5 hours ago, wwwat said:

I wrote a ~2k line script, was undetected for around a week or two, running for 5+ hrs a day, but after the first ban the other bans came in after using the script for like 2 days (stealth injection, didn't use proxies to be fair). I don't think writing your own script makes much of a difference anymore (I'm sure it used to years back).

Proxies don’t matter unless you’re running a farm, even botting on 2-3 accounts on home IP is fine.  You’re using injection though.. as others pointed out, that’s the main problem.

Only way is to hire someone to write you a script built with machine learning and train it with data.

On 5/20/2020 at 1:49 PM, shreder said:

Only way is to hire someone to write you a script built with machine learning and train it with data.

That's a way to do it, but clearly not the only way.

For training ML, you need to have an effective feedback mechanism. I.E. do X, score goes up. The sooner the feedback happens after the action, the better.

For getting banned, there isn't a whole lot to go on. Whatever you do that gets you banned, doesn't trigger immediately. Most bans are delayed bans. As well, no single action causes the ban. That's a really poor situation for trying to train a machine learning script.

At the current state of the art, people are far better at learning when there is very little data to go on, than computers. Computers need loads of data and iterations to train machine learning. Imagine creating 200k accounts and getting them all banned in order to train a machine on what gets accounts banned? Well, the thing that gets accounts banned is having large numbers of accounts doing statistically indistinguishable actions, so.... yeah that's not going to work.