Purely Formulaic SNG

[cagedman]

What do you folks think of the possibility of making a profit a the 10+1, 20+1 (say, 5-10% ROI) with nothing more than a player-ignorant decision tree?

A purely formulaic approach.

Think it's possible?

It is for No Limit ring (25NL, tested over 750,000 hands). I've done it, and I'd be willing to share in exchange for a low level SNG decision tree.

[pergesu]

Yes, you can write a bot that will beat the $11s and $22s

[cagedman]

Care to discuss the general concept (strictly in regards to actual play)? The idea of formulaic bubble play seems complicated to me as a successful ring player and a semi-green SNG player.

[flyingmoose]

Do a search for ICM. A bot would basically make all +$EV plays according to ICM, and that's all that's needed to beat the 11-33s.

[cagedman]

It seems like the difficult part would be weighting the results of the ICM computations against the likely results of an action. It's difficult to put people on a range of call hands at this level, so it's difficult to determine the relative frequency of wins, losses, etc. (i.e. for a given push/fold scenerio)

Maybe this point is not important at this level? (i.e. if a play is profitable with a conservative hand distribution, then it will only be more so with poorer play)

Thoughts?

[Isura]

[ QUOTE ]
It seems like the difficult part would be weighting the results of the ICM computations against the likely results of an action. It's difficult to put people on a range of call hands at this level, so it's difficult to determine the relative frequency of wins, losses, etc.

[/ QUOTE ]

Good points. But from my time experimenting with ICM, exact hand ranges don't change the EV of play that much. Basic tight, normal, and any two type ranges are all that are really necessary to figure out a reasonably accurate decision.

[cagedman]

That's what I was hoping for.

[flyingmoose]

By the way, if you're thinking about making a bot, I wouldn't advise it. Party is pretty good at catching them.

[cagedman]

I'm not--more interested in a mindless strategy for massive multitabling...

[pergesu]

How massive is massive? Plenty of people here 8-table with somewhat mindless play, and a number of guys 12-table. Just takes some practice.

[flyingmoose]

Who besides Raptor 12-tables?

Jesus, I don't even want to think about trying it. Playing 12 tables at the same time would make my eyes bleed.

[pergesu]

ZeeJustin does, and I'm pretty sure a couple other guys do too.

Not sayin it's easy...but people do it.

[Freudian]

[ QUOTE ]
Do a search for ICM. A bot would basically make all +$EV plays according to ICM, and that's all that's needed to beat the 11-33s.

[/ QUOTE ]

A big weakness of ICM as it is, is that you have to guess as to what the others players range is (and often be wrong, which is often overlooked when people claim a call is +1.3% or whatever).

A player independent bot doing ICM calculations would be like a blind leading the blind.

Of course one could always assume a medium range and be slightly off most of the time, but it would still probably be able to play profitable SnGs at the lower limit. Just not optimal.

[flyingmoose]

[ QUOTE ]
[ QUOTE ]
Do a search for ICM. A bot would basically make all +$EV plays according to ICM, and that's all that's needed to beat the 11-33s.

[/ QUOTE ]

A big weakness of ICM as it is, is that you have to guess as to what the others players range is (and often be wrong, which is often overlooked when people claim a call is +1.3% or whatever).

A player independent bot doing ICM calculations would be like a blind leading the blind.

Of course one could always assume a medium range and be slightly off most of the time, but it would still probably be able to play profitable SnGs at the lower limit. Just not optimal.

[/ QUOTE ]

I agree with you that it wouldn't be optimal, but I do think it could beat the god-awful 11 and 22 players for a decent clip.

[Freudian]

If Saabpo could do it without postflop play, a bot knowing when to push or check/fold a flop would win more.

[jeffraider]

It's doable, but it would be a bunch more complicated than the NL Ring strategy that you've got. I think it would need to at least somewhat take into account the stack sizes of the other remaining players.

I'd like to see the NL Ring strategy one because I keep getting pwned at NL Ring.

[Blackjack]

what was your rate in the 25 NL?