Recently an old credit account that I've had sitting idle was closed due to inactivity. I get credit-score updates from the issuer (who still maintains the credit account that I actually use), and this event caused a small drop in my score (812 to 805, so -7 points) apparently because my total available credit decreased.
So we've got a drop in total credit from $25000 to $11000 (the fact that the bulk of my credit was with the card I never use came as a surprise to me, too), with a corresponding drop of 7 points to the credit score. That looks like (but probably isn't) 0.5 points per $1000 of available credit. And fair enough, I guess.
The part I'm wondering about is that the card issuer also provides a "simulator" tool that predicts what your credit score will do if you take various actions. I thought I could get my 7 points back by just increasing the limit on the card I still have up to $25000, but when I put that into the simulator it kicked the score down to an atrocious 780-something. Strangely, if I simulate adding a second card to bring my total credit up to $25000, instead of a plummeting score the simulator gives 5 of the 7 points back.
I guess the core question is, why does having $25000 in credit available across two accounts produce a significantly better score than having the same amount of credit in a single account? If someone is trying to max out their reported credit score, what's the optimal configuration of total credit and number of cards/accounts?