Hmmmm.... what?
confused? you will be.
You did you measure the rate of growth?
well there arent stats available for this and I dont have the time/inclination to sample the arena populations for several days worth of 24h periods. however considering the max total numbers playing, and assuming a roughly normal distribution for total player numbers during the main sessions, the max rate of joins would be at 300-400 total players for weekend evenings and a little lower for weekday evenings. this would be even lower for the lunchtime sessions. to work best, the min cap should also be dynamic, based on the profile for that particular session. given the data for unique login attempts it would be fairly easy to determine the best min cap for each session.
edit: you take my point though, that the min cap for a session should be set to the number where the join rate is highest?