Anti detection and anti ban

Anti ban - I have a long history of botting, including Texas Hold'em bots, back in the mid to late 00's, when that scene was vibrant, before "Black Friday". I made bots that made a lot of money, so I had a very strong interest in ensuring that those bots did not get detected as such by the casinos. I have no evidence that SuperCell is actively searching for bots, but they very well might be, so lets err on the side of caution.

As an example of bot detection methods, the main way to detect bots is by click patterns. No human touches the same button in the same place every time, but rather in a clustered random fashion, both in space and time. This bot uses the algorithm described here for every click: http://stackoverflow.com/questions/23700822/weighted-random-coordinates

There are detection weaknesses in this bot as well. The troop deployment method in particular, although still using the clustered random approach, has a fairly unique signature. At the moment, there is not a better solution - detecting the red box that bounds where troops can be deployed is a non-trivial exercise, and thus there is not an implementation that I am happy with. Also, the bot deploys troops a bit faster than most humans could do, but not unreasonably so. Barching in particular is more effective the faster you can get your troops out there, so I am content to live with this the way it is, for now.