What kind of hit for carrying balances [update 7/3]

[NRB525]

Now I'm wondering if derogs in the file affect how FICO looks at number of accounts. If derogatories, then increasing the number of accounts reporting brings little skeletons out in the algorithm. Speculation, of course, but plenty of files seem to have more noise if they have baddies.

3 cards costs 3 points, in your experience. Seems like a small amount of points to not have to worry about keeping all but one at zero.

[Revelate]

---

@NRB525 wrote:

Now I'm wondering if derogs in the file affect how FICO looks at number of accounts. If derogatories, then increasing the number of accounts reporting brings little skeletons out in the algorithm. Speculation, of course, but plenty of files seem to have more noise if they have baddies.

3 cards costs 3 points, in your experience. Seems like a small amount of points to not have to worry about keeping all but one at zero.

---

Keeping yes, I let my balances fall where they may when I don't care; however, mortgage process or any other application you actually care about you'd be stupid not to in my opinion.  Quite literally on that one: difference between my 3.875 and 4.125% on my so cali sized mortgage, thank you but no thank you.  That's a massive mistake financially: thousands of dollars a year over 30 years.

Also FWIW dirty files tend to sort into limited buckets as a result of the old derogatories... they tend to be flat moreso than clean ones.

It's pretty consistent so far, JagerBomb's who had a gorgeous file pre forclosure and I believe 5 tradelines saw a drop at 2/5.  Jamie at 2/7.  Me at 3/9, there's a couple of people in the 14-15 range who have suggested it's higher for them.  This doesn't have to be complicated: just because the FICO algorithm isn't fully disclosed and admittedly has some complexity on the whole, doesn't mean there aren't dirt simple mechanics in it and this might just possibly be one of them.

[NRB525]

difference between my 3.875 and 4.125%

Yes, that is a substantial interest rate difference on a mortgage. Honestly, though, you wouldn't be at that boundary except for the unfortunate times in 2010. So the number of cards becomes a way to make the best of a bad situation from file history, not the root cause of being at that boundary.

Cheers!

[Revelate]

---

@NRB525 wrote:

difference between my 3.875 and 4.125%

Yes, that is a substantial interest rate difference on a mortgage. Honestly, though, you wouldn't be at that boundary except for the unfortunate times in 2010. So the number of cards becomes a way to make the best of a bad situation from file history, not the root cause of being at that boundary.

Cheers!

---

Which fits 99% of the people who find this place.  

Need better emoticons here to actually convey what I think regarding your statement 

This board is about understanding the underpinnings of the FICO algorithm.

[Thomas_Thumb]

With regard to understanding underpinnings of FICO, a good starting point might be to define the framework of a model and then validate/adjust the defined framework. Once that is in place then start listing and quantifying affect of factors on score by profile category (or bucket assignment). So how about reverse engineering a simplified Fico model on MyFICO?

Here is a start - if people like the approach perhaps someone could champion a "build a model thread". 

Background assumption 1: Fico 8 has 12 scorecards - 8 for clean files and 4 associated with non clean files. Credit profiles are assigned to a scorecard and then scores are determined within the confines of the assigned scorecard. So what defines your scorecard assignment? 

Background assumption 2: There are three criteria  used for score card assignment for clean files. These are: Total # accounts (active & inactive) in the file - thin file or not, average age of accounts in the file - seasoned veteran or not and age of youngest account - changes in spending habits or not. If each of these factors is assigned a + or - then you have 8 combinations for these three factors - same as # of scorecards for clean files. Additional criteria are used for assigning non clean files to the four remaining scorecards.

Presented below is a hypothetical scorecard framework for clean files. 

Fico 8 model - framework assumptions
Score card #	Lower limit score	Upper limit score	Total # Accounts on report	AVG age of accts (yrs)	Youngest account age (yrs)
1	760	850	8 or more	8 or more	2 or more
2	720	829	8 or more	8 or more	under 2
3	720	829	8 or more	under 8	2 or more
4	720	829	under 8	8 or more	2 or more
5	680	799	8 or more	under 8	under 2
6	680	799	under 8	8 or more	under 2
7	680	799	under 8	under 8	2 or more
8	640	759	under 8	under 8	under 2

Let's further hypothesize that what does/does not SIGNIFICANTLY impact score is conditional on whether a file (is/is not) clean and to a lesser extent the scorecard within the (clean/not clean) category. Here are some thoughts on factor with associated deducts  for illustration - nothing validated just examples presented to illustrate a possible method for consolidating information. No doubt there are interactions among factors but trying to capture that adds undue complication - at least initially.

Comments	1) no installment loan => deduct 30 points from scorecard max
	* installment loan under 1 year age deduct 10 points
	* Installment loan over 90% remaining deduct 20 points
	* Installment loan over 60% to 89% remaining deduct 10 points
	2) Aggregate credit line utilization
	* Utilization 10% to 19%, deduct 5 points
	* Utilization 20% to 29%, deduct 10 points
	* Utilization above 30%, deduct 30 points
	3) Cards reporting non zero balance (depends on score card)
SC #1	* Five or more with 100% active cards reporting deduct 5 points
SC #2-4	* Five or more regardless of % reporting, deduct 5 points
SC #5-8	* Three or more with over 50% reporting deduct 10 points
SC #5-8	* Five or more regardless of % reporting, deduct 10 points
	4) Individual card credit line utilization
SC #1	* No deduct regardless of utilization
SC #2-7	* Utilization @ 80%, deduct 10 points
SC #8	* Utilization @ 80%, deduct 15 points. Deduct 10 points @ 50%

Food for thought.

[CreditDunce]

Are these FICO 08 scores?  I know when I let most of my cards report balances with overall utilization less than 5% (although some cards at 20-35%), my EX-08 and TU-08 scores are hurt, but not enough to worry about.  My EQ-04 score takes a much bigger hit.  At least that is my impression, but I don't have enough data for proof.

[Thomas_Thumb]

Please don't accept the examples in the proposed model as accurate. The numbers presented are merely placeholders.

The intent of this proposal is to develop a reasonable educational tool - over time- based on actual data provided by posters.Not an easy task and not sure anyone has the dedication/motivation to manage the initiative.

That aside, my experience is that Fico 4 is more responsive to changes in credit card utilization relative to Fico 8.

[CreditDunce]

---

@Thomas_Thumb wrote:

Please don't accept the examples in the proposed model as accurate. The numbers presented are merely placeholders.

The intent of this proposal is to develop a reasonable educational tool - over time- based on actual data provided by posters.Not an easy task and not sure anyone has the dedication/motivation to manage the initiative.

That aside, my experience is that Fico 4 is more responsive to changes in credit card utilization relative to Fico 8.

---

I am sorry, I was asking if the scores in the OP were FICO 08.  I assumed they were.  I agree my experience is my EQ-04 score is hurt more with higher utilization/more cards reporting a balance.

I am not quite sure how your ealier post fits in with the OP.  You may want to start your own post if you wish to discuss scorecards per say.   Or you could look for threads more aligned to scorecards.  IIRC there was a thread not too long ago that provided some evidence that we all belong to multiple scorecards that all get applied.   I haven't read there evidence yet, but I did mark it for further reading:

http://www.fico.com/en/node/8140?file=7900

[NRB525]

Based on the proposed ScoreCards TT,

I would fit into the Scorecard #5 for general account ages, so I take it the 799 is the max?

I have an installment loan under 1 year, under 60% remaining, so -10

Utilization 20%-29% so -10

For cards reporting, SC 5, I've got 5 or more reporting, but none over 50% (anymore) so -10

No lines are at 80%+ so no deduct on individual card utilization.

Calculates to 799-10-10-10 or 769.

My latest: EQ 750, TU 760 EX 764.

The missing factor is some INQ counts affecting the scores, EQ=3 TU=5 EX=5 (several EX are very aged)

I did just see a 36-40 point increase in the last month on all three bureaus, so I think the INQ are a serious factor for me, offset by the rapid reduction in total Utilization, and getting to the point where none of my cards or LOC have over 50% outstanding. 45% is currently highest utilization on two cards.

[NRB525]

---

@Revelate wrote:

---

@NRB525 wrote:

difference between my 3.875 and 4.125%

Yes, that is a substantial interest rate difference on a mortgage. Honestly, though, you wouldn't be at that boundary except for the unfortunate times in 2010. So the number of cards becomes a way to make the best of a bad situation from file history, not the root cause of being at that boundary.

Cheers!

---

Which fits 99% of the people who find this place.  

Need better emoticons here to actually convey what I think regarding your statement 

This board is about understanding the underpinnings of the FICO algorithm.

---

I use the term unfortunate in the commiserating sense, not a judgemental sense. We are all indeed here to learn, and learning comes from facing reality. Like when I realized that it's time to pay down the cards when the total outstanding balance is $118k on $168k of pre-CLD, pre-account closure, and pre-payment accelleration credit lines