Interesting observation and nice data you have provided. Me not being a guru of any kind, I think the explanation you have come up with appears to make some sense.
One big question I have is what is holding your scores back? 4 years AAOA pristine file should be 800+ on FICO 8 by most reports we've seen here; almost wonder if your bucket based on oldest account is so lofty that you look lame in comparison but I wouldn't think your bucket would go that low typically. Also your other models look way too similar to mine in some cases even discounting my stupid EQ FICO 9 scores. Interesting datapoints though.
Are there still balances on the closed accounts? If so I would suspect they count as they apparently do for revolving utilization; if not though, at least with my dirty scorecard (and it may change for those of you in clean ones) I've always seemed to drop based on current accounts on my data: 33% and 50% for EQ FICO 8 of open revolvers reporting a balance. If I get time I'll see about factoring in closed accounts too but I suspect it's going to come out with awkward numbers which doesn't seem to be the case for the rest of FICO scoring in general.
I guess the correct way to test this theory would be for another clean file to get good data on number of cards with balances and see if it corresponded with yours.
In general 20 point differential is "within design tolerances" were your scores more similar prior to carrying the balances? Also to add to that, your TU scores mostly moved upwards, though at least based on my data EQ doesn't do cards with balances on the same percentage as TU/EX at least for my scorecard.
@Revelate wrote:One big question I have is what is holding your scores back? 4 years AAOA pristine file should be 800+ on FICO 8 by most reports we've seen here; almost wonder if your bucket based on oldest account is so lofty that you look lame in comparison but I wouldn't think your bucket would go that low typically. Also your other models look way too similar to mine in some cases even discounting my stupid EQ FICO 9 scores. Interesting datapoints though.
Your guess is as good as mine. Probably better! 
I'd be happy to send you my last five 3B Reports (with only the personal info page removed) if you would be interested in wading through them yourself. They range from 24 to 60 pages though, so it's not a small project. Just PM me with your email address if you want 'em.
All are free of any derogatories-
11/17/14 - 766/753/767 (day my last derogatory aged off)
07/05/15 - 739/735/746
12/11/16 - 744/740/753
02/11/16 - 743/727/748
06/13/16 - 737/719/744
My guess is that it's a combination of new accounts, number of cards with a balance, and number of cards over 30%
A little history...
On 11/16/14 I still had multiple derogatories and scores all around 665
On 11/17/14 last derogs dropped and I jumped to 766/767/753 with 3 open installments, 5 closed installments, 3 open CCs and 1 closed CCs. Oldest acct: 16yrs 5mths, AAoA: 9 years, 31% revolving utility
Since then I've added one new auto loan and 21 new credit cards. So, my AAoA is really being held up by five 17+ year old student loans.
I did climb briefly into the 775-785 territory after my first 3-4 prime cards were added to my clean reports. Then the long slow decline began with additional new cards, the new car loan, and then (about 6 months ago) a series of large BTs to cover several back-to-back family emergencies.
I believe this new 3B report will be the low point for my scores (fingers crossed) as I'm just starting to get my feet back under me again to pay down balances and have zero cards I'm interested in apping for in the future. So balances should drop and "new card penalties" should just diminish for the next 6 to 12 months.
@Revelate wrote:Are there still balances on the closed accounts? If so I would suspect they count as they apparently do for revolving utilization; if not though, at least with my dirty scorecard (and it may change for those of you in clean ones) I've always seemed to drop based on current accounts on my data: 33% and 50% for EQ FICO 8 of open revolvers reporting a balance. If I get time I'll see about factoring in closed accounts too but I suspect it's going to come out with awkward numbers which doesn't seem to be the case for the rest of FICO scoring in general.
No, all closed accounts were fully paid before being closed.
Think your guess is actually on the money but I don't have time to tease it out unfortunately (really need to do the pre-work for my Chem lab starting in a couple of days, sigh). Your revolving utilization just doesn't sound that bad (30% on a couple of tradelines would be wtfever on my report, but admittedly our scorecards are not equivalent). If someone else does it I'd certainly be interested in seeing the results.
Comparison data since I find it interesting heh. Minimal balances on 5/13 revolvers, recent mortgage and two nearly paid off baby installment loans open; 8 scoreable inquiries on EX including 2 from a few days prior, 6 accounts within 1 year (none within six months, though that'll change shortly and be interesting to see) same 6 accounts within 2 years on this report since my 2014 app spree just aged out end of May (didn't apparently get any bump up from that though or it was so small as to be lost in the noise). Different scorecards (tax lien, 30/60 late, and a partridge in a pear tree from 2010) just I'm surprised given what you've posted that you wind up basically equivalent absolute.
| Score | Aahz 6/13 | Rev 6/12 |
| EX FICO 8 | 719 | 721 |
| EX FICO 8 AU | 705 | 720 |
| EX FICO 8 BC | 724 | 741 |
| EX FICO 98 | 725 | 723 |
| EX FICO 98 AU | 699 | 735 |
| EX FICO 98 BC | 692 | 722 |
| EX FICO 04 | 699 | 677 |
| EX FICO 9 | 712 | 722 |
| EX FICO 9 AU | 706 | 735 |
| EX FICO 9 BC | 708 | 720 |
