Speculation Why FICO Score Increases In Chunks

[Anonymous]

I have learned in life that, when a business does something that seems counterintuitive, there is a generally deliberation behind it, and 99% of time that deliberation leads to their making more money.

I have already written about why, when a consumer clicks to check the free FICO score at his her bank account, the click in the web  browser causes a sequence of events, behind the scenes, where the Fair Isaac Corporation is notified of each click: In a nutshell, Fair Isaac Corporation is able to guage the desperation of the consumer to know their FICO score. Dependening on the level of desperation, FICO might hold the update of the score for a particular month a bit longer than "usual". In my case 3 months ago, the "longer than usual" turned out to be 13 days. The goal is to induce the consumer to call FICO and ask:

Why is my FICO score taking so long to change this month?

At which point they will be informed by the CSR that they may purchase the score right there on the spot.

On a different topic, it just occurred to me why FICO might bump scores in chunks, or as we scientists/engineers say, "the score trajectory is not continuously differentiable". Many people have come to the forums asking why their FICO remains static for a long period of time. Then, suddenly it bumps, but not by two or three points. Sometimes it is 20+ points, without any significant change in profile. Some of you might say that the reason is age, but that does not make sense. Consider the following analysis:

Let's say that you are going to get a loan of some kind. The best rate will be had at 760. You think about applying on Tuesday, but get caught-up in traffic on way to bank and miss 5PM closing time. Your score is 740 on that Tuesday. You decide to try again the next day, Wednesday, and are surprised to find that your score is now 765. The only thing that changed over night is that your report is one day older. This seems counter-intuitive. After all, if the purpose of a score is for lenders to guage risk in servicing you, certainly you are not 25 points less risky on Wednesday than you were on Tuesday. If anything,  you might be a point or 2 less risky: The plot of your credit score should be "continuously differentiable", in technical terms, "smooth" in layman's terms.

Now the question becomes: Why does FICO boost in chunks?

I used to think that, perhaps, they wanted avoid revealing a smooth trajectory of a consumer's score because scientists would be able to see a correlation between lapsed-time and score-change, and therefore, figure-out the secret sauce of time-lapse on score. But now I think it is this:

If FICO were to boost score in accordance with smooth progression of time/smooth increase in score, it would eliminate uncertainty that FICO wants to be in the consumers mind. The consumer would look at their trajectory and think:

Hmmm...no baddies...looks like I am going up 3 points per month on aging. I'm at 742 now. I'll be at 760 in 6 months.

FICO does not want the consumer to realize that. They'd rather sandbag the score at 742. Leave it there for 6 months. Then, suddenly boost it to 760. Why? Because if the consumer is waiting for that 760 for some specific reason (mortgage), and at month 5, score is still stuck at 742, they might become midly frustrated at constantly checking and seeing no change, and buy score-monitoring. With a smooth trajectory, however, the consumer has a degree of certainty, and they have no reason to check because they can clearly see when it is going to reach 760. 

Yes, it's a somewhat lame speculation...but not having a smooth trajectory does not make sense! 

What do you guys think?

[Anonymous]

---

@AnonymousIn a nutshell, Fair Isaac Corporation is able to guage the desperation of the consumer to know their FICO score. Dependening on the level of desperation, FICO might hold the update of the score for a particular month a bit longer than "usual".

Your score is 740 on that Tuesday. You decide to try again the next day, Wednesday, and are surprised to find that your score is now 765. The only thing that changed over night is that your report is one day older.

---

I think your understanding here is a bit off on how scoring works.  FICO doesn't "hold the update" at all.  The algorithm yields a score based on the report data used at that time.  Same data, same score.  Different data, possibly different score.  FICO has no control over the input data. 

In your illustration above, the only thing that changed over night was not the report being 1 day older.  It's impossible for a score to change 25 points based on the aging of 1 day (really 1 month in the eyes of the algorithm if we're talking a move to the 1st of a month).  The only time you could come close to this would be if your AoYA was 11 months and crossed to 12 months over night, good for 15-20 points perhaps, then your AAoA also crossing a threshold at the exact same time, perhaps good for another 5 points.  Very unlikely that this is the case... more likely is that something else in terms of the information present on the CR changed from one day to the next... perhaps a negative item falling off, multiple inquiries from a big spree 365 days ago becoming unscoreable, a utilization paydown across threshold(s) happened, etc.

[Anonymous]

---

@Anonymous wrote:

---

@AnonymousIn a nutshell, Fair Isaac Corporation is able to guage the desperation of the consumer to know their FICO score. Dependening on the level of desperation, FICO might hold the update of the score for a particular month a bit longer than "usual".

Your score is 740 on that Tuesday. You decide to try again the next day, Wednesday, and are surprised to find that your score is now 765. The only thing that changed over night is that your report is one day older.

---

I think your understanding here is a bit off on how scoring works.  FICO doesn't "hold the update" at all.  The algorithm yields a score based on the report data used at that time.  Same data, same score.  Different data, possibly different score.  FICO has no control over the input data. 

Let me be clearer by what I mean by "holding the update". If anyone of you have cards with Chase, Capital One, or Discover; and you call the to ask why your "Free" FICO score is not updating on (roughly) the same day of the month as "usual", they will be quick to tell you that they have no control over the FICO page on their own website. Chase, especially, kept saying this to me, on the phone, over and over as if they'd been indoctrinated to say that each time a customer asked about any weirdness with their "Free" FICO score. After thinking about how adamant the Chase CSR's were in stating how much "they had no control" , I thought that was a little weird, so I dug further (I'm an engineer) and discovered why they kept saying that: It's true, and likely, their managers told them to say that to customers. What they mean, which they, themselves, do not fully understand, is that, when a customer clicks on the "Free" score link, score information is pull from FICO each and every time. I explained in my OP why FICO might conract with banks to set it up this way, instead of just giving the banks' computers information once per month. This detail is exceptionally important if my hypothesis regarding "Free" credit scores is correct.

Now about the "holding the update". Yes, it is true that FICO computes score based upon the data, and that never changes. And it is true that, if the profile data remains the same from one day to the next, the score will remain the same from one day to the next. But what I mean to say was...the banks do not control when the bank receives the "Free" FICO score from FICO. FICO controls that. Indeed, this detail is exceptionally important from a business perspective, if you are an executive at FICO, as mentioned earlier. That was ths subject of another thread. 

In this thread, however, I am talking about a non-linear bump in score which is related to what you wrote. I can explain with a question:

Let's say that the year is 2020, and you have been tasked with creating a scoring algorithm. Which of the following two algorithms is better for your customer (banks, auto-loan shops, etc.):

1. The score can jump as much as 30 points if consumer is about to cross an AAoA threshold.
2. The score cannot jump as much as 30 points if consumer is about to cross an AAoA threshold, instead, gradually increasing over a period of months from its before-jump value to its after-jump value.

I argue that #2 is better from a risk-management perspective. A person whose score is 742 on a Tuesday night, does not suddlenly become 28 points less risky the next day if score bumps to 770 on Wednesday. If the person has a 770 on the Wednesday, it is technically more accurate that the "risk" was, say, 768 on Tuesday, 766 on Monday, etc. Since the algorithm chosen by FICO is just that, chosen, they could have just as easily made the scoring "ramp up" smoothly. My argument is that FICO actually gets financial benefit from deliberately making the score not smooth, but "jumpy".

[dragontears]

Very interesting theory and an entertaining read (it does make feel like I need a tin hat lol).
The flaw in the theory is that you assume that with a stable profile that points are gained in "chunks". In my experience this is not the case. My score has a steady, gradual increases over the last several months of 2-4 per month. In fact, I have never had my score go up >20 points in 1 day that wasn't correlated to paying down utilization or the removal of a baddy.

[Anonymous]

That is the question isn't it:
Is it possible for FICO score to jump a significant number of points soley because of AAoA stuff, or does increase occur gradually?

[Anonymous]

The only time FICO scores increase significantly (say, 15-20 points) is from a major event... not day to day normal stuff.  In terms of Age of Accounts factors, in my previous post I mentioned the only one (AoYA crossing 12 months) that typically results in scoring impact that large.  Other events that cause a score to shift in a "chunk" [15-20 points or more] are generally related to the two largest slices of the FICO pie, Payment History and Amounts Owed.  If a significant change happens in either of those sectors, certainly a more dramatic score change can happen... and in that case I would say yes, there would be a significant (relative to the score change) difference in the risk level of that profile from 1 day to the next.  For example, someone goes to sleep on a Monday night with their score of X and they wake up on Tuesday to find their final 90 day late payment fell off their report and their score is 75 points higher.  Yes, they are significantly less risky on Tuesday due to a major negative item no longer being present on their CR.

[Anonymous]

Indeed, you pointed-out a very interesting fact: AAoA can result in significant jump. From a technical perspective, this does not make any sense. A person whose AAoA is 11 months, 30 days, 23 hours, 59 minutes, and 59 seconds, is not, say 25 points more risky than someone whose AAoA is 12 months. In that regard, instead of letting score jump this way, it makes more sense to let score increase gradually, according with AAoA, in a linear way. It can be proven that the gradual method would be more appealing to an actuary, because, it reveals a truer risk-profile of the borrower at any given point in time than the jump method. [Baddies, OTOH, can be considered a different matter].

[Anonymous]

---

@Anonymous

But what I mean to say was...the banks do not control when the bank receives the "Free" FICO score from FICO. FICO controls that. 

---

I'm not sure how you figure this.  A free FICO score is pulled by the lender usually at or around the same time every cycle.  I've had Discover for several years and can only think of once or twice that my score didn't update on the 4th or 5th of the month.  It's very consistent.  CO offers VS (not FICO) so they're sort of irrelevant and depending on your Chase product (I know Slate offers FICO) most offer the VS flavor as well.  If a lender decides to wait a period of days or longer to display your new FICO score, that's on them, not FICO.  Any time you're provided with a FICO score, there will be a date attached/associated with that pull.  Citi for example is notoriously slow in providing the customer with their update.  Citi may pull your score on the 25th of a month (for example) and then not provided it on their web site until the 12th or so of the following month.  That has nothing at all to do with FICO "controlling" when the bank gets the score, as the score is time-stamped with the date of the pull.  Whether or not the lender updates that [monthly] score on their web site promptly or waits a bit is up to them, but your score is still your score at the moment that it's pulled.

[Anonymous]

---

@Anonymous

Indeed, you pointed-out a very interesting fact: AAoA can result in significant jump.

---

Not AAoA; AoYA.  I've never seen an AAoA threshold crossing result in a significant score gain... mostly 4-6 points, which I would classify as "gradual" over time.  AoYA on the other hand tells a much different story.  It's been proven that those that haven't opened an account within the last year are less risky than those that have.  It's that simple.  You can argue if you want of those 15 points or so should come back more gradually, but I don't see the point.  If people at < 12 months AoYA as a group are more risky than people at > 12 months AoYA, I see no problem with a 15 point or so bump at that point in time crossing.

[Anonymous]

---

@Anonymous

---

  Any time you're provided with a FICO score, there will be a date attached/associated with that pull.  Citi for example is notoriously slow in providing the customer with their update.  Citi may pull your score on the 25th of a month (for example) and then not provided it on their web site until the 12th or so of the following month.  That has nothing at all to do with FICO "controlling" when the bank gets the score, as the score is time-stamped with the date of the pull.  Whether or not the lender updates that [monthly] score on their web site promptly or waits a bit is up to them, but your score is still your score at the moment that it's pulled.

The presumption here is that it is the bank that is the "culprit" for delaying the update. Example: Let's say you are with Chase. Your new score is normaly visible around the 5th of the month. One month, check on the 5th, and it is not there. Then you check on the 6th, not there. 7th, nope. 8th, nope, 9th, nope. Finally, on 10th of month, it says, "FICO score as of 3rd of month is 700". What I am saying might be happening:

The bank actually does not have your new score in their computers at all. What's happening is that, when user clics on the VIEW SCORE button, the computers at Chase go pull a record of score from FICO. So every time user clicks, info is pulled from FICO (not credit bureau). Then, the "culprit" would be FICO: They grab your credit file from EQ/EX/TU on the 3rd. They compute your score on the 3rd. And then they sit. You come to Chase on the 5th, thinking it will probably be updated, and every time you click, it has not updated. You try 6th, 7th, etc. At some point in all of this FICO, says: OK, we've made him wait long enough.... And FICO finally provides the update that FICO has been sitting on to Chase.

I decided about 3 months ago to check this hyothesis. If FICO is doing what I think they are doing, then clicking on the VIEW SCORE button over and over, every day, starting with the "usual" day, should result in FICO witholding score. So that's what I did. I went to my Amex account, and banged on the VIEW SCORE button over and over, and sure enough, the gap between my "usual" update day and the day that it was finally shown to me on my Amex account was 15 days!