Poker game demo - BETA

POKER game demo BETA - v0.8:



http://www.mediumedge.com/poker/poker_v08_2.zip
DOWNLOAD DEMO

[EDIT: updated link to new v08.2 demo]

Summary:
It's primarily a test of the AI and Poker rules. *ALL* of the graphics are temp!!! The finished product will look completely different. Trust me! So there's no need to comment on the less than stellar graphics.

Player and Computer start with $500. You play until someone loses his/her $500. Betting is limited to four raises per round which limits the overall hand to no more than $400. (1st round: bet$25/raise$25, raise$25/raise$25, raise$25/raise$25, raise$25/raise$25 --- 2nd round: bet$25/raise$25 ,raise$25/raise$25 ,raise$25/raise$25 ,raise$25/raise$25)

Hit escape to quit the demo.

The purpose of this beta test is to check for the following:
* Errors - scoring errors, misreading hand ranks, etc.
* Exploits - is there an easy routine to deprive the AI of it's cash?
* Challenge - the AI is programmed to be tough. Is it? (And no, it doesn't cheat)
* Length - how long does it take to win/lose?

Feedback is appreciated. Thanks!

Enjoy!

STRIP poker... would improve this 1,000% :)

"*ALL* of the graphics are temp!!! The finished product will look completely different. Trust me!"

Help me check this part and then... :)

yeah the numerous strip pocker games on the Amiga got me through my teens.

I found an error in your reading of hand rankings. I should have one this hand with an Ace high:



but it gave it to the computer. Also, I don't know what you're referencing for poker, but usually when both players "Stay" (in the US, it's "Check"), the hand continues to play out. I found it very irritating that we would redeal if neither of us bet. I think the AI is 'OK'.. I found that if I bet heavy, it usually folded unless it had 3 of a kind or better. Finally, it seemed like there were quite a few times that I would Draw and get the same exact same cards in the exact same order. Overall it's a nice start though.

@NTense: Good find. It was incorrectly ranking hands when both players had a high card. *FIXED!*

As for how to handle dual "Stay"s (or Checks). According to "How to Play Poker" by Peter Arnold (pg. 39): if every player, including the dealer, checks on the first round, the deal comes to an end. The cards are collected, reshuffled and cut, and dealt by the next dealer, who puts in another ante of six chips, making the pot twelve chips for the next round."

Finally, in the final round the computer folds a majority of the time if it only has a high card and the player bets $25. It rarely folds with a low pair, even less with a high pair and virtually never with 2 pair or better.

The demo has been updated and is available here:
Download Demo v8.1

More feedback please! If everyone feels the AI is at least "OK" and there are no more errors then I'll move on to the graphics. That will be a very good thing! Trust me.

I tried to test it here on this laptop that has no 3d - the game tries to use 3d mode, so unfortunately I couldn't test it. maybe later.

Hi Dustin,
I played the game quite extensively yesterday (not the 8.1 fix, but the original). It seemed like when I drew extra cards, I would often get the exact same 'value' cards in the exact same positions (suits were different). This happened VERY frequently, causing me to wonder if the dealing or shuffling algorithm needed looking at. Otherwise, I thought the game played pretty well. Good Job.

BTW, I play poker pretty extensively with a large group of people (5 card draw, Texas Hold'm, Omaha Hold'm, 7 Card Stud, Baseball, Follow the Queen, etc..). I found the computer to be a little bit predictable for me. You might want to try to have the AI "test" the player a few times. Maybe track how often the player folds to a heavy bet in the first betting round. If the player repeatedly folds often, the computer then knows that they can probably bluff with nothing but High Card. If their bluff fails and they get nothing on the draw, they can always fold out, and it only cost them their initial Bluff bet. Just a thought to make the AI a little more interactive. :)

I found another error.. We both had 2 pair, and I had the hi pair of the 2 pair, yet the computer claimed the win.

***NEW VERSION!!!***
Demo Version .082


Once again, NTense comes through with the bug hunt!!!

@NTense: Big thanks for giving the program such a workover! I think I addressed all the issues you raised (see below). As for improving the AI, I couldn't agree more. There are key areas that I could greatly improve the AI including detecting player betting patterns, AI betting streaks, and dedicated (versus statistical) bluffing. If the end result was to create a true poker sim then I would continue to improve the AI like you suggested. But I'm not creating a *traditional* poker sim...

[In case I haven't been obvious enough about what I'm doing then reference BloodLocust's & coffeedotbean's previous posts. Hint! Hint! :)]

***VERSION .82 CHANGES***
Okay, I've gone through the ranking system with a fine tooth comb and, I believe, corrected them all. Mainly the mistakes appeared when both had the same hand rank (ex. pair, straight, flush, etc) and the program compared them incorrectly but it all should be fine now.

I added one additional feature. At the end of each hand the program writes out the entire order of the deck to a file named "log.txt" The main purpose is to ensure that the deck is being properly mixed up during the shuffle. As an unintended bonus, if you or the computer folds and it's just killing you to see just what the computer had you can now glance at the "log.txt" file and find out, after the fact, just what cards the computer was holding.

Yes. It wouldn't be poker unless a few baps were thrown in there.