topic Re: Boost vs Ding - Your Thoughts in Understanding FICO® Scoring

Boost vs Ding - Your Thoughts

Anonymous — Fri, 05 Oct 2018 06:07:27 GMT

I've been doing quite a bit of research particulary on 850 Fico8 profiles. It would seem that if the max attainable score is 850 and we know that many people here have 850 scores with one or more of the following:
-Inquiries
-Loans with high balance ratios
-AoYA <1yr
-Less than 4 or 5 revolvers

...then clearly 850 is a "soft" ceiling and thus certain things that most people see as "dings" are in fact improperly thought of as dings. My opinion is thatinstead of a ding it is a "loss of bonus points"

Example given...AoYA....many people see a nice boost when their AoYA hits 1 yr. In my opinion, they were not losing 10pts for AoyA<1yr but rather were not getting bonus points.

I think this also applies with several other factors as well but Ive not done enough further research to identify them. Another one might be the huge bonus people see when their installment loans report <8.9% balance. I believe these are bonus points and not base points.

My theory suggests the following are bonus points:
-Cards reporting balances is 0 to 20%(i.e. AZEO)
-Total amounts owed <$20,000 (incl non-mortgage loans)
-Loans <8.9% remaining
-Revolving util <5% (or between 1 to 5% rather)
-0 or 1 hard inquries we've seen profiles stay at 850 after a hard pull) possibly even 2
-Oldest account >10yrs

I believe these are dings:
-0% revolving utilization
-AoYA between 1month and 6months
-Only 1,2 or 3 revolving accounts (some say less than 5)
-No open loans
- More than 1 or 2 inquiries
-More than 50% cards with balances

I believe these are neutral score factors:
-AoYA 6months to 11months
-Less than 1 or 2 inquries in the last 1 year
-20% to 49% cards reporting balances

I am happy to hear any feedback. I think their are arguments to be made as to whether something is a ding or a boost. I realize there may be supporting/contradicting information out there already so I am not claiming to be accurate here. I also think that in order to remove the soft 850 ceiling, fico introduced some models which go up to 900.

Re: Boost vs Ding - Your Thoughts

Anonymous — Fri, 05 Oct 2018 07:47:20 GMT

I suppose it's a glass half full glass half empty argument.

To me, anything is a ding if it generates a negative reason code. If you're losing points from anything it means you're missing out on the max and are experiencing a ding. That's how I see it.

The fact that some "buffer" can exist to a point greater than 850 simply allows for a few dings to be present and one still possess an 850 score. They are still dings, though. A single scoreable inquiry, AoYA < 12 months, etc. are ding factors that are lowering score. Just because someone may still have an 850 score to show doesn't mean that they aren't still experiencing that ding. They are, it's just that it isn't seen due to top-end buffer being present.

Re: Boost vs Ding - Your Thoughts

Anonymous — Fri, 05 Oct 2018 14:41:19 GMT

I agree with BBS. This is ultimately just an issue of how a person wants to define certain words. It's like asking whether the freezing point of water is really zero degrees or 32 degrees. You have to define first what "degree" means (in the context of F or C) and then you get an answer.

BBS suggests that the simplest way to think about any FICO model is that the starting place for anyone is a perfect score with a perfect profile that perfectly meets every scoring factor. Then as you deviate from that perfect profile you experience scoring penalties. This happens to people even in the 850 group, since the starting place is likely 870 or so.

PS. Curiously to have an absolutely perfect profile turns out to be an abstraction: it's actually impossible to perfecly meet every scoring factor. For example to perfectly meet the scoring factor of having a low number of open accounts with a positive balance you'd want zero such accounts. But if you do that, you'll have to have zero open loans and all cards at zero, which would give you two different penalties. So in that case you have to compromise by having exactly one open loan and one card reporting a balance.

Re: Boost vs Ding - Your Thoughts

Anonymous — Fri, 05 Oct 2018 16:23:39 GMT

Good post from CGID. To piggy-back on his final point, on my file I have one open installment loan. For maximum scoring benefit with respect to my revolving accounts, I want one to report a small balance against the 8 revolvers that I have. In doing this, I avoid the 15-20 point penalty associated with all revolvers being at $0 balances. We refer to this as AZEO of course. However, I naturally have a balance on my only installment loan. This means that I really have 2 accounts (out of 9 total) that have balances. I receive the negative reason code that I have too many accounts with balances. If I were to drop the installment loan or drop to no revolvers with a reported balance, I'd nix the too many accounts with balances reason code, but replace it with no open installment loan present or no revolving credit use. CGID is correct that it is impossible to have all of these reason codes eliminated at the same time.

Re: Boost vs Ding - Your Thoughts

Anonymous — Sat, 06 Oct 2018 05:48:07 GMT

I have to backup my original claim a bit - if the max score is 850, and humans exist with 850 scores who have inquiries, new accounts, and high balance loans, then clearly it cant be an "only deductions" or "only additions" ball game. If the range is 300 to 850, that leaves 550 allocable points. If that were the case, a single inq or new acct or non-AZEO profile would 100% result in a less than 850 score.

There must be certain factors which add up to >550 points in order for this to hold water

Re: Boost vs Ding - Your Thoughts

Anonymous — Sat, 06 Oct 2018 05:53:26 GMT

If we are assuming an "only penalties" game its simply impossible to have a single inquiry and an 850 score. Example of my theory is as such......

In a penalties only game:
Inquries scored as:
0 inq = no points lost
1 Inq = minus 5 points~
2 inqs = minus 10 total points~

In a ding/bonus game inquries are scored as per my theory follows:
0 inqs = +5 bonus points [i.e. greater than 850 points)
1 inq = no points
2 inq= -5 points

Re: Boost vs Ding - Your Thoughts

Anonymous — Sat, 06 Oct 2018 08:22:04 GMT

I think the part you're missing out on here or failing to take into consideration is that the max score able to be obtained isn't 850. It's likely something closer to 870 like CGID suggested earlier. 850 is the max score that will ever be displayed, but due to a top end buffer built in one can actually possess a score greater than 850 by probably around 20 points give or take.

So yes, one can go by the only penalties perspective and still possess an 850 score while incurring a ding or two.

Re: Boost vs Ding - Your Thoughts

iv — Sat, 06 Oct 2018 21:06:00 GMT

@Anonymous wrote:
If we are assuming an "only penalties" game its simply impossible to have a single inquiry and an 850 score. Example of my theory is as such......

In a penalties only game:
Inquries scored as:
0 inq = no points lost
1 Inq = minus 5 points~
2 inqs = minus 10 total points~

In a ding/bonus game inquries are scored as per my theory follows:
0 inqs = +5 bonus points [i.e. greater than 850 points)
1 inq = no points
2 inq= -5 points

You are over-thinking this entirely. While the exact scoring criteria and weighting aren't fully known, the basic method is fairly simple.

Picture a (very) simplified 0-100 scaled model, with 5 scoring criteria:

Criteria A: 0-25 points
Criteria B: 0-25 points
Criteria C: 0-25 points
Criteria D: 0-25 points
Criteria E: 0-25 points

Notice that a "perfect score" on all five catergories actually adds up to 125 points? But the scoring model is defined as 0-100...

Anything over the "official" max is simply capped at the max - no need for complicated "boost vs ding" methods.

Each individual criteria is just assigned a range of points from 0 to it's category max (which is different per scorecard!), and the individual criteria scores are added together, and then capped to the official range.

(This is also how the reason codes are determined - those criteria with the largest point gap between your score and the category max determine the displayed reason codes.)

On the flip side, some of the older scoring models have actual maximums that are LOWER than the "official" max, where actually reaching 850 is impossible.

Re: Boost vs Ding - Your Thoughts

NRB525 — Sun, 07 Oct 2018 12:16:29 GMT

OP you have a long list of boost vs ding items. In general I think your concept is correct, but I suggest thinking about it like this:

The basic level of your FICO score is determined by how long you have had accounts open, whether you have made all payments on all obligations (in terms of what your credit report can see) on time, and what your utilization is for those accounts. Once you have all those inputs, you get to a basic FICO score.

Let's say that is an 800 score. The cardholder does not know that yet.

I think this 800 score is an "anchor score" which, in the short term, the cardholder can work some of the credit decisions around, so there is also a band of scoring around 800 where the cardholder could influence the score. For discussion, let's say the band is 20 points either way, up or down. A quick reduction in utilization may result in some score improvement. Apping a new card may reduce the score somewhat, but as long as the cardholder continues to pay on time, the anchor of 800 is still pretty much there, and slowly improving.

Now we take into account the nuances of what you have as potential boost or ding. As the cardholder score approaches 850, that band would expand to include potential scores over 850. FICO 8 cannot score 851 however, so even if the cardholder gets all the boosters, the score pegs out at 850.

The list of ding items, if it is only one and only minor, I think we have seen examples of those minor situations not moving a person off the 850 peg. So there are other boosting items the same person has, ensuring the score remains at 850.

Having said all that, I would think that only one ding is allowed at 850, and those would have to be minor. The main factors to get to 850 are no lates over a long time, very little borrowing, token borrowing really, to get that FICO "Risk of Default" measure to as close to zero as it can possibly be. If the user is borrowing, 850 isn't really available, it is something lower, even if still over 800.

Re: Boost vs Ding - Your Thoughts

Anonymous — Mon, 08 Oct 2018 02:45:27 GMT

NRB525, while I agree with the meat and potatoes of your post, I definitely wouldn't say that very little borrowing or token borrowing is required. Mortgages almost by definition are large amounts borrowed and someone can easily possess an 850 FICO score with their mortgage still at 75% utilization or so. This could be someone that owes around $500k on a $650k mortgage, for example. That's definitely significant borrowing/debt.

Re: Boost vs Ding - Your Thoughts

NRB525 — Mon, 08 Oct 2018 05:13:08 GMT

That could be. Mortgages are likely a different perspective on FICO 8 influence.

Re: Boost vs Ding - Your Thoughts

Thomas_Thumb — Wed, 10 Oct 2018 03:16:32 GMT

@NRB525 wrote:
That could be. Mortgages are likely a different perspective on FICO 8 influence.

My data suggests mortgage B/L is treated differently than other loans on both Fico 8 and Fico 04. Probably treated differently on Fico 9 as well although my mortgage B/L had dropped below 40% when I 1st saw a Fico 9 score. Really no change in Fico scores associated with mortgage from ~~55%~~ 52% to 21%.

As we know there are a couple Fico reason statements that specifically state "non mortgage installment loans" when referring to negative impact of a high installment loan balance or balances.

Re: Boost vs Ding - Your Thoughts

Revelate — Mon, 08 Oct 2018 18:06:23 GMT

@Thomas_Thumb wrote:
@NRB525 wrote:
That could be. Mortgages are likely a different perspective on FICO 8 influence.
My data suggests mortgage B/L is treated differently than other loans on both Fico 8 and Fico 04. Probably treated differently on Fico 9 as well although my mortgage B/L had dropped below 40% when I 1st saw a Fico 9 score. Really no change in Fico scores associated with mortgage from about 55% to 21%.

As we know there are a couple Fico reason statements that specifically state "non mortgage installment loans" when referring to negative impact of a high installment loan balance or balances.

Well hang on, my own data that I originally documented installment utilization as a thing on suggests there's no change associated with non-mortgage installment loans in FICO 8 from the same ranges you have, so how are mortgages different in that regard?

This is the FICO 8 one straight from Transunion:

Proportion of loan balances to loan amounts is too high.

That's all loans AFAIK which matches the data.

FICO 9 with myFICO's massaging it a bit, looks conflated there too:

The remaining balance on your mortgage or non-mortgage installment loans is too high.

Where it goes sideways is FICO 98, namely EX 2:

The balances on your non-mortgage credit accounts are too high

It looks like there may have been a change in how they are handled, and I've never seen any installment utilization changes on FICO 04 nor have I seen the reason codes on either EQ FICO 5 or TU FICO 4 (even when I've had 3 reason codes only and it clearly showed up on FICO 8 models at the time) but at least today with FICO 8/9 it appears both from data and reason code nomenclature that all installment loans are counted basically identically.

Re: Boost vs Ding - Your Thoughts

Thomas_Thumb — Tue, 09 Oct 2018 00:39:50 GMT

As I recall you have a mortgage with over 80% B/L remaining. I would expect that your score is being affected by the high B/L and receive a reason statement to that effect. Yes, Fico does look at B/L for mortgages at some level. However, that reason statement does not prove lack of differentiation between mortgages and other types of loans. What proves differentiation to me is the "NON-MORTGAGE LOANS" reference in the last statement in the paste below (received for EQ Fico 04).

Re: Boost vs Ding - Your Thoughts

Revelate — Tue, 09 Oct 2018 01:34:13 GMT

Think we may be talking different elements of the FICO algorithm, maybe more to that FICO 2 high debt bit we were talking about previously though that's an interesting find, thank you for posting it. I didn't see an installment utilization factor with EQ FICO 5 with either auto loans or the small secured loans in my testing anyway.

Also I'd note I have the same reason codes now with my >80% mortgage as I did with the original share secured / CD loans I was testing with, and the reason code didn't go away until <10% (some rounding factor in there to be sure) but there was clearly an upper bound shift it was just north of 55%.

Anyway hoping I can track that upper tier shift for sanity's sake as I walk my mortgage across several potential breakpoints over the next few years, but unfortunately my file isn't as good for testing on anymore as the scores aren't fixed like they were with my tax lien.