@Anonymous wrote:
@Anonymous wrote:I am at 74% on my auto loan right now. I will see if getting it below 70 (which should be in the next month or two) will make a difference. For science.
It's also extremely difficult under most circumstances to determine if such a point is indeed a threshold, because any gains realized are likely very small; perhaps we're talking 3-4 points for example. That being said, there are always a ton of different things going on with profiles that can result in 3-4 point gains like inquiries becoming unscoreable, revolving balance changes, number of accounts with a balance decreasing, AAoA/AoYA/AoOA increases, etc. For someone to really know for sure, they'd need to have access to daily FICO scores/reports in order to isolate a score change to only the [reported] balance change on the installment loan. It could also be isolated with two (before & after) carefully timed $1 CCT trials, but not many people are willing to test on that level
I'm tracking my credit through like 8 different places so I will do my best to determine what happens. Espcially since I know the date my loan balance reports to the bureaus I can narrow the window down a bit.
I have been checking lately with my various installment loans and aggregate (which is individual if only one loan).
There is no breakpoint between 78% and 100%.
@Thomas_Thumb wrote:Based on isolated data from CAPTOOL and more ambigious data from other sources there appears to be an installment loan threshold somewhere between 65% and 69%.
Fits my original data which found a line somewhere between 64.67% and 82.93%
I decided to clean up my balance sheet and am throwing my ridiculous 3 gig income towards my auto loan now and USBank is good for reindeer games on that front. I'll concretely test 75%, and then push to 70.1% or whatever and then just make the monthly mortgage payment to walk the range from that 70.1% down through 65%.
Best I can do for concrete testing, but should be pretty close since my mortgage payment reports separately from literally everything else on my credit report.
@Anonymous wrote:
@Revelate wrote:Best I can do for concrete testing, but should be pretty close since my mortgage payment reports separately from literally everything else on my credit report.
True, but it is worth noting that you're talking a mortgage here and not a non-mortgage, which could (and likely does) have different threshold points "worth" different point values at those thresholds.
Well to be clear every statement that a mortgage is handled differently is conjecture and there's a bunch of admittedly not clean data that suggests that they're counted exactly the same way... to say nothing of the singular reason code that exists for both mortgage and non-mortgage loans (bureau language not mine).
Actually I'll toss out one more datapoint: if I exclude the mortgage the payoff of my personal loan I went from 59.5% to 80.4% and no change on any model which contradicts multiple datapoints and I know my original one was a good one.
Really it's a fairly simple test if a long one, get the auto loan paid down to under 8.9% and see if there's a change which is seriously doubtful because of the above datapoint, walk across the assumed breakpoint with the mortgage, pay the auto loan off and see if the score change is the same, and then once the mortgage clearing starts happening see if the line is the same but I'll bet you and everyone else that it is frankly. I've never understood the distinctions people tried to draw between them TBH at least under modern algorithms. FICO NG I do not care, nobody uses it might as well be FAKO and anything that predates it like FICO 2 might be awkward too.
@Anonymous wrote:
@Revelate wrote:Well to be clear every statement that a mortgage is handled differently is conjecture and there's a bunch of admittedly not clean data that suggests that they're counted exactly the same way... to say nothing of the singular reason code that exists for both mortgage and non-mortgage loans (bureau language not mine).
I thought we've seen multiple people report negative reason codes related to loan balances being too high going away on a mortgage with utilization being somewhere in the 70% range, where that reason code for other loan types is often visible until the 8.9% threshold is crossed. I've never tested it personally, just going off of what I've read on the forum.
I guess my question would be whether someone with (say) 50% left on their mortgage sees any negative reason codes pointing to the loan balance being too high. This would of course mean that the mortgage needs to be the only open installment loan on file. We know for certain that crossing 8.9% on a non-mortgage results in a gain (SJ verified this gain to be significant, too) meaning that at 9%+ prior to that gain a negative reason code could/would be present due to those FICO points being held back.
The problem in general with more minor reason codes is they can get pushed off the table, especially if we're talking higher breakpoint already passed... I think if we nail that down we'll see.
Actually if we want to be specific I'm the one who originally verified that reason code goes away completely under that... what we don't have good information on is mortgages because it takes a long ass time to make a meaningful dent in most people's mortgages, and then worse we get stuck with FICO 8 not giving reason codes anyway to anyone over 800 as a result of stupid CMS implementations, whereas FICO 04 doesn't appear to care about it though I do see it on EX FICO 3 which is a thing.
We'll see as I might have a fantastic testing opportunity given I have some dirt holding down EQ/TU; if I do find a difference with mortgage handling I'll be the first to announce it, but Occam's Razor does apply here.
Really thinking about it, it's simple: we know the 9% line is absolutely significant, the higher one is like 6 points based on my testing though that scales by scorecard probably to 10 points I suspect. Finding the top breakpoint and then walking the mortgage across it, if I get the same score magnitude score shift think that can pretty concretely figure it out one way or the other.