BIG BRAIN

A Poker Playing AI created by

team Half Suited

Kesha, Joe, Ted and Tom

AI

MS

• Learn a new language

• Apply all our learning at Makers

• Explore data structures and back end development

• Produce a tangible product

• Write our own code, unassisted

Our Approach

• Gambled with a taking on such a huge project

• Step by step approach

• Solid communication on project direction, stand ups and retros

• Flexible MVPs

Process

• Started simple to help learn Java, deck of cards, dealer and bot

• First bot analysis was dealing with the first two cards

• Slowly built up the model to include a card evaluator, hand comparer, hand weigher and second player

• Linked all together using a GUI

Tech

How BIG BRAIN "thinks"

• Takes the first two cards and assigns a weight

• Adds their values together and applies bonus

+2 if it contains a high card (10 or higher)

+10 if both cards are a high card

+8 if the hand is suited

+5 if the hand has to consecutive cards

+8 if the hand is suited

x2 if the cards are a pair

How BIG BRAIN "thinks"

• Cards from flop, turn and river are given to its hand

• Calculates the best 5 card hand

• Bets accordingly

How BIG BRAIN "thinks"

• Calculates possible draws

• Adds a bonus weight

+26 if the hand is one off a flush

+13 if the hand is one off a straight, straight flush or fullhouse

-13 if the cards in hand are not in the best 5 cards

+7 if one card in hand is in the best 5 cards

+20 if both cards in hand are in the best 5 cards

What was tough

• Initial decision of MVP

• Using an IDE for the first time

• Calculating a straight

• Java

What was good

• Working bot that beats James at poker
   

• 99% test coverage

• Strong OOP, and SRP

• Taught ourselves a new language

• Daily stand-ups and retrospectives, pair programming

• Had fun

Thanks for listening

Check out our blog

pokerprogressioninjava.wordpress.com