# Winning Algorithm?

#### zachvac

##### Legend
This is probably impossible at high limits, but for low limits, do you think it would be possible to develop an algorithm for winning poker (beating the rake as well)? I generally follow one myself for small pots, but as pots get big and I've got a good hand but not necessarily the nuts, I have to make a read based on betting patterns, which would be tough to develop into hard and fast rules. This is important because if one were to be developed, first of all bots would be profitable, but secondly if one were to utilize such an algorithm, even if it were legally (applying it on their own), they would avoid tilt. Yes it would be boring, but if it's free money for someone who doesn't have to have any poker skill, I'm sure people would do it. So does anyone think an algorithm could beat the rake at low limits? PF would be easy, many exist already using position and starting hand requirements. Most of the time postflop would be
easy, but for example when you hit top pair it'd be tough to determine if you're beat with an algorithm. And when you hit the J high flush and have to decide if the other guy was drawing to a flush as well, no way a computer could figure that out.

Of course we could use a very tight aggressive model, meaning no flush/flush draw would be any lower than Q, maybe J, and just assume we always are winning. 2 pair on the board could slow us down but other than that we can assume a flush means we're ahead. The people pushing and calling with 8 high flushes or trips would most likely balance out when our opponents actually have the nut flush. Thoughts on this?

M

#### maltz

##### Guest
Yeah I think a profitable bot is highly possible in micro limits. To make it more human you can randomize its delay time and preflop hand selection, raise amount, etc.

As for hand reading I think there are a few obvious clues:

(1) Opponent's stats - % see flop, % raise PF, aggression factor
(2) Bet/Raise amount
(3) Thinking time

Also I read that there can be an easy bot. You are constantly shortstacked - whenever you get the top 10% starting hands, you push preflop and hope some guy with deep stack would call. If you win, you hide away the extra and remain shortstacked (probably goes to another table). Believe or not this strategy actually makes you money.

#### pigpen02

##### Legend
It would take time to develop this algorithm, but I am working on it. It keeps getting better the more I work on it, too. I call this algorithm "Experience".

#### zachvac

##### Legend
Yeah I think a profitable bot is highly possible in micro limits. To make it more human you can randomize its delay time and preflop hand selection, raise amount, etc.

As for hand reading I think there are a few obvious clues:

(1) Opponent's stats - % see flop, % raise PF, aggression factor
(2) Bet/Raise amount
(3) Thinking time

Also I read that there can be an easy bot. You are constantly shortstacked - whenever you get the top 10% starting hands, you push preflop and hope some guy with deep stack would call. If you win, you hide away the extra and remain shortstacked (probably goes to another table). Believe or not this strategy actually makes you money.

Have you tried this? I never even thought about this, but playing at \$0.05/\$0.10, I'm surprised how often a 3x raise from even UTG gets no calls. Doesn't seem like this would work, unless you get a string of hands and push a couple times in a row and someone thinks you're pushing with crap.

M

#### maltz

##### Guest
Yeah I try it all the time when I get short stacked in a tourney.
Wait I actually pushed with craps, too.

I think your stack has to be relatively short... like 5-10X big blind.
Also, in NL10 people are generally better than... yeah NL2. Try it at NL2 and push with \$0.5. I think you'll get callers.