Are FICO scores competitive?

[Adidas]

I've been reading a lot here on the forums and elsewhere online trying to understand all the ins and outs of FICO scoring. I've learned a lot but one thing I'm unclear on is whether FICO scoring is competitive? What I mean is if everyone in the US started making 100% of on time payments would that raise the bar for what it takes to get an 800+?

I remember reading somewhere that the main reason scores vary from different bureaus is because each one has slightly different data on you (most commonly the # HPs) but that the algorithms for scoring also vary from each lender even if you are only looking at one specific type of FICO score like FICO 08. The idea was that FICO 08 is optimized based on the data within each database at each bureau so that a FICO 08 based on your data from TU might be different than your FICO 08 from EX even if your personal data is identical from each one. This would most easily be tested in someone without any HPs for 2 years where all 3 bureaus have everything correct from that person's history. I also remember reading somewhere that the scoring might have adjusted during the '08 financial crisis as more people were late/defaulted on loans. The idea was that a default around that time might have affected your score less because it wasn't only happening to you.

Can anybody shed some light on this? I ask because the new FICO 09 algortihms are supposed to be more forgiving of medical debt and collections that are PIF. I know medical debt is one of the largest reasons for BK in the US so if these algorithms are competitive does that mean some people who don't have medical debt might have their scores lowered to compensate for the people whose scores will go up? The idea would be that under the old FICO 08 you were getting a boost by not having medical debt pulling you down but in FICO 09 you wouldn't have that boost because the algorithm doesn't care about medical debt.

[Anonymous]

Your own individual scoring performance is based in part on competing with other people.  That's because any given FICO model (e.g. FICO 8 Classic) will group consumers into what are called scorecards.  (You will hear people around these Forums often call them "buckets.")  A score card is a group of people that FICO considers somewhat similar.  I'll give you some examples.

In FICO 8 classic, there are 12 scorecards total.  Four of these are for people with substantial derogs.  The remaining eight are for people with no derogs or perhaps only a couple Day 30 lates.  So already you are seeing how FICO is putting people into similar groups: people with derogs vs. people with clean profiles.

For the clean profiles there are eight scorecards.  Three factors determine which scorecard you get placed into: Age of Oldest Account, Age of Youngest Account, and Total Number of Accounts.  (All three factors consider closed accounts as well as open.)

So one scorecard (at one end of the spectrum) would be for a person very new to credit: few accounts, and all of them under a year old.  At the other end of the spectrum might be your mom and dad: a dozen or more accounts, the oldest being 25 years old, and they haven't open a new one in at least 25 months.  Each of those scorecards has similar kinds of people (at least based on those three factors).

Once you are in a scorecard, you are scored (in part) against the other people in that group, much like your teacher in high school grading everybody on a curve.  At this point all the various scoring factors from FICO come into play: CC utilization, etc.

So yes, in this sense, scoring well on FICO is a kind of competition, just as it was in Mrs. Flynn's history class back in high school.

In that sense, you are right to view with skeptcism the universal joy people seem to have when FICO relaxes some previously harsh standard of theirs.  This is because There Ain't No Such Thing As A Free Lunch.

[iv]

---

@Anonymous wrote:

Once you are in a scorecard, you are scored (in part) against the other people in that group, much like your teacher in high school grading everybody on a curve. 

---

What? No. Not at all. (You know better, CGID!)

You are absolutely NOT "...scored against the other people..." - to answer the OP question, no, FICO scoring is not "competitive".

You ARE being scored against the people in the reference dataset used to develop that particular model (and yeah, that does mean that a new model like FICO 9 can "move the goalposts").

But you are NOT being directly scored against other people - the model is static.  No amount of behavior change in other people will affect your score in an existing model.

Behavior changes can (and do!) change scoring criteria in future models, though - so OP's concern about changed scoring criteria is somewhat valid.  But remember that the scoring is meant to be calibrated to a projected "rate of borrower default", not a standard bell curve.  If average borrower behavior "improves", then the curve can just move to the right...

[Anonymous]

That's interesting, iv.  My understanding (from a lot of folks here) was that within a scorecard you were being scored against the other people in that scorecard, at least in part.  It sounds like you are saying that (a) once a model is released it is fixed (no further changes to it are made, though the developers working on the next model might make changes) and (b) that the person within a scorecard is being scored against those people from the past that the developers used when they were creating it -- but not scored against people from this year.

As I say, very interesting!  (Assuming I am understanding you right.)

But also, as you go on to say, perhaps it doesn't matter, given our OP's concern.  Basically, he's saying that he used to stand out from the crowd because he had no derogs.  Now, in the brave new world of FICO 9, he'll stand out less, because all kinds of people will have no derogs.  With FICO 9, you can have a history of several things being turned over to collection agencies (which in the past scoring system was understandly flagged as a serious risk) and yet if you pay them off you will be considered just as good as our OP, who was diligent and never got anything turned over to collections.

Regardless, thanks for the feedback.  Much appreciated.

[Adidas]

Thanks for the feedback! Now that I think about it, it seems to make more sense to keep the credit score fixed to a probably of default since that's what the CCCs care about. That would also explain the new rules on PIF collections since they got their money in the end. I know there are figures online estimating the probability of default based on FICO but I always figured they are more estimates than the underlying logic of FICO. If that's the case then the goalposts wouldn't move even with new models if they are fixed to the % chance of default?

[Revelate]

---

@Adidas wrote:
Thanks for the feedback! Now that I think about it, it seems to make more sense to keep the credit score fixed to a probably of default since that's what the CCCs care about. That would also explain the new rules on PIF collections since they got their money in the end. I know there are figures online estimating the probability of default based on FICO but I always figured they are more estimates than the underlying logic of FICO. If that's the case then the goalposts wouldn't move even with new models if they are fixed to the % chance of default?

---

Even if you fix certain scores to be representative of liklihood to default, that which leads to default does change over time hence the changing of medical vs. regular collections, and paid vs. unpaid collections, and more weight on recent history, etc ad naseum.

That said the optimization path on virtually any algorithm (certainly true when we're talking FICO models) is the same, so just make the file as pretty as possible and receive a pretty score regardless  of specific algorithm chosen.

Or in my case put as much lipstick on the pig as possible .