What could possibly go wrong? Please, tell me :)

[tobi4255]

I have made a program that I run alongside the poker table i play at. When cards are dealt pre-flop, my program will detect the hand I was dealt, the current pot and the number of playing opponents.

Then it simulates poker games to find out what the chances of winning are (approximately, but still) and then uses that data to calculate the expected value of going all-in pre-flop with the hand dealt.

If the expected value is significantly in the positive, it goes all-in, still on the pre-flop. The expected value is calculated under the assumption that at least 1 player calls the all-in, if more players calls it the EV goes even higher.

If no players call, the blinds are collected along with any raises/calls that might have happened before my turn.

If an all-in is lost, the table is left and re-entered with same buy-in.

If an all-in is won, the table is left and re-entered with same buy-in, rinse-repeat.

If it ties, the game is continued unless balance was significantly lowered in which case its treated as a loss.

I started out with 0.01 BTC and I am currently at 0.016 + a withdraw of 0.005 btc.

I must admit though, I had some crazy luck in my test phase. First attempt ever at this strategy, 3/7 players call my all-in and then I win it with a three-of-a-kind from the flop.

So... Any poker adepts out there who can help me optimize this strategy, or tell me if im asking to bust in the long run despite my (carefully) calculated EV.

[Wendigo]

Are you sure that getting more players pre-flop will increase the EV you are getting? I think more players going all-in will lower the EV because this will increase the variation and the possibility for your better hand (pre-flop) to get beaten on the river for example. I think it's better to get one or two players calling at most and in general try to get isolated calls.

My 2 satoshis.

Disclaimer: I am not a poker pro.

[tobi4255]

Disclaimer: I am not a poker pro.

Neither am I

If more players raises, I stand to win more money. The chances of some1 having a better hand is also that much larger, but simulations are run with the max amount of possible opponents.

if against 4 opponents, it will generate 4 random hands (excluding unavailable cards) and then play out the round. So it accounts for all players being active in the round before going all-in.

I would think that by assuming the "worst" scenario, it would counter the effect you mention. And since i am all-in, people cant force more money out of me in that round, but simply make a separate pot im not a part of.

It is very mind boggling though, and perhaps better suited for a mathematics forum

[kodes88]

I have made a program that I run alongside the poker table i play at. When cards are dealt pre-flop, my program will detect the hand I was dealt, the current pot and the number of playing opponents.

Then it simulates poker games to find out what the chances of winning are (approximately, but still) and then uses that data to calculate the expected value of going all-in pre-flop with the hand dealt.

If the expected value is significantly in the positive, it goes all-in, still on the pre-flop. The expected value is calculated under the assumption that at least 1 player calls the all-in, if more players calls it the EV goes even higher.

If no players call, the blinds are collected along with any raises/calls that might have happened before my turn.

If an all-in is lost, the table is left and re-entered with same buy-in.

If an all-in is won, the table is left and re-entered with same buy-in, rinse-repeat.

If it ties, the game is continued unless balance was significantly lowered in which case its treated as a loss.

I started out with 0.01 BTC and I am currently at 0.016 + a withdraw of 0.005 btc.

I must admit though, I had some crazy luck in my test phase. First attempt ever at this strategy, 3/7 players call my all-in and then I win it with a three-of-a-kind from the flop.

So... Any poker adepts out there who can help me optimize this strategy, or tell me if im asking to bust in the long run despite my (carefully) calculated EV.

what program is it?did you build it with programing language?or just using software?
i think it can used for poker only,but its really amazing for me,because i never use that thing,i just play normal and combining between strategy and luck..

[YuginKadoya]

I have made a program that I run alongside the poker table i play at. When cards are dealt pre-flop, my program will detect the hand I was dealt, the current pot and the number of playing opponents.

Then it simulates poker games to find out what the chances of winning are (approximately, but still) and then uses that data to calculate the expected value of going all-in pre-flop with the hand dealt.

If the expected value is significantly in the positive, it goes all-in, still on the pre-flop. The expected value is calculated under the assumption that at least 1 player calls the all-in, if more players calls it the EV goes even higher.

If no players call, the blinds are collected along with any raises/calls that might have happened before my turn.

If an all-in is lost, the table is left and re-entered with same buy-in.

If an all-in is won, the table is left and re-entered with same buy-in, rinse-repeat.

If it ties, the game is continued unless balance was significantly lowered in which case its treated as a loss.

I started out with 0.01 BTC and I am currently at 0.016 + a withdraw of 0.005 btc.

I must admit though, I had some crazy luck in my test phase. First attempt ever at this strategy, 3/7 players call my all-in and then I win it with a three-of-a-kind from the flop.

So... Any poker adepts out there who can help me optimize this strategy, or tell me if im asking to bust in the long run despite my (carefully) calculated EV.

what program is it?did you build it with programing language?or just using software?
i think it can used for poker only,but its really amazing for me,because i never use that thing,i just play normal and combining between strategy and luck..

You can certainly play with your strategy and luck with poker, but I am curious too if what kind of programming language he is using because his post is certainly not complete, he just start by saying that he is playing with the simulation test, and another questions is does this kind of poker includes PVP, or is it all an AI kind of a poker? and how many tries till you get to a 0.016 winnings, the calculation is really needed here so you can get clear answers!

[tobi4255]

--snipped--

what program is it?did you build it with programing language?or just using software?
i think it can used for poker only,but its really amazing for me,because i never use that thing,i just play normal and combining between strategy and luck..

It is a simple python program that I made myself, with help from Deuces.

[tobi4255]

--snipped--

--snipped--

You can certainly play with your strategy and luck with poker, but I am curious too if what kind of programming language he is using because his post is certainly not complete, he just start by saying that he is playing with the simulation test, and another questions is does this kind of poker includes PVP, or is it all an AI kind of a poker? and how many tries till you get to a 0.016 winnings, the calculation is really needed here so you can get clear answers!

... What? I get the feelz you are only here because of your sig campaign

If not, feel free to ask concise and clear questions...

It is regular texas hold 'em poker against other humans. If you play poker against "AI" then there is likely a house-edge making it impossible to get positive expected value.

Starting from 0.01 BTC, it took roughly 100 hands to reach the 0.016 BTC (withdrew 500k sats) and now approx. 250 hands in the balance is 0.0085.

[marlboroza]

You are what we like to call "bingo player"  and you only had some luck, nothing more.
Most of people will just fold to your all in if you keep doing all in on high pair or whatever - unless they are bingo players too or they have high pair too.
Try this strategy on high stake tables, you will notice that everyone will sit out or just simply leave table.

[tspacepilot]

I'd like to know how you're calculating EV.  Your comment that more people calling leads to greater EV seems pretty suspicious.

[tokeweed]

I have made a program that I run alongside the poker table i play at. When cards are dealt pre-flop, my program will detect the hand I was dealt, the current pot and the number of playing opponents.

Then it simulates poker games to find out what the chances of winning are (approximately, but still) and then uses that data to calculate the expected value of going all-in pre-flop with the hand dealt.

If the expected value is significantly in the positive, it goes all-in, still on the pre-flop. The expected value is calculated under the assumption that at least 1 player calls the all-in, if more players calls it the EV goes even higher.

If no players call, the blinds are collected along with any raises/calls that might have happened before my turn.

If an all-in is lost, the table is left and re-entered with same buy-in.

If an all-in is won, the table is left and re-entered with same buy-in, rinse-repeat.

Poker sites don't allow you to exit the game and then re-enter with the original buy in amount, you should buy in with the same amount you left with if it's over the buy in maximum.

If it ties, the game is continued unless balance was significantly lowered in which case its treated as a loss.

I started out with 0.01 BTC and I am currently at 0.016 + a withdraw of 0.005 btc.

I must admit though, I had some crazy luck in my test phase. First attempt ever at this strategy, 3/7 players call my all-in and then I win it with a three-of-a-kind from the flop.

So... Any poker adepts out there who can help me optimize this strategy, or tell me if im asking to bust in the long run despite my (carefully) calculated EV.

What you're trying to do is sub optimal.  If there really was a way of winning without actually playing poker then somebody would have thought about it already.  What you're doing is not only sub optimal, it's not fun and doesn't help you or the game at all.

[sir.humus]

If it ties, the game is continued unless balance was significantly lowered in which case its treated as a loss.

I started out with 0.01 BTC and I am currently at 0.016 + a withdraw of 0.005 btc.

I must admit though, I had some crazy luck in my test phase. First attempt ever at this strategy, 3/7 players call my all-in and then I win it with a three-of-a-kind from the flop.

So... Any poker adepts out there who can help me optimize this strategy, or tell me if im asking to bust in the long run despite my (carefully) calculated EV.

What you're trying to do is sub optimal.  If there really was a way of winning without actually playing poker then somebody would have thought about it already.  What you're doing is not only sub optimal, it's not fun and doesn't help you or the game at all.

It's right, you won't get anywhere with this tactic. you can roll the dice as well and save some time, but don't expect too much winnings.
But I did like the idea of playing with a simulator next to you, and do your moves when you know all the odds.
Providing I have some Python skills, how long will it take to build something like this?

[tspacepilot]

To people  "it wont' work": I would consider that to be an empirical question.  In point of fact, people on low-stakes poker tables aren't necessarily the best poker players.  Whether OP's strategy works depends (in part) on the competition.   malboroza is almost certainly right about higher stakes players: they'll just leave the table.   but with low-stakes players, you never know, they may want to play bingo---i've seen some funny stuff on sealswithclubs back in the day.

tokeweedis right that most sites aren't going to let you leave and rebuy immediately with less than you left with.  You'd have to wait for some time limit to expire in order to re-enter with just the buyin amount.

I still want to hear more about the OP's calculations for EV.  Please.

[tokeweed]

To people  "it wont' work": I would consider that to be an empirical question.  In point of fact, people on low-stakes poker tables aren't necessarily the best poker players.  Whether OP's strategy works depends (in part) on the competition.   malboroza is almost certainly right about higher stakes players: they'll just leave the table.   but with low-stakes players, you never know, they may want to play bingo---i've seen some funny stuff on sealswithclubs back in the day.

tokeweedis right that most sites aren't going to let you leave and rebuy immediately with less than you left with.  You'd have to wait for some time limit to expire in order to re-enter with just the buyin amount.

I still want to hear more about the OP's calculations for EV.  Please.

Low stakes poker are full if nits now.  It wasn't like pre 2006 where almost all the players in the table are there to gamble.  Today all the info on how to play the game are available everywhere online, most of them now follow a hand chart on what to do preflop and in what position they are, then they play accordingly postflop depending on the strength of their hand.

OP's strategy might be profitable pre 2006 but today it's break even at most.

[NorrisK]

You are not maximizing EV by making over sized bets at the wrong time.

You need to milk the opponent when you are ahead. This is called value betting.

I would recommend reading up on poker strategy to optimize the bot. Basically what you are doing is what pros are always doing in their head. They know the winning percentage of their hand pre-flop and consider the range the opponent is playing depending on his stats.

[tobi4255]

I'd like to know how you're calculating EV.  Your comment that more people calling leads to greater EV seems pretty suspicious.

The program knows my hand, and the number of opponents playing. It calculates the % i win, the % i lose and the % i tie out of ~55k simulations.

I then add the pot to my table balance times two, to also simulate at least 1 call. So, it calculates EV for playing against all opponents for the pot made up of only 1 call, so more calls will make the risk more worthwhile since the odds stay the same but the possible return increases.

Suspicious is a strong word... I am just having fun with Python, and my interest for gambling.

[tobi4255]

You are not maximizing EV by making over sized bets at the wrong time.

You need to milk the opponent when you are ahead. This is called value betting.

I would recommend reading up on poker strategy to optimize the bot. Basically what you are doing is what pros are always doing in their head. They know the winning percentage of their hand pre-flop and consider the range the opponent is playing depending on his stats.

Exactly, maybe i was not clear about this in above posts.

The reason i came up with trying this out, was after seeing how many flush-draw hopium i take during a poker session. People will see an all-in preflop, and be plagued by what-ifs, while my bot has the odds down.

I dont intend to win millions from this, although I wouldnt mind. ;-)

[tobi4255]

tokeweedis right that most sites aren't going to let you leave and rebuy immediately with less than you left with.  You'd have to wait for some time limit to expire in order to re-enter with just the buyin amount.

Yes, exactly. After sitting out for approx. 3mins depending on activity ( i assume ), my balance is cleared from table and i can re-enter.

I still want to hear more about the OP's calculations for EV.  Please.

Example calculations:

My hand: 9s,td
Table/community cards: None (pre-flop)
opponents: 4
current pot: 50
balance on table: 200

Simulation (http://i.prntscr.com/1G1p8s9uQp2MFrN1jZrq-A.png)

so 21.18% chance to win 450 (assuming at least 1 call to the all-in of 200) and a 75.93% of losing 200 (if no blinds), and 2.89% chance of a tie that results in a shared pot (0).

I hope that helps

This is my source for the mathematics

[tobi4255]

Providing I have some Python skills, how long will it take to build something like this?

Not long at all, see my above post where i put a link to the python module i use to simulate poker games. It is called Deuces, and it is truly awesome!

The tough part is "connecting" to a poker table and be able to act on it. In order to run as many simulations as possible it is vital to get the parameter data asap

[tspacepilot]

I'd like to know how you're calculating EV.  Your comment that more people calling leads to greater EV seems pretty suspicious.

The program knows my hand, and the number of opponents playing. It calculates the % i win, the % i lose and the % i tie out of ~55k simulations.

More on this below, but I assume that the number of simulations is because you're trying to model the indeterminacy of the opponent's behavior?   I guess I'm not completely clear why you'd need to run simulations.  If you have a model, can't you just read the parameters and make a decision?

I then add the pot to my table balance times two, to also simulate at least 1 call. So, it calculates EV for playing against all opponents for the pot made up of only 1 call, so more calls will make the risk more worthwhile since the odds stay the same but the possible return increases.

So, this makes me think you've got something wrong already.  Lets say you go in with a high hand, so you have a high probability of winning.  But the more opponents you have against this hand, the less likely it is to win because of how poker hands work.  I'm pretty sure that if you're correctly modelling your distribution, you'd find that the chances of AA winning against one opponent are higher than the chances of AA winning against two opponents (and so on).

Suspicious is a strong word... I am just having fun with Python, and my interest for gambling.

Sorry if I was misunderstood.  I'm suspicious of the correctness of your calculations, nothing more.

tokeweedis right that most sites aren't going to let you leave and rebuy immediately with less than you left with.  You'd have to wait for some time limit to expire in order to re-enter with just the buyin amount.

Yes, exactly. After sitting out for approx. 3mins depending on activity ( i assume ), my balance is cleared from table and i can re-enter.

On poker sites I've used, it's more like 20 minutes to an hour.  But I guess you know the principal anyway.

I still want to hear more about the OP's calculations for EV.  Please.

Example calculations:
 
My hand: 9s,td
Table/community cards: None (pre-flop)
opponents: 4
current pot: 50
balance on table: 200

Simulation (http://i.prntscr.com/1G1p8s9uQp2MFrN1jZrq-A.png)

so 21.18% chance to win 450 (assuming at least 1 call to the all-in of 200) and a 75.93% of losing 200 (if no blinds), and 2.89% chance of a tie that results in a shared pot (0).

I hope that helps

Erm, not really.  I mean, that's a screenshot that shows the output of your program, not how you got there.

This is my source for the mathematics

I took a brief look here.  I'll look more closely later.  Thanks for sharing your source.  I note right away that this discussion says things like

"To answer this, we'll first have to determine how often he calls, how often he raises, and how we fare against each of those. Let's assume that we somehow know ..."

it's not so clear to me how you can determine this in real life without a lot of data about particular opponents.  I suppose you can do something like what this article does---model some generic opponent---and see how they fare.  Is this what you're doing?

Providing I have some Python skills, how long will it take to build something like this?

Not long at all, see my above post where i put a link to the python module i use to simulate poker games. It is called Deuces, and it is truly awesome!

The tough part is "connecting" to a poker table and be able to act on it. In order to run as many simulations as possible it is vital to get the parameter data asap

Did you link to something other than that screenshot?

----

Responding to your question about connecting to a poker room to get the data: did you try OCR?  What have you tried?  If your poker program sends the hand data in the clear you should be able to use wireshark to read that straight from the network socket.

[tokeweed]

You are not maximizing EV by making over sized bets at the wrong time.

You need to milk the opponent when you are ahead. This is called value betting.

I would recommend reading up on poker strategy to optimize the bot. Basically what you are doing is what pros are always doing in their head. They know the winning percentage of their hand pre-flop and consider the range the opponent is playing depending on his stats.

It takes more than reading to learn the most optimal poker strategy.  It takes logging a lot of hands in Poker Tracker or HM and study all the numbers and see what works for you and what doesn't.  tobi4255 is scamming himself when he thought he could make a lot of money doing what he's trying to attempt.