Comments on: How Is Your Credit Score REALLY Calculated? http://www.myinvestingblog.com/how-is-your-credit-score-really-calculated/ Teach Your Money To Think! Fri, 10 Feb 2012 18:37:00 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Edwin http://www.myinvestingblog.com/how-is-your-credit-score-really-calculated/#comment-1134 Edwin Wed, 16 Apr 2008 11:13:27 +0000 http://myinvestingblog.com/2008/04/14/how-is-your-credit-score-really-calculated/#comment-1134 --- Andrew said: the answer is because it has been shown objectively that, on average, people with more cards are more likely to default. Period. --- Wrong, actually. You must have multiple credit cards to get the best score you can. It seems that FICO likes for you to have less than half of your credit cards reporting a balance, and half or less of all of your accounts showing a balance, so if you have a mortgage and a car loan, and you want to allow 2 cards to report a balance because you use them and can't pay them off and have them report 0 at statement time, you are going to need at least 6 credit cards. If you have 5, your score will be lower, all else being the same. —
Andrew said:
the answer is because it has been shown objectively that, on average, people with more cards are more likely to default. Period.
—
Wrong, actually. You must have multiple credit cards to get the best score you can. It seems that FICO likes for you to have less than half of your credit cards reporting a balance, and half or less of all of your accounts showing a balance, so if you have a mortgage and a car loan, and you want to allow 2 cards to report a balance because you use them and can’t pay them off and have them report 0 at statement time, you are going to need at least 6 credit cards. If you have 5, your score will be lower, all else being the same.

]]> By: hank http://www.myinvestingblog.com/how-is-your-credit-score-really-calculated/#comment-1137 hank Wed, 16 Apr 2008 04:16:26 +0000 http://myinvestingblog.com/2008/04/14/how-is-your-credit-score-really-calculated/#comment-1137 Ah - Andrew thanks for the insight; it is very informative and interested to hear what a person in the field of FICO has to say about it. Thank you VERY much for stopping by Andrew and feel free to stick around! Your input is indeed helpful! Ah – Andrew thanks for the insight; it is very informative and interested to hear what a person in the field of FICO has to say about it. Thank you VERY much for stopping by Andrew and feel free to stick around! Your input is indeed helpful!

]]> By: Andrew http://www.myinvestingblog.com/how-is-your-credit-score-really-calculated/#comment-1135 Andrew Wed, 16 Apr 2008 04:08:47 +0000 http://myinvestingblog.com/2008/04/14/how-is-your-credit-score-really-calculated/#comment-1135 Guys, your description of how this works isn't quite accurate. What is really going on is that Fair Isaacs (and others) are building predictive models on large sets of historical data. Statistically, what is happening is that each of these factors are being measured, correlated with an outcome variable that the user would like to predict, and weighted, usually using a technique known as logistic regression, to develop a predictive score of the likelihood that a given debtor will default in a year. It's akin to developing a model of how often 7's will roll up when you roll a pair of dice. You can't predict whether the next roll will be a 7, but you can estimate the probability of the event. The formula is somewhat secret because capturing the data, doing the analysis and building the models is expensive. FICO has literally hundreds of people developing and updating these models. If you share the formula, then any Joe Schmoe can go and calculate scores, and that's what Fair Isaacs wants to sell. Why should they give up their hard work? No-one is going to like to hear this, but the FICO score is used because it is well known in the industry that FICO knows their stuff, and that the score works for it's intended purpose. To the previous poster, asking why it matters if he has two cards rather than one with twice the balance, the answer is because it has been shown objectively that, on average, people with more cards are more likely to default. Period. The score doesn't predict your behavior, it predicts average behavior, which is what a lender cares about. Maybe you are the most responsible debtor on the planet, but if you look like a defaulter, that's what the lendor has to go on. And defaulters have a lot of common characteristics, such as a lot of accounts, missed payments, high available credit relative to income, etc. Disclosure: I work for a competitor to FICO, doing predictive modeling on these types of problems. Guys, your description of how this works isn’t quite accurate. What is really going on is that Fair Isaacs (and others) are building predictive models on large sets of historical data. Statistically, what is happening is that each of these factors are being measured, correlated with an outcome variable that the user would like to predict, and weighted, usually using a technique known as logistic regression, to develop a predictive score of the likelihood that a given debtor will default in a year. It’s akin to developing a model of how often 7′s will roll up when you roll a pair of dice. You can’t predict whether the next roll will be a 7, but you can estimate the probability of the event.

The formula is somewhat secret because capturing the data, doing the analysis and building the models is expensive. FICO has literally hundreds of people developing and updating these models. If you share the formula, then any Joe Schmoe can go and calculate scores, and that’s what Fair Isaacs wants to sell. Why should they give up their hard work?

No-one is going to like to hear this, but the FICO score is used because it is well known in the industry that FICO knows their stuff, and that the score works for it’s intended purpose.

To the previous poster, asking why it matters if he has two cards rather than one with twice the balance, the answer is because it has been shown objectively that, on average, people with more cards are more likely to default. Period. The score doesn’t predict your behavior, it predicts average behavior, which is what a lender cares about. Maybe you are the most responsible debtor on the planet, but if you look like a defaulter, that’s what the lendor has to go on. And defaulters have a lot of common characteristics, such as a lot of accounts, missed payments, high available credit relative to income, etc.

Disclosure: I work for a competitor to FICO, doing predictive modeling on these types of problems.

]]> By: Jesse http://www.myinvestingblog.com/how-is-your-credit-score-really-calculated/#comment-1136 Jesse Tue, 15 Apr 2008 18:59:09 +0000 http://myinvestingblog.com/2008/04/14/how-is-your-credit-score-really-calculated/#comment-1136 One thing that bugs me is the whole "keeping the formula a secret" thing. Should having 1k on one card and 1k on another get a better score than 2k on one card and 0 on another? Absolutely not. Should not carrying any balance hurt you? Absolutely not...but those things do matter. <em>Jesse's last blog post..<a href='http://feeds.feedburner.com/~r/thepennysaved/~3/267776363/'>Starbucks Vs Gatorade Vs Gas: wow</a></em> One thing that bugs me is the whole “keeping the formula a secret” thing. Should having 1k on one card and 1k on another get a better score than 2k on one card and 0 on another? Absolutely not. Should not carrying any balance hurt you? Absolutely not…but those things do matter.

Jesse’s last blog post..Starbucks Vs Gatorade Vs Gas: wow

]]>