Understanding FICO® Scoring

[Walt_K]

@nyancat, There are reasonable disagreements on the level of risk that is associated with the same data. The different scoring models weight the data differently. Even within FICO there are differences across scoring versions. People report 40+ point differences from their TU98 and TU04 scores. The data is the same, but the model has been updated to assign risk differently. Ideally, the more recent version is based on a more accurate model, but who knows. The different models assign different weights to different data. This is why you should not expect them to track the same.

It is no different than valuations of companies by analysts. Some analyst may think a company's stick is underpriced while another thinks it is a good buy. We're not surprised that they disagree because we understand that they are using different metrics. But people seem to expect a credit score to be an objectively verifiable metric rather than one analyst's prediction of risk that may or may not be correct.

[cleansheet]

+1 @Walt_K

There is a real push by FICO to make it seem that they have only a single "magic number" that everyone uses, and that all other scores are meaningless.  The reality is that even within FICO there are multiple versions, with differences between the 3, and then scores biased for auto loans etc.  I will be very interested to see how the FICO scores change once they update the formulae to account for all the foreclosures in recent times.