Ahh... okay. That makes sense. I like it better then 
OP, be glad your scores went the way they did! I applied for an auto loan and my CU pulled a score typically used for mortgages. It turned out my score on that report was 40+ points lower than the one on this site. So my surprise was a bad one, while yours was at least a good one! 
I've had MF's Scorewatch for years and about a year ago EQ FICO scores transitioned from EQ 4 to EQ 8 since that is what most credit issuers use. I didn't see any real change. The newer score was less sensitive to the accounts with balances while the older one ignored Amex charge card balances. There was some fluctuation due to these difference and probably others.
Mortgages are probably the most important type of credit most of us will ever use but credit cards are the most numerous use of credit scores.
Mortgages tend to stick with older credit scores because they have very long time horizons and the newer scores, while more predictive for credit cards, take more time to establish validity for mortgage purposes.
Also, each issuer of credit has their own portfolio based on earlier credit scores so even when a new score comes out it can take several years before a CC issuer will feel comfortable switching. Credit scores are a market. There is continuous improvment in these products and lenders want the most predictive scores they can get consistent with validation on their own customer base and these vary bank to bank. There just isn't such a thing as a "perfect" credit score. These are all statistical estimates and the economy and behavior of consumers varies over time. Even between bank's overall customers.
Remember that the purpose of improving credit scores is to be able to lend to more people, not fewer. Since the large majority of people with older credit scores will repay future loans the goal of these score algorithm improvements is to sus out those folks so they can be approved and not declined. There are some with good older scores that may be at higher risk and suffer declines with new scores but these are fewer in number. Doing this allows FICO and other score providers to sell their scores for a reasonable profit and recoup the rather large investments needed for creating highly predictive algorithms.
