I would broadly endorse Sid's findings except for the last one.


  • The next star drop in gain I'm seeing was what I expected from reading other posts. No idea if there has been an overall nerf to speed because I have no access to historic data.


What surprised me was the wide range of values that were achieved for the same player doing the same type of training from one session to another - that highlights the need for a decent sample size.

Its risky to say this because the sample sizes are small but I've been recording the start and end condition for each training session (and each type) and so far I'm not seeing any influence of Condition Level on Gain. (But take that with a pinch of salt atm)

PS. At some point I'll take a further look at Gain per 1 point of condition and see if anything jumps out.