Comparing FICO with other scores

[Roarmeister]

Since yo'all is American and I am Canadian and we use different scoring systems and our economic situations are different I've been thinking about how to best compare the scores so they relate to each other to some degree.

FICO's score range is 300 to 850 with a median of 723 and average of 664 according to one website.  http://www.credit-report-101.com/average-fico-score.html

Canadian Equifax range is 300 to 900 with a 25 percentile of 699, 50 percentile of 759, and 75 percentile of 808 -- estimate from my own credit reports.  I also use another Canadian scoring system provided by my bank; it has a 399-862 range with a 25 percentile of 680, 50 percentile of 750 and 75 percentile of 800 (estimates).

Therefore, I have worked up a table based on information I have located on a number of sites plus my own credit reports.  My own EQ score is 807, which puts me in the 1% probability of defaulting range.  That is just about as good as it gets but I figure I can get it a bit higher; 4.5 years ago, it was only 732.

The best I can figure from these table is that a Canadian score is VERY roughly about 50 points higher for the equivalent FICO score for percentiles which I am going to assume is because of different scoring systems and because the Canadian range is 50 points higher.  The Canadian economy is also in better shape so there may be a bit of a difference from that as well.

The most glaring difference is in the delinquency rates, especially from scores 649 and down.  For the moment, I am going to attribute part of that to difference in the housing markets (house prices are rising in Canada while they may not have even bottomed out in the US yet; and the difference in unemployment rates 7.2% vs. 8.2%

I thought these tables would come in handy in case someone out there had information on Vantage or other systems scoring and wanted to cross compare.

Forgive the formatting issues.

FICO                                                                                                      Canadian Equifax

Score                     %            Percentile           Delinquency       Score                     %            Percentile           Delinquency Rate

up to 499             2              2              83%                                        300 to 549            4              4              55%

500 to 549            5              7              70%                                        550 to 599            4              8              33%

550 to 599            8              15           51%                                        600 to 649            6              14           21%

600 to 649            12           27           31%                                        650 to 699            11           25           11%

650 to 699            15           42           15%                                        700 to 749            19           44           5%

700 to 749            18           60           5%                                          750 to 799            27           71           2%

750 to 799            27           87           2%                                          800 to 849            24           95           1%

800 plus                13           100         1%                                          850 to 900            5              100         1%

[RobertEG]

A major problem in attempting any such analysis is that even FICO scores are not "A" FICO score.  Their is no single FICO scoring algorithm to act as a uniform basis for any comparison.

Each CRA licenses its own flavor, as do different industry segments, such as banks, credit cards, auto, mortgage, insurance, etc. etc.  Even within industry segments, they have a menu of FICO versions to choose from, and many still use older versions to avoid the cost of update/retraining, etc.

With no standard, baseline scoring to make comparisons, and inability of consumers to even get industry-specific scores, it is an almost impossible undertaking.  The additional task of comparing FICO scores to non-FICO scores is impossible, as their innards are totally unknown.

But interesting in general terms.

[Roarmeister]

Impossible you say?  Well perhaps, but you should be able to get reasonably close once you cross compare the percentiles and delinqency rates.  And by reasonably close, I mean close enough to matter so that you can predict.  There will never be an exact match. But if you knew the above information between a FICO and a Vantage score you should be able to predict a score +/- 10 pts, all other things being equal.  The protential problems may occur because different rating agencies may have different information being reported to them and there is also the possibility of errors in one but not the other.

[Mailak]

---

@Roarmeister wrote:

Impossible you say?  Well perhaps, but you should be able to get reasonably close once you cross compare the percentiles and delinqency rates.  And by reasonably close, I mean close enough to matter so that you can predict.  There will never be an exact match. But if you knew the above information between a FICO and a Vantage score you should be able to predict a score +/- 10 pts, all other things being equal.  The protential problems may occur because different rating agencies may have different information being reported to them and there is also the possibility of errors in one but not the other.

---

I was just mentioning this in another thread. FICO + (disturbance) = Vantage/Plus. Given enough FICO/Vantage or FICO/PLUS pairs, we can see if there's a pattern to the disturbance, all else equal. Taken further, we can try to decompose the disturbance into variables. 

That's truly what FAKO scores are - reengineered scores (although I wouldn't venture to guess how much reengineering it is - could be the disturbance model or could be total reformulation). I think one could try to statistically un-reengineer them  Won't approach 100% but, the more predictive power you get, the better!

[llecs]

I'll expand on RobertEG's post in a sec, but I'll throw in there that we all sometimes try to compare a FAKO with a FICO. I love analogies, but comparing any FAKO (e.g. Vantage, PLUS, CreditXpert, CE Score, Equifax Credit Score, TransRisk, and so on) with FICO is like comparing apples and oranges. It's like counting and finding 7 seeds in your apple and assuming your orange would have just as many. It's impossible to do.

You can compare if both FICO and FAKO score similarly I suppose, but they do not. FAKO scores count stuff that FICO does not, and vice-versa. You can have a FICO increase and a FAKO decrease for the exact same event. Inquiries for example are counted all 2 years vs. FICO's one. At one point AU accounts were totally ignored by Vantage, whereas FICO fully scored them. Mix of credit is treated differently. With most FAKOs (PLUS included), a student loan is scored differently from a personal loan from an auto loan. In FICO speak, there are considered one the same. For some FAKOs (TransRisk included), collections are fully scored into the length of history and removing an old CA can drop your FAKO score. FICO ignores a CA's age. I can go on and on with the differences, but the most stark is scoring buckets.

FICO is the only scoring system out there that I know of that compares EVERYONE's FICO to your own. Your FICO is dependent on everyone else's. When the economy tanks, baddies aren't scored as harshly because others are being lumped into that derog bucket and the impact of a new baddie lessens. In other words, in a bad economy a new baddie won't hurt as bad. In a great economy (e.g. increasing FICOs), then the impact is more severe.

On a personal standpoint, I've done my own comparison for entertainment purposes. In regards to FICO vs. Vantage (from CreditKarma), my FICO is mostly higher than Vantage despite the 501-990 range vs. FICO's 300-850. On avg. it is usually higher by 20 or so points but my FICO has been as high as 50 or so and even lower than Vantage by about the same. VantageScore doesn't like me I guess. I can guess by a +/-50 range, or a 100 point swing. Not too accurate. In regards to PLUS (from USAA), my FICO is usually higher by 40-50 but there are times that my PLUS is higher by 10 or so. That's a 50-60 point swing. My favorite is my CC-enhanced TransRisk (from Barclay's SeaMiles updated regularly). My TU FICO is always higher than the TU-based FAKO.  Sometimes my FICO is only higher by 50. I've recorded it as high as 225 points. That's at least a 175 point swing. In short, if I see a FAKO, I can guess within 100 points where my FICO might be at, usually.

On to REG's point, there are other FICO versions out there. If you have a CC company pull your FICO, a mortgage lender pull your FICO, and an auto dealer pull your FICO all within the same day from the same CRA, you'll likely have 3 different FICO scores. There are different versions out there. TO make it more complicated, there are different version years, kinda like model years on a car. The newest Classic versions ignore CA's under $100, as an example. Inquiries are score differently. There are 2 more scoring buckets in the FICO08 versions. Lates are treated differently. These come to mind first. So, if you did accurately predict within a +/-10 point change on your FICO from a FAKO, that same +/- change won't occur on a different version, as an example.

I definitely encourage you to plot it though. I did the same. In the process you'll learn how FICO treats stuff differently than the FAKO you are using. I learned quite a bit and got a better understanding of FICO scoring and credit scoring in general.

[marty56]

I just got back from Vegas and it couured to me the odds of  generating your FICO score from a FAKO one was about the same as Marty leaving Vegas with 100k in slot winings.

In case you were wondering, I left Vegas + $60 from slots and like trying to compair FAKO to FICO, that was a fluke.