Hi guys,
I was linked to this thread by multiple people, so just wanted to jump in. I was one of the persons who helped Schwein (and PS) busting bots over the last two years mainly playing the great game of PLO. We found more than 20 accounts on PS. Massive credit to Schwein! (
http://forumserver.twoplustwo.com/15...-them-1537778/).
I'm not familiar with NLHE and NLHE stats these days, and I do not have the time to do it, but I'll recommend that someone builds a simple data model using 25-30 different parameters and gathers a ton of data. Either from rPTR or across multiple HEM-DB's. The data model should include the basic stats like VPIP, PFR, 3b, 4b, fold to 3b, to the more exact ones like check/raise turn, check/raise flop as initial raiser, steal SB/CO/BTN, bet river checked to etc.
Compare the numbers and find out how many stats are aligned inside a 2-5% interval/pattern. In my experience the bots will have 85-90% of the stats within a 2% interval. Afterwards calculate the euclidean distance between stats from suspected accounts and compare regulars (and control groups). Visualize this in simple diagrams as people tend to understand it way better this way.
Happy hunting!