Quote:
Originally Posted by jalfrezi
The bolded is what I based my research on, and it revealed that the site whose framework your code sample came from does not integrate with PT (unless you call hard coding player aliases and player types "integration").
I'm not sure how you didn't run across PT integration. My code sample was 6mo old btw and from a previous bot I first started with. I replied that I won't give out any current code because of reasons enumerated in past replies. But, even so, I'm surprised you didn't find it. I'm pretty sure I would get banned if I posted it publicly so best I can do is if a mod PMs me, I will share the info for PT database integration from both the github repository for the open source framework and user-manual how-to's for a couple of cheap, plug-n-play products. On your last point, I did also hard code for player aliases for certain regs who would end up being on almost every table the bot was playing on.
Quote:
Originally Posted by AmazingErvin
OP thank you very much for doing this thread and your honesty about an unknown topic. LOL at people with their swords drawn who want him to out himself just to give him the ungabunga.
OP what are challenges/limitations you've had with botting/strategies and what do you think can be improved in the future? I'm assuming the challenges are more for programming and not so much poker strategy - we already have bots who play limit HU optimally but there are infinitely more variables in NL so I'm guessing you have a general strategy the bot plays but constantly have to make alterations for weird spinoffs? Or is the bot doing wild stuff like deciding someone's range on a river is straight garbage, it 100% has the best hand, overbetting 2.8x pot mixed in with perfect frequency etc?
Creating bot behavior for different situations is still inelegant grunt work. It basically involves beginning by coding the fundamentals of what types of hands a bot should play based on position, stacksize, etc. and how they should play those hands post-flop. Then you run the bot, get a few thousand hand sample size, go over the hand histories and start coding for specific conditions and situations that you see come up that you haven't differentiated yet. Then you run the bot to get a few thousand hands and repeat the above step, over and over. To get my bot to perform to my standards at the 50/100nl level was roughly 3 months of daily brute work. It's pretty labor-intensive to be honest. For example, when I first started, I may have coded the bot to play full houses a specific way but then I'd run across a hand where there are trips on the board, and the bot has the best full house available via the board but would lose to some overpair pocket pairs in the villian's range. Well, I would now have to add instructions for the bot so it can deal optimally with situations like this and other situations comparable to this (like if the bot has a pocket pair itself, etc.) FYI, I'm not a noob so I took this into account on my first version--the above situation is just for demonstration purposes
Once you get a lot of this grunt work down, you can start incorporating more opponent-specific actions for these situations and so on and so forth.
Quote:
Originally Posted by TennisGolfandPoker
I'd rather play against bots than nobody at all. OP, would you rather have 3 bots at your table or 3 good human regs? Also not sure if you addressed if bots are common in mtts and sngs? Very insightful thread op, thanks for transparency
This is a fantastic question. I would much rather have 3 bots in my ring tables than 3 really good human regs.
Bots are more common in MTTs and specifically certain types of sngs. Fundamentally, the narrower the scope of decisions that a bot will have to make, the easier it is to build it to play well. I focused all of my attention on 6max ring but I know that in general, from observing threads and chats, sngs and MTTs where far more popular. Specifically, hypers seemed to have the greatest interest. Games that start off shallow or that degenerate quickly to <25bb or push/fold situations are godsends for bots. Some botters also seem to like the idea of having the bot play through the early rounds of a bigger tournament and transfer control to a human player once it gets close to the money or at the final table.
Quote:
Originally Posted by xStormy
The way this post reads to me: "I made X amount of money doing something and never got caught, its a process that involves no input from me....but I swear I quit "
If you read through some of my previous replies, this wasn't the case for me at all. No "free money". Every day that I ran the bot I would have to put in hours after the completion of all sessions to examine, fix bugs, modify/add code, etc.
Quote:
Originally Posted by AltruisticRaven
I'm absolutely shocked that the OP continued to reply even after all of the ******ed speculations and feedback he has gotten. I guess I shouldn't expect anything different on a forum without upvotes/downvotes/thanks frequented by some of the most dense people around.
Very interesting thread OP. Thank you for shedding a little light on the botting situation.
Thanks! I wanted to structure this as a well so people are more informed and familiar about botting culture. I expected to receive a fair amount of abuse but I was/am hoping to also field some more interesting questions.