ProPokerTools came up with an interesting way to rank poker hands using an evolutionary sim as described in this blog entry:

http://pokercoder.blogspot.com/2006/...-of-hands.html

I decided to try the same approach for Lazy Pineapple. I started with a ten-handed table and did ten iterations. Then for comparison I did the same thing for a five-handed table. Here are the hands for each sim.

Ten-handed Lazy Pineapple High-Low 8.

Five-handed Lazy Pineapple High-Low 8.

Any surprises? Not really.

On the ten-handed evolutionary sims, the hands that lost the most rank when compared to ten-handed show-down sims were hands similar to 932r. I assume with some players folding and others staying with aces, these hands didn't earn enough when they hit.

The hands from the evolutionary sims that gained the most when compared to showdown sims were hands such as (87)5 and similar.


When comparing ten-handed evolutionary sims with five-handed evolutionary sims, the ten-handed gainers were similar to T(98). The five-handed gainers were hands with mid- to big pairs such QQQ, TT2, TT5 etc.

Of the positive-EV evolutionary sim hands, the biggest ten-handed gainers were 543 and 654 type hands. The five-handed gainers were A99, (A98) and AJ7.

Very cool!

To give credit where it is due, when I did this for omaha and hold'em, I stole the approach from this paper:

http://www.cs.ualberta.ca/~duane/pdf/2002aij.pdf

bachfan, ProPokerTools guy

That's absolutely fantastic, Mack!

Very interesting!

It's the complete list of all 1755 uniquely different three card hands arranged in order of starting hand value for the keep all high low three game, as per the ProPokerTools models (which are also fantastic)!

It will take a while to digest.

Thank you.

Buzz

That link doesn't work for me, Bachfan. (I get a "Not Found on this server" message). I'll try it again and delete this post if it works.

Prokofievfan

I felt I should have mentioned that, but couldn't get the link to work. Here's another link to, I believe, the same paper:

http://webdocs.cs.ualberta.ca/~darse/Papers/AIJ02.pdf


PS: Just want to add that ever since I saw the first PPT Flopalyzer™, I've been fascinated by the work you are doing.

I've never tried showdown pineapple sims with a discard on the flop because I thought the results wouldn't teach me much, but I think that with the ProPokerTools-UofA approach, a discarding sim could produce something of interest. If I come up with a straightforward discarding algorithm I may give it a shot, but it could well take forever to run...

Thanks. That link works.

I got there and read the paper.

Although I didn't understand the paper leading to your ranking system, I think I understand your ranking system.

I'm vaguely surprised that trip aces rates so highly. My understanding, and the way we play the game, is you get to keep all three cards but may only use two of them to make a high hand (and at most another two, or a different two to make a low hand.

Thus you may use
• two hand-held cards + three common board cards,
• one hand-held card + four common board cards, or
• no hand-held cards + five common board cards
to make a five card poker high hand, and
• two hand-held cards + three common board cards,
• one hand-held card + four common board cards, or
• no hand-held cards + five common board cards
to make a five card poker low hand (8 or better).

I assume it's the same deal on your simulations.

Absolutely fantastic lists! Very impressive achievement! You are the LP-8, CP-8, and P-8 guru!

I'll study your lists and maybe set my own thinking straighter.

Thank you again.

Buzz

Here are the results for PPT-style sims for ALL-HIGH Lazy Pineapple.

Ten handed:

Five handed:

I think one of the reason AAA might rate so well is that these are survival sims, and locking up so many aces limits the number of opponents that hero plays against--if that makes any sense.

It does. Thanks.

Does each hand have a total associated with the number of wins? (I can see how this would be difficult since poorer hands are more or less continually dropped out of the simulations). Do I have the correct impression?

Buzz

Here's an animated gif showing the ten iterations of the all-high ten-handed sim attempting to stabilize:



I may have transgressed in hosting this image, so it may stick.