgage: one inquiry from 9/10/14
Mortgage fun: 2 inquiries (reasonable assumption of they're all being done right based on other bureaus and TU's score too)
JCB: 3 inquiries, 8/4/15
RBFCU: 4 inquiries (maybe, this one is coded differently according to CK)
FIA CS: 5 inquiries
C1: 6 inquiries
9/4/15: -4 points, reason code states too many inquries w/in 12 months
9/10/15: +4 points, back down to 5 inquiries, reason code gone
Sure looks like grace period is legit on that JCB inquiry, but I should've taken a drop earlier on the mortgage one if we didn't have binning and I didn't from the looks of it though I'll check my score shifts later tonight after dinner to be sure but I looked for one earlier when summing up the data and didn't see one matched to that.
Looks like either binning or dedupe on the RBFCU/FIA/C1 inquiries, to 6, back to 5 expected number and the score changed.
If dedupe:
Pre-mortgage: 1
Mortgage fun: 2
JCB/FBCFU/FIA/C1: 3
9/10/15: 2
**bleep** it I need to take some dumb inquiries on TU to test this I guess, if we can nail down the breakpoint of TU inquiry bins assuming they exist by doing say an inquiry every 31+ days over a few months from a known state think we can knock this out based on my dataset. Anyone see problems with that test plan?
Why not try duplicating my test? I took 5 TU 08 HPs over 4 days. 5 days might be better. No FICO score change until 31st day.
Or duplicate my EQ data. I took 4 EQ 04 and NextGen HPs over a week. No FICO score change for 31 days.
I'm still extremely interested to see how Experian treats BofA HPs.
EDIT: The test would be more conclusive if you could get your reports free of major delinquencies. It seems like you should be able to get rid of your liens. However, I know nothing about dirty files.
The problem is there's no baseline to your test and actually the data I posted looks supiciously like yours as far as testing based on your theories, and the inquiries weren't spaced far enough apart to be counted discretely potentially. As a result I don't think it'd tell us anything new... TU's never going to be utterly clean for too long in my case, and the important thing is to test on a fixed file and that mine is at least once this most recent CC gets tacked onto my report. Dirty file to dirty file should be enough on this one, I really REALLY doubt the grace period / dedupe if it exists is different on scorecard, binning might be.
The rationale for my test plan is this: if we get inquiries that will count one at a time, let the grace period of 30 days pass assuming it's legit (as mentioned I think my data supports your assertion on that count) so see if it counts or not, and then take the next at call it day 32, wait another 32 days, etc.
In this manner using something dumb like random CLI requests of $100 from Barclays which I don't care what happens to that tradeline at this point, we should be able to see whether or not binning occurs and if so, what the breakpoints are. Then compared to my own data (and maybe yours, I need to look more closely at it) we can determine if inquiries are individually counted (in which case pretty obvious dedupe exists on CC apps for TU, again I don't think this holds on EQ/EX on my data) or if they're binned, what the breakpoints are and then we can compare those breakpoints to our previously clustered apps.
That's the logic behind my test plan anyway.
@JLK93 wrote:
@Revelate wrote:
JCB: 3 inquiries, 8/4/15
RBFCU: 4 inquiries (maybe, this one is coded differently according to CK)
FIA CS: 5 inquiries
C1: 6 inquiries
9/4/15: -4 points, reason code states too many inquries w/in 12 months
9/10/15: +4 points, back down to 5 inquiries, reason code gone
Maybe I'm missing something. It looks as though you've already confirmed my data. You've already proved buffering and deduplication of credit card inquiries on Transunion. My file is clean and yours is dirty. Yet, we have the same result.
Actually we haven't; the grace period I agree looks solid, but the dedupe could just be binning on both of our tests. As a result I want to design a test strictly for binning so we can rule that out one way or the other.
@JLK93 wrote:
@Revelate wrote:the dedupe could just be binning on both of our tests.If binning is responsible for the results of my tests, then that would mean that EQ bins HPs 1 through 4 , and scores HP 5 discretely,
and TU scores HPs 1 and 2 discretely and bins HPs 3 through 7 . Why would EQ score HP 5 discretely and TU score HPs 1 and 2 discretely and bin HPs 3 through 7. I don't see any consistency. It seems to defy logic.
I would also point out that I have an 850 on TU with 7 HPs. In the past I had an 850 HP on EQ with 5 HPs. No one else on myFICO or any other credit forum has ever achieved 850's with so many HPs. To the best of my knowledge, no one else has ever achieved a 850 with more that 1 or 2 HPs. Binning could not possibly explain the uniqueness of my 850 score with 7 HPs.
For one, there's a buffer north of 850, the best analogy I have would be my 93% is just as good as Susie Superstar's 105% in GPA terms at any university. It's been conclusively demonstrated that with many different things which drop those not at 850 by some amount, that the 850's still stay at 850. It's arguably the worst place to test from as a result.
My profile is capped positively especially on something like EQ Beacon 5 at an upper bound, a bunch of inquiries should still hit that part of the scorecard and I think it's conclusively demonstrated above that I can affect my reason codes if I get enough inquiries on there for TU FICO 8 since it showed up in my data.
The old testing moderators like HaulingTheScoreUp and illecs both described binning and I've seen nothing to discredit it.
The reason people were so up in arms previously on this thread was it boiled down to he said / she said arguments; we now have another dataset which agrees with your assertions but does not rule out the obvious forum wisdom. When multiple reasons are available, further testing is required; hence further testing. Just because you don't believe binning exists, and I happen to trust some of the old time members' who I had a bunch of interaction with, is similar: this isn't about being right or wrong, it's about proving or disproving theories... and for that we have to test. Same as any other "scientific" endeavor.
We're already seeing a marked disparity in my data between EX/EQ and TU when it comes to the grace period... namely I don't think it exists on either of the other bureaus but it might well on Transunion based on both of our datasets, this experiment simply needs to be controlled against TU alone and see if binning exists on TU with regards to FICO 8... it doesn't apply to any other bureau, though I will state it's well known across all 3 bureaus that if you rack up enough inquiries you stop losing points for additional ones, and that's certainly not dedupe. Anyway, while I'm willing to admit that the old theories may well be wrong or maybe at least not applicable to the modern FICO algorithms (almost had to be tested against EQ Beacon 5.0 from Scorewatch back in the day), I'm not about to discard it without concrete proof and that simply hasn't been provided yet in my opinion.
@JLK93 wrote:
@Revelate wrote:For one, there's a buffer north of 850,
When you make comments like this, it seems obvious that you did no look at my data. Only in the last test did I have an 850 score. The results of that test were based on TU04 and TU08 Bankcard scores. Did you actually look at the data and come to the conclusion that an 850 FICO buffer was at play?
Ok please please stop being so confrontational; you brought up the 850 example not me, and as for the EQ comment I'm going off my own data.
I don't see how anyone thinks that a single datapoint or even data set means they're right and everyone else is wrong; that goes for everyone, and that's insane. Life doesn't work that way, and this forum doesn't either as been proven over and over again. That goes for me too with my waspish snap in this thread at some folks who have decried my mortgage loan being the same as my share secured loan data, so I'm no saint.
You can complain about us all you want, but you seriously go back and forth between thanking people and picking fights, and that's just not the way to do this in any sort of community.
If I've misread this I apologize but I've around the net for a while and was at the heart of this forum for a bit too, and this isn't how this community is supposed to work. If you want help trying to prove or disprove your theory fine, but otherwise I expect folks will continue to dismiss your controversial statements without further proof. Could be wrong, but I don't think I've read the goat entrails incorrectly in this: it's not lost on me that nobody else said boo while you and I were dissecting this in the past few days, and I'm 100% confident they read it.