Interesting info Credit Algorithm

Discussion in 'Credit Talk' started by Calypso, Aug 25, 2003.

  1. Calypso

    Calypso Well-Known Member

    Had a fascinating talk with a Palo Alto computer whiz who explained a little bit about how the credit scoring algorithm works.

    Apparently it is a "learning system" that modifies the algorithm as it inputs consumer behavior.

    So it works a little bit like this (totally simplified, of course) They analyze the credit risk associated with different scenarios-- presence of a mortgage (and dollar amount), number of finance companies (if any) percentage of credit used, etc. Then everybody in the group that screws up gets dinged. (Like I said, this is really simplified).

    But it explains a lot about the credit scores I see and the fact that they often seem to be so random.

    Last year, people who carried over 10 cards may suddenly have had a higher rate of default and that is why my score suddenly dropped and the reason code about "too many cards" appeared, even though *my behavior* had not changed.

    It is really interesting. This guy seemed to know A LOT about the program. He was a stock guy, but knew people who worked on the FICO model.

    It's funny. I came to this site to learn as much as I could about credit scoring, and even though I have *succeeded* in reaching my goals, I am still drawn here, because I constantly learn new stuff. I'm so impressed by the power of joint effort here.

    I have learned so much here and hope that I can help someone else --

    Cheers-

    Calypso
     
  2. GEORGE

    GEORGE Well-Known Member

    HIGH F.I.C.O. 739...PAY OFF $40,000+ F.I.C.O. TAKES A DUMP UNDER 700!!!

    ONE WITH A BRAIN WOULD ASSUME THAT IT WOULD INCREASE ABOVE 739!!!

    CLOSE ONE ACCOUNT LOSE A FEW POINTS MORE...EVEN THOUGH IT IS NOW PAID IN FULL EVEN THOUGH IT "WAS" ALMOST "MAXED"

    YOU CAN'T WIN!!!

    It is like a car...you change the oil and filter every 3,000 so the motor will last 250,000+ but the door falls off...motor is purring like a KITTEN!!!
     
  3. catnap1972

    catnap1972 Well-Known Member

    All you need to remember is

    "YER GUN(NA) GET SCREWED"

    No need to remember algorithms or any other complex mathematical formuals!
     
  4. klingsor

    klingsor Well-Known Member

    That is very interesting. Still, I wonder how large those differences are -- i'm inclined to believe that the types of shifts you're talking about wouldn't account for more than a couple points in terms of a drop. for the effect to be any larger those would have to be massive default rates in various groups.
     
  5. Calypso

    Calypso Well-Known Member

    But it must be true to explain some of the things I have observed (I see A LOT of credit reports)

    For instance-- I have noticed that people with large mortgages (over 400 or 500K) can have several 30 day delinquencies and still have scores over 700. I have never seen scores over 700 with several 30s on reports without large mortgages. There must be data that people with large mortgages have a lower rate of default. Student loans also seem to give FICO scores a huge boost-- even when they are still deferred, with no payment record. We all know how important length of history is, and low ratios. High credit lines are also beneficial (proving just how tricky Cap1 is with their no reporting game)

    It is fascinating to observe. We only have so much power. We can pay down balances, carry a mix of credit, make our payments on time, but there are factors beyond our control.

    The whole concept of a "learning system" explains part of the mystery. The scoring constantly changes as the system processes results.
     
  6. lbrown59

    lbrown59 Well-Known Member

    << but there are factors beyond our control.
    Calypso
    =============
    More reasons it's a scam

    >>

     

Share This Page