Fico Threshold technical question

[Ambrosius]

I maintain a paid subscription to FICO. I generally keep my credit card usage in 10% range, aim for under 8.9%

I recently had to rely on my credit cards. My primary card with credit limit at $8,200 went from 11% to a reported 20% credit usage. I try to keep my usage well under 28.9%. I got a 14 point drop on my Experian score and 7 point drop on Transunion.  There are no other extenuating factors, and the alerts specifically cite to this increase credit card usage. I know these things bounce back and there is no "memory" of credit utilization, but I thought I knew all the rules. This rather irks me, I am in the middle of arranging credit for a new large purchase and am trying to guage the timing. I still have 3-6 months to arrange this, so unforeseen hits like this are important.
I may be under a mistake impression, but I thought that a hit this large would only occur if I passed one of the threshholds. Is there a new threshold between 8.9 and 28.9? Is it common to have a 10+ point drop by going from 11 to 20% credit utilization? BTW, this is FICO 8, the overall score seen on the FICO main dashboard.

[Yasselife]

@Ambrosius 

Yeah, even if you don’t cross the big 8.9% or 28.9% thresholds, you can still see drops within those ranges depending on your profile. FICO has smaller “sub-thresholds” that aren’t officially published, so going from 11% to 20% can still trigger a score hit; especially if it’s on a card that reports as your highest balance or is a major contributor to your overall utilization. The good news is once you pay it back down, the points should rebound quickly.

[Patient957]

I think the 9er thresholds, like 9% and 29%, are for aggregate UT.  There are also thresholds for single-card UT.  I once lost 5 points on EQ8 when I let one card report 15%, even though my aggregate UT was around 3%.  Interestingly, neither EX8 nor TU8 budged.  Of course, once I got that 15% card back down, I got my 5 points back.

[CaneVariant]

---

@Patient957 wrote:

I think the 9er thresholds, like 9% and 29%, are for aggregate UT.  There are also thresholds for single-card UT.  I once lost 5 points on EQ8 when I let one card report 15%, even though my aggregate UT was around 3%.  Interestingly, neither EX8 nor TU8 budged.  Of course, once I got that 15% card back down, I got my 5 points back.

---

I had a similar experience with my EQ8 dropping 6 pts going from one card reporting 3% to 14%. Aggregate utilization was still below 8.9%. No movement on TU8 or EX8

[Thomas_Thumb]

@Ambrosius 

---

What was your before/after aggregate revolving UT? Many posters have reported a score penalty when aggregate revolving UT crossed above 5% but stayed below 9%.

Everyone should test their own files. Results differ by scorecard. Many profiles would see no score change while others, such as yours, show heightened sensitivity.

The nice thing about a high revolving UT score penalty is it is temporary. If you are concerned about 10 point shifts, try maintaining cards that report balances under 9% UT. If not, just optimize UT when pursuing new credit - and test your file repeatedly.

[Zoostation1]

@Ambrosius  I just had a drop in aggregate utilization from 8.4% to 4%, and while I didn't gain anything on TU8, I did gain 4pts on EQ8.  Experian hasn't reported yet so I don't have data on that one but nonetheless it follows the pattern of what Thomas was stating.

[FicoMike0]

Some of you big spenders will have to test this. I don't ever spend 5% period. Maybe next big vacation...

[Ambrosius]

One begins to wonder if the thresholds even matter. From what I am seeing and what has been posted, it is more of a sliding scale. So 8.9 isn't really a magic number then, nor 28.9, are they?

[Thomas_Thumb]

I don't recall seeing your before/after aggregate utilization being listed. What was the change?  How many cards do you have and any change in number reporting balances? Do all CRAs list the exact same account data on the dates associated with the before/after scores? How do you know? Often CRAs have timing differences in file updates. 

You have not provided much information on your overall file. Have you opened an account in the last 12 months? Any recent inquiries?

I  would not see a point change for taking a single card from 11% to 20% - unless it was a high limit card that pushed AG UT above a milestone threshold. I suspect something else in your file(s) also changed between the 2 pulls.

[RedKhaos]

If you want to maximize your credit score for a loan, you should just do AZEO for two months and call it a day.

OP, you actually did cross one of the speculated threshholds, which are in increments of 10%. The thresholds are thought to be 8.9%, 18.9%, 28.9% etc. FICO does seem to much more heavily weight the aggregate compared to the individual cards. Going from 11% to 20% utilization crosses one of these thresholds so getting a 10 point penalty would be consistent with my current understanding. If it's a sub-5 point swing, I would say it becomes a bit hard to distinguish and it could easily be some other factors or just the inherent variability of the model.

For smaller score changes, I tend to put less weight on the exact ouput conclusions and a bit more on the logical implications. Statistical models like FICO by definition are not deterministic. So even IF you can control every single variable, there's an inherent error margin baked into the output. I.e. if you run the same exact FICO experiment 1000 times, you should NOT expect the same score result 1000 times. You should expect very "similar" results most of the time that make sense and aren't counter-intuitive. 

For example, if your utilization went from 8% to 3%, "all things being equal" which is easier said than done, consider the following possibilities:

1) No change => I'd expect this for clean, mature, thick credit profile and also for most profiles as well.

2) <5 point increase => Plenty of data of these. I lean toward this as maybe less likely but also reasonable. This is "logical" since a lower (non-zero) utilization should concide with a *slightly* better score and by proxy a more credit worthy person. Utilization should be inversely related to credit score and this relationship should be consistent. This is also known as a monotonic relationship (as one variable increases, the other variable consistently either increases or decreases - or maybe stay the same - but not go up and down randomly)

3) <5 point decrease => I would be much more surprised by this from FICO. As it leads to an "illogical" model interpretation that a lower (non-zero) utilization could coincide with a less credit worthy person.

Can 3) still happen once in a while with someone's specific profile? Absolutely, because any model can error. But it is far, far more likely for something like FICO to lean toward 1) and 2) than 3) because of built in mechanisms. My point is as long as you see 1) or 2), then yeah the model is likely behaving as expected and there's probably some logical reason for it.