Meaningless Credit Scores - Why are they so different?

MattMattMattUK

So first of all I know credit scores are meaningless, the issues that a bankrupt, someone with £20k of unsecured debt or an 18yo can have a perfect score and someone who has never borrowed and has millions in the bank can have a low score, but I was wondering if anyone has an insight into the different methodologies that they use to calculate these gimmicks. Note that I this is a curiosity thing as I know that these scores do not matter. I keep a regular eye on my credit reports themselves as several years ago I was a victim of impersonation fraud, so I check for any searches etc. and all three of the reports themselves contain identical information. 

The reason I ask is that my scores are different between the three main agencies. Experian is 999/999, Credit Karma is 699/710 and ClearScore is 472/700. The Experian account has always been at maximum apart from a two month period when I took out a mortgage and moved house a few years ago and then it only dropped to 850. The Credit Karma account randomly bounces up and down between around 630 and 705, even when no changes occur on my credit report. The ClearScore rating moves between 470-500 on a monthly basis, again with no changes to my credit report. 

This leaves me wondering, is there any method to their madness?

zzyzx1221

Presumably, they have different takes on what will keep you coming back for another hit.  Seems like it's working

MattMattMattUK

zzyzx1221 said:

Presumably, they have different takes on what will keep you coming back for another hit.  Seems like it's working

I am not there for the scores though, I log in monthly to check the reports for any searches that are not mine as I was previously a victim if impersonation fraud. I know score itself is meaningless, the only reason I notice it is that it is on the front page when I log in, but I have an inherent curiosity and that leads me to ponder why they would seem to be using such different methodology. I have no interest in increasing or "improving" my scores, but I do have a curiosity as to why they are so different. 

Maybe some of them include an RNG?

RelievedSheff

No idea why they differ but mine differ wildly as well even though they all show the same information.

My Clearscore randomly generated number dropped by 300 points a couple of months ago. No idea why as nothing had changed other than credit balances had reduced. The other two didn't budge.

Still none the wiser why it dropped but it didn't come back up this month. No doubt it will jump back up again for no reason at some point.

I tend to just ignore the numbers and make sure that the reported information is correct instead.

zzyzx1221

MattMattMattUK said:

zzyzx1221 said:

Presumably, they have different takes on what will keep you coming back for another hit.  Seems like it's working

I am not there for the scores though, I log in monthly to check the reports for any searches that are not mine as I was previously a victim if impersonation fraud. I know score itself is meaningless, the only reason I notice it is that it is on the front page when I log in, but I have an inherent curiosity and that leads me to ponder why they would seem to be using such different methodology. I have no interest in increasing or "improving" my scores, but I do have a curiosity as to why they are so different. 

Maybe some of them include an RNG?

Well, the very fact that you're posting about this shows that they've successfully managed to make you keep thinking about your score long after you've left.

So whatever they're doing definitely is working.

[Deleted User]

I believe the MSE Credit Club one is bugged, I have been 999 since I signed up, the other 2 fluctuate (regardless of the fact the scores don't mean anything, the changes at least show something is going on, a fixed 999 for 2+ years suggests something is wrong).

Their scores are based on factors like available credit, credit repayment history, total debt, stability at an address, credit utilisation etc

daringlion

Scorecard developer here. I'll try to explain the main reason for the large differences in scores but basically it's a scaling thing. Behind each score is a probability between 0% & 100% of some event happening. Maybe it's the probability of default, or it could be something else. This probability is then converted into a score using a particular mathematical formula, but a developer can input their own "baseline" score into this which impacts of the range of scores that come out. Each company, bank or bureau, will usually have a baseline score they commonly use, there is no industry standard, so this gives rise to different scores.

To explain why scores can change each month even when "nothing happens". So, let's say for example that "age of most recently opened account" happens to be an input into the score. This is now 1 month older, so that may result in a different score. That's one example, but there could be several things like that impacting a score.

[Deleted User]

The scores are all rubbish.

The payment histories, electoral register, public registers are handy but the score is there to always try and “sell” you some form of credit to “build up your score” etc

I only ever pay attention to factual issues, and I have picked up the odd mistake over the years and contacted companies to correct them. 

zzyzx1221

daringlion said:

Scorecard developer here. I'll try to explain the main reason for the large differences in scores but basically it's a scaling thing. Behind each score is a probability between 0% & 100% of some event happening. Maybe it's the probability of default, or it could be something else. This probability is then converted into a score using a particular mathematical formula, but a developer can input their own "baseline" score into this which impacts of the range of scores that come out. Each company, bank or bureau, will usually have a baseline score they commonly use, there is no industry standard, so this gives rise to different scores.

To explain why scores can change each month even when "nothing happens". So, let's say for example that "age of most recently opened account" happens to be an input into the score. This is now 1 month older, so that may result in a different score. That's one example, but there could be several things like that impacting a score.

I think the large difference between the scores is because they're making it up as they go along and have all decided on entirely different weighting for different attributes, which bear no resemblance to any actual lenders weightings and are therefore far too generic to offer any real value.

[Deleted User]

zzyzx1221 said:

daringlion said:

Scorecard developer here. I'll try to explain the main reason for the large differences in scores but basically it's a scaling thing. Behind each score is a probability between 0% & 100% of some event happening. Maybe it's the probability of default, or it could be something else. This probability is then converted into a score using a particular mathematical formula, but a developer can input their own "baseline" score into this which impacts of the range of scores that come out. Each company, bank or bureau, will usually have a baseline score they commonly use, there is no industry standard, so this gives rise to different scores.

To explain why scores can change each month even when "nothing happens". So, let's say for example that "age of most recently opened account" happens to be an input into the score. This is now 1 month older, so that may result in a different score. That's one example, but there could be several things like that impacting a score.

I think the large difference between the scores is because they're making it up as they go along and have all decided on entirely different weighting for different attributes, which bear no resemblance to any actual lenders weightings and are therefore far too generic to offer any real value.

CRA scores yes though the basis of the principle is fine. Of course as mentioned, as the lender has their own scoring system based on their chosen factors / weightings the CRA score is rendered meaningless (not least as lender sees salary)

zzyzx1221

Deleted_User said:

zzyzx1221 said:

daringlion said:

Scorecard developer here. I'll try to explain the main reason for the large differences in scores but basically it's a scaling thing. Behind each score is a probability between 0% & 100% of some event happening. Maybe it's the probability of default, or it could be something else. This probability is then converted into a score using a particular mathematical formula, but a developer can input their own "baseline" score into this which impacts of the range of scores that come out. Each company, bank or bureau, will usually have a baseline score they commonly use, there is no industry standard, so this gives rise to different scores.

To explain why scores can change each month even when "nothing happens". So, let's say for example that "age of most recently opened account" happens to be an input into the score. This is now 1 month older, so that may result in a different score. That's one example, but there could be several things like that impacting a score.

I think the large difference between the scores is because they're making it up as they go along and have all decided on entirely different weighting for different attributes, which bear no resemblance to any actual lenders weightings and are therefore far too generic to offer any real value.

CRA scores yes though the basis of the principle is fine. Of course as mentioned, as the lender has their own scoring system based on their chosen factors / weightings the CRA score is rendered meaningless (not least as lender sees salary)

Well of course, that's what we're talking about after all (CRA scores.)