I have 3 residential IP's in different towns (family home, my home, place of work), multiple proxies from different companies that come recommended from reputable people on here, VM with MAC spoofing, some of the more notable premium scripts, no more than 3-4 hours a day at maximum, 2 hours break for every 1 hour minimum during those times, aged accounts and fresh accounts and so forth and I cannot dodge the banhammer for more than 72 hours. The only accounts to survive are ones using scripts that I have written myself and I have a feeling I'd have the same kind of outcome if I were to purchase a private script from those same premium script writers, I can only go from my personal experience and this is it.
Of the same premium scripts that have gotten me banned for less than 4 hours botting total, 3 of them have had other people saying they have gotten banned after 1-3 hours of use as well within the past week.
It's not hard to imagine that the data collected is fed into some kind of neural network or the like and used for future detection which would mean more users over time will make it easier to detect.
As you can probably tell, when I said I hold mostly the same opinion, that didn't include the IP part. I'm also not saying the client or the programmers are bad, in fact I said the opposite but in the days of machine learning, I just don't see how they wouldn't utilise that information.