Rocklike NL by TylerK

So, this came up on IRC, but I had to have a tabula rasa to do the math. TylerK was playing a 10 NL ring game, going all-in PF with AA/KK/QQ/AKs, and folding everything else. Also playing the blinds.

The question is, how profitable is this? How many callers do you need?

First simplistic model - AA only.
You get AA 0.2% of the time, or 1 in 440

If everyone folds, you win 0.15 (blinds) plus ~.20 from limpers or raisers
If you get a caller, you are going to be a 4:1 fav - assuming stack of ~$6, you should win $4.80 on average.

Assume that you are -.10 on blind structure of .05/.10., so 440 hands costs you $4.40, so if you get calls 100% of the time, you will make about $0.001/hand, or if you prefer 0.40 (>3xBlinds) per AA.

It's actually disasterous, because you are only going to get called 25% of the time and be losing $4!

If you losen up and play KK/QQ/JJ AKs the same way, then you are playing 1 in 100 hands, so at the same payoff (4:1), you are winning $3.00 per all in. So if you are called every time, you make .03/hand (woo hoo).

Except... you don't get called 100% of the time, 5% of the time you are looking at worse odds.

So 5% you are losing $10, and 95-X% you are down about $1.15 (10xblinds + 1xblinds + 2 limpers), X% you win $3.00.


So... 3.0X = .05(10) + (0.95-X)(1.15)
3X = .50 + 1.09 - 1.15X
1.85X = 1.59

X = 86% of the time, not gonna happen.

Best to play like Rippy...

Holy crap, you're some kind of futuristic scientist sent back to destroy us!!

Nice post though. Doing the math to figure it out was probably a better way to find out than to just go field test it.

if by "futuristic" you mean "unemployed" then, ya, you're right on

Originally Posted by TylerK

Holy crap, you're some kind of futuristic scientist sent back to destroy us!!

Nice post though. Doing the math to figure it out was probably a better way to find out than to just go field test it.

If by sent back to destroy you, you mean helping you not lose money, then yes, you're correct. I think a thank you is in order .

Here is true rock NL poker:

Preflop, the only hands to reraise with are AA and KK.
Raise from any position (maybe occasionally from EP a limp-reraise w/ AA or KK or a limp-call with QQ and AK): AA KK QQ AK
Limp Early/Raise Mid-Late: JJ-TT, AQ
Limp and call raises from any position with 99-22, looking to hit sets and win big pots.
Limp and occasionally raise (or depending on how tight the rest of the table is, maybe frequently raise) all the "20+21" suited hands in mid-late.
Muck everything else preflop and play defensively postflop unless you have better than top pair. Basically, wait for big pocket pairs and flopped top-two pair or better before getting a lot of money in the pot.

You're almost guaranteed to win money this way, or at least to not lose much.

Tyler if you actually tested that method you would have one boooooooring day, zen how many hands are you going to get AA,KK,Aks, or QQ in 1000 hands? Another thing, were did you learn to do that kind of math, I get 100 in algebra 2 but cant understand that


-anto

actually you'll get AA 0.45% of the time, or 1 in 221.

you can calculate this with (4/52) * (3/51) or by (4C2) / (52C2)

Didn't I tell you to always check my math???


P(AA) is 1 in 220, not 440. I was looking at a spread sheet I made with an error in it, so you are making 4.80 - 2.20 = 2.60/call on average, instead of 40 cents

If everyone folds, you are down about $1.80 with AA, so

2.60X = (1 - X)1.80
X = .40
You break even if you get a caller only 40% of the time!

Now expanding to KK/QQ/JJ/AKs, you will hit 1 in 50, not 1 in 100!
BUT a caller will have a hand as good or better ~10% of the time not 5, and you are winning $4.20/caller, AND if you steal, you are only down about .15

So
4.2X = (.90 - X).15 + .10*10.0

4.35X = 1.135

X = 26% plus the 10% that you are in coinflip or dominated, which is not that far from what you were getting...

Originally Posted by elanto

Tyler if you actually tested that method you would have one boooooooring day, zen how many hands are you going to get AA,KK,Aks, or QQ in 1000 hands? Another thing, were did you learn to do that kind of math, I get 100 in algebra 2 but cant understand that

I was sitting in class bored out of my mind and I thought it'd be a nice diversion, or at least a little funny.

You'd be surprised how often one of these hands comes up when you have 4 tables open at once, though.

Originally Posted by zenbitz

Didn't I tell you to always check my math???


P(AA) is 1 in 220, not 440. I was looking at a spread sheet I made with an error in it, so you are making 4.80 - 2.20 = 2.60/call on average, instead of 40 cents

LOL...I remember saying "1 in 220" and thinking "oh, super scientist math guy must be right, I'll shut up" when you said 1 in 440.

Originally Posted by zenbitz

If everyone folds, you are down about $1.80 with AA,

Hmm. By my calculations you will get your hand once in every 24.5 orbits around a full 9-handed table. In the example case with blinds of .10 and .05, doesn't that mean you're posting about $3.60 in blinds for each AA?

I was assuming table of 10, not 9 silly poker stars...

BUt, I probably still made a mistake.

So, 10 handed, in in 22 orbits - you pay 3.30 in blinds BUT I assumed that you might win from the the blinds occasionally, so I only charged you .10, not .15.

That's where the $2.20 came from.

For Elanto - this is basic probability with a little algebra. I think I took probability my junior year in high school - honors math. I got a C, dropped honors math! (tough school though). I didn't really understand it until I took statistical mechanics (physical chemistry) in college.

I never learned a damn thing in math class, after algebra. Learned it all in physics and chem.

Nice stuff.. I gave this strategy some thought not long ago.. I was actually pondering the idea of whether it would be profitable to make a bot that plays this way. Just program to go all-in on AA, KK, etc.. Even if you could show just a tiny profit, you should be able to make a killing by having it play on multiple tables at the same time 24 hours a day..

Anyways, I didn't really have it in me to do the math, so I'm glad to see this thread.

One thing you might want to consider is whether there is an optimal stack size for this strategy. For example, suppose you only bought in for $5 instead of $10. This would likely increase the number of calls that you get from your all-in bets, but what would it do to your EV?

There was a thread about this at 2+2..

Any NL rock strategy is just plain ineffiicient if pushing AA KK and QQ hard and limping other pocket pairs mainly looking to jam the pot after flopping sets aren't the key elements of it.

Actually, if you look carefully (and why bother) you will see that I assumed that average caller's stack is only $6, not the full $10, so you are only risking 6$ in this calculation.

The problem with a bot is two fold :
- 1, I assumed you make back 1/3rd of the blinds,
- 2 it is a trivial strategy to beat - only call the guy when YOU have AA/KK/AK!

Even if you don't realize it's a bot, it won't take many orbits to realize what the strategy is...

Originally Posted by zenbitz

Even if you don't realize it's a bot, it won't take many orbits to realize what the strategy is...

Not if you program it to switch tables and/or poker rooms every 2 orbits.