practical value of a huge credit limit?

[JLK93]

---

@Anonymous wrote:

JLK93, it would seem that cashnocredit and TT simply have opposing data sets to some degree.  It doesn't necessarily mean that either is "wrong" just that they are different.  This wouldn't be the first time we've come across this on the forum, that's for sure.

---

Exactly. They are different data sets. cashnocredit's data relates to a score change caused by charge card balances being included in aggregate utilization by EQ FICO 8. TT's data relates to a score drop due to only a charge card reporting a balance. EQ FICO 8 does not factor individual charge card utilization. The reason code TT received was "There are no recent balances on your revolving credit accounts." That reason statement is correct. However, it can't be extrapolated to mean that the balance is not being scored as part of the aggregate utilization.

TT said that his charge card balance was not included in aggregate utilization by EQ08. I asked for the source of that information. I never received an answer to the question. I suspect that the source of that information is summary information included with a 3B pull. That is not data. It is a consumer product.

Then, I was hit with a barrage of screenshots. The next morning, I received a simplified chart, as though I couldn't understand the data. I understand it just fine. I completely disagree with the conclusion not the data. This is not new data. Many people have had score drops due to only a charge card reporting a balance.

There are at least 3 most possible causes for the score drop on EQ.

1. Charge card balances are not included in aggregate utilization. This doesn't seem to be a reasonable conclusion since it contradicts strong data. It was proven conclusively through testing that EQ FICO 8 scores aggregate utilization for Amex charge cards. Logic can't invalidate testing.

2. Charge cards are not included in the number of cards reporting a balance. It has supposedly been proven that charge cards are, in fact, included in the number of cards reporting a balance. So, this would not seem to be a reasonable conclusion.

3. FICO 8 does not factor individual charge card utilization. The reason statement "There are no recent balances on your revolving credit accounts" seems to suggest that this is the cause of the score drop. This would seem to be the most reasonable conclusion. 

There is no reason to arbitrarily pick a conclusion that contradicts strong real world data. Conclusions can't be made in a vacuum. Testing shows conclusively that EQ FICO 8 scores aggregate utilization for Amex charge cards. Logic can't invalidate the results of testing. It's a choice of logic or testing. Both can't be right.

[Thomas_Thumb]

As mentioned, the conclusion starts with a strong data that is reinforced by reason statements and summary data provided on the reports.

I recognize other have and will continue to present and analyze different data sets. People are free to reach their own conclusions based on what is presented - hopefully without bias. Unfortunately, viewpoints are often ingrained which gets in the way of objectivity regarding "what conclusions are reasonable" based on the information presented.

Based on the data I presented, my conclusion is: Enhanced Fico 08 scores dropped across the board between 7/23 and 8/6 due to no revolving accounts having a balance as indicated in the reason statement and 0% aggregate utilization.

A review of each account on the report shows fewer open accounts with balances in August relative to July for all CRAs and lower total open account balances being reported. Both these factors should have a positive impact on score - certainly not negative. Indeed, earlier Fico algorithms do show score increasing in August.

Why do I think this is happening? 

1) Earlier versions of Fico are recognizing the AU card balance in August whereas Fico 08 is not.

2) Fico 08 is NOT recognizing the only other card showing a balance in August in calculation of aggregate utilization, namely the AMEX NPLS charge card.

3) Unable to say whether Fico 04 and/or Fico 98 utilize the NPSL charge card in revolver Ag UT% because those scores did not drop. The lack of drop could be do solely to the AU card counting but, the AMEX may be counting as well. 

The data is clear: Enhanced Fico scores dropped on Fico 08 but not on Fico 04/Fico 98. 

[JLK93]

---

@Thomas_Thumb wrote:

 reinforced by summary data provided on the reports.

---

Summary data is not data. It can't reinforce anything. It is a consumer product. 

[tacpoly]

---

@JLK93 wrote:

Summary data is not data. It can't reinforce anything. It is a consumer product. 

---

I have no horse in this race, but I don't understand what you mean when you call something as a "consumer product".  How is it different from data?  I guess it would be best if you could define "data" and "consumer product".  It just seems like you're talking past each other and it would be helpful to operate on the same vocabulary and assumptions. 

And I went to the link you posted and, granted took only a cursory look, but from what I gather, cashnocredit and Thomas Thumb are both basing their conclusions on score changes to some extent.  Is this wrong?

[JLK93]

---

@tacpoly wrote:

---

@JLK93 wrote:

Summary data is not data. It can't reinforce anything. It is a consumer product. 

---

I have no horse in this race, but I don't understand what you mean when you call something as a "consumer product".  How is it different from data?  I guess it would be best if you could define "data" and "consumer product".  It just seems like you're talking past each other and it would be helpful to operate on the same vocabulary and assumptions. 

And I went to the link you posted and, granted took only a cursory look, but from what I gather, cashnocredit and Thomas Thumb are both basing their conclusions on score changes to some extent.  Is this wrong?

---

The summary is something created for the benefit of the comsumer. It has no connection to the FICO scoring algorithms.

[Thomas_Thumb]

---

@JLK93 wrote:

---

@Thomas_Thumb wrote:

 reinforced by summary data provided on the reports.

---

Summary data is not data. It can't reinforce anything. It is a consumer product. 

---

@Thomas_Thumb wrote:

A review of each account on the report shows fewer open accounts with balances in August relative to July for all CRAs and lower total open account balances being reported. Both these factors should have a positive impact on score - certainly not negative. Indeed, earlier Fico algorithms do show score increasing in August.

Why do I think this is happening? 

1) Earlier versions of Fico are recognizing the AU card balance in August whereas Fico 08 is not.

2) Fico 08 is NOT recognizing the only other card showing a balance in August in calculation of aggregate utilization, namely the AMEX NPLS charge card.

3) Unable to say whether Fico 04 and/or Fico 98 utilize the NPSL charge card in revolver Ag UT% because those scores did not drop. The lack of drop could be do solely to the AU card counting but, the AMEX may be counting as well. 

The data is clear: Enhanced Fico scores dropped on Fico 08 but not on Fico 04/Fico 98. 

---

---

Data is also in alignment with the reason statement - which pulls from the Fico database.

As mentioned, Fico score results agree with what one would expect with a drop from 1% utilization to 0% utilization - which can only be true if the AMEX charge and AU cards are not included in the calculation.

[tacpoly]

---

@JLK93 wrote:

---

@tacpoly wrote:

---

@JLK93 wrote:

Summary data is not data. It can't reinforce anything. It is a consumer product. 

---

I have no horse in this race, but I don't understand what you mean when you call something as a "consumer product".  How is it different from data?  I guess it would be best if you could define "data" and "consumer product".  It just seems like you're talking past each other and it would be helpful to operate on the same vocabulary and assumptions. 

And I went to the link you posted and, granted took only a cursory look, but from what I gather, cashnocredit and Thomas Thumb are both basing their conclusions on score changes to some extent.  Is this wrong?

---

The summary is something created for the benefit of the comsumer. It has no connection to the FICO scoring algorithms.

---

I see; you have an issue with using the "reasons for score change" to support his position (and I agree with you).  But are Thomas Thumb's conclusions supported by FICO score movements alone? 

[Thomas_Thumb]

It is generally well understood that reason statements provided on reports often do not describe the true cause for a score change. Many reason statements start with: "Too many", Too Few", Not enough", Too high", Too low" or "Too short". These statements are ambigious and you don't know the specific threshold(s) the FICO model uses to trigger reporting of said statements.

For example, "proportion of loan balances to loan amounts is too high", (B/L ratio) is one reason statement. I have a mortgage under 35% B/L and did get this reason although my B/L is sufficiently low not to impact score. Perhaps the lowest threshold is 10% which appears to be a true for SSL and possibly Auto loans.

FICO does have some definitive reason codes/statements that are not ambigious. One of those is: No recent revolving balances

I contend that in order for this statement to be listed on a report; the data set analyzed by the FICO algorithm had to trigger pulling the statement. Since the statement is not ambigious - (condition is either NO or not NO), FICO did not count the AMEX or AU accounts - both of which show balances. That is why I include the reason statement as supporting evidence in addition to exhibiting the changes in Fico scores.

I do not use the statement as proof of why score changed  - particularly since Fico mortgage scores and Fico 08 scores moved in opposite directions. However, I do make the arguement: the FICO model must have determined the "no recent revolving balances" condition existed which cannot be true if the AMEX or AU balances are included.

The data does speak for itself but, the reason (Score Factors) statement aligns with the conclusions drawn from the data.

[Anonymous]

To walk back a little bit closer to the original focus of this thread, it was primarily (a) whether people knew of practical uses for extremely high individual CLs and secondarily (b) whether there was a definite and reliable answer about when FICO models dropped a tradeline from one's total CL.  I.e. where was the cutoff as of today?

To circle back to that part B.... do we know?  And is it model dependent?  For example, is it 50k for FICO 8 but a different number for FICO 98, FICO 04, or FICO 9?  Is it CRA dependent?  I.e. since the EQ FICO 8 model is slightly different from the TU FICO 8 model, does it dependent in part on which CRA we are tallking about?

I agree with JLK93 that "to know" should not be based on the "summary" information that a particular credit monitoring service provides (myFICO, Credit Check Total, Credit Karma, Credit.com, etc.).  Thus, as I said far earlier in the thread, I was pretty sure that in many cases people who said that a particular big tradeline was "being included" were basing that on seeing it being listed as part of their "total CL" in the summary.

But I also agree with TT that formal reason FICO reason codes, when absolutely unambiguous, are reasonable as part of an argument

JLK93 gave a very straightforward method of testing where the cutoff might be.  Basically, if a person has several cards, with one of them being at 50k (say)....

     * He keeps all cards reporting $0, except for the 50k card and a smaller card (e.g. 20k).

     * He keeps each of those two cards reporting a small balance ($100 say) for a couple months. 

     * Then he pays the 20k card down to $0 leaving just the 50k card reporting

If we see a big score drop (and also see the No recent revoving balances reason code) then the cutoff must be at 50k or perhaps lower.

The test just has to be repeated by enough people with a variety of individual CLs.

What do we know thus far?  FICO 8 should be very easy to get an answer for, since there are so many people capable of testing with that model.  The mortgage scores and FICO 9 would probably need several people each with access to the $40/month myFICO product that gives monthly updates for all FICO flavors.

[Anonymous]

If I ever get a credit line with a monster limit, I'll be more than happy to test this out