Algorithms

[Anonymous]

 

Please stop paying your bills on time so that my score will go up.  Sound crazy?  I hope so!  But, what an interesting idea (though perhaps misguided) to think that a change in your credit has an effect on mine.  I’ve given thought to this idea, and I’m sharing it with you guys and gals for review.

 

So, what is the basis for even thinking such a thing given that we all know that our credit scores are a direct reflection of what’s in (and only in) our credit files?  In other words, how could it possibly be that your not paying bills on time could ultimately affect my credit score?

 

If only one of you didn’t pay your bills, my score wouldn’t change, and even if a few thousand people were not paying their bills on time, that would not effect my score either, as that would be an insignificant number to effect change, but if enough people didn’t pay, then it would positively (yet slightly) affect the rest of us, or so is the implication of this crazy idea of mine.

 

If there’s one thing we know about the secret formula, it is that it’s complicated, and the more we learn and try to figure it out, the more we come to understand the degree to which it is dynamic.  It draws on many different elements that are in (or missing from) our credit files, and we know this, yet we still can’t successfully predict with accuracy what our scores will be—despite the fact some rather smart people have tried to figure it out, and that tells me that we’re missing something, and for what that might be, I go back to the basics. 

 

The goal is to assess risk, and one way to do that is to do so comparatively … compare credit files to other credit files.  Just as we can rank test scores in percentiles, so too can those that have access to credit files rank each element of people’s credit files to see how they compare to others overall.

 

For example (and for illustrative purposes only), if you have no late payments in your credit file, then you are not (metaphorically) at the bottom of the class (as far as that one element is concerned), and if you have more late payments than anyone, then you’re at the bottom of the class.  It’s a big class with millions upon millions of students (metaphorically), but we could still figure out who is in the top ten, the bottom ten, and all the tens in-between.  Could our individual scores be linked to the overall statistics that are based on the actions of others?

 

In analogy, consider the statistics of all credit files as ocean swells while thinking of individual credit files as waves.  The better the elements in your credit file, the higher your wave (or score), but the degree to which the height changes may also be affected by the underlying ocean swells.  For example, if you suddenly have two recent 30-day late payments, that looks bad and your score will drop, but if a massive number of people were to suddenly have three 60-day lates, you could be thrown into a higher percentile ‘bucket’ such that your score will not drop as dramatically, as how you are doing comparatively hasn’t worsened to the degree that it would have had no others had those 60-day lates.

 

Keep in mind, this is all pure speculation—more of a crazy idea really.  Brainstorming, if you will.  But, I wanted to share it.  Who knows, maybe one of you guys in the know can offer some thoughts.

[Anonymous]

I would almost agree that you are at least partially correct. Many people on here are familiar with "scoring buckets" where FICO groups you with others with similar credit file characteristics. The better the bucket the higher your potential score depending on how you rank with others in your same "bucket."

It is a very intriguing concept and like you said the more we figure out about FICO the more difficult it becomes.

But I like the way you put it and it makes sense.

[RobertEG]

Good food for thought.

The FICO algorithms, for which there are at least a dozen, are not updated instantaneously. Fair Isaac does not maintain consumer credit files independently of the CRAs, so must get periodic mass information from the CRAs to trim their algorithms.  An expensive process that is not done every day or every month.  So it would be a period of time before any major shifts in mass credit profiles would even be factored into the scoring algotithms.

Take your example of a major overall shift in consumer payment profile for the worse.

FICO might very well view this is increased risk in all of its algorithms, and thus it might also hurt the higher-end consumer scoring algorithms for those who have thick and clean credit files as well as those in the lean and dirty categories. 

For example, a massive increase in major derogs might cause the FICO algorithms at all levels to look more harshly on any new minor derog (30-day late), now recognizing its increased potential to become a major derog.  It might result in a shift in the weight given to other categories, such as length of credit history or credit mix.

FICO is not a system that gives you points.  You start with 850 max, and it chops off points for risk areas.  You cant just assume that if someone else’s score goes down, your score goes up commensurately.  Increased risk by others might also be viewed as increased risk for all.

It is not necessarily a zero sum game.  One going down does not necessarily result in another going up.

If overall consumer risk took such a serious turn for the worse, it would probably cause a tightening of credit by lendors, and harsher, more restrictive terms on new credit. Also, maybe an increased incentive for original creditors to do charge offs and collections at any earlier time.  The raw consumer credit market might suffer dramatically.

[Anonymous]

Thanks!  That was insightful.