Computer scientists' basically broke the 'limit of two Texas Hold'em' Em Poker players

Computer scientists’ basically broke the ‘limit of two Texas Hold’em’ Em Poker players

The computer scientist team has developed a program to play the perfect two-seat Texas Hold’em in the long run.

One purpose of game theory is to “complete” the game. This assumes that you are playing perfectly, and if you track perfectly, you will find the strategy of each player to give the best results for each player.

In a paper published in the journal Science a team of computer scientists led by the University of Michael Bowling Alberta announced that they had broken Head-up-Limit Hold ‘Em in nature.

Incomplete information

Tic-tac-toe is a very simple game and has a fairly well-known solution. If two players follow the appropriate strategy, the game will always end with a draw. Checkers were also resolved this way in 2007. Through the process of carelessly analyzing all dam games, available for several years, researchers have identified strategies always leading to draws, such as tic-tac-Tech.

However, poker has an additional dimension that complicates analysis. To see the cards shown to other parties, but you cannot see the other side of the country as opposed to the dam or chess-like game.

This makes decision making significantly more difficult. There are several ways to play a pair of 10 games. You must evaluate your choice based on how your opponent wins and how the community card will appear later in the game. If you have a teenage partner, there are many possibilities for the possibility of others, and there is no way to distinguish between these possibilities.

This poker “incomplete information” character, while having a computer program can not lose in chess and programs that can defeat the grand master of chess, the development of computer poker players can not lose weight has slowed down.

Texas Hold’em

Computer scientist Michael Bowling and his team at the University of Alberta have been playing computer poker for years. Their aim is to solve the relatively simple poker version: the Texas Hold ‘Em head-up limit.

The Texas Hold ‘Em game has four transaction cards, followed by a betting round. Initially, each player flips two cards and creates a secret “hole card” part of the hand. (What is meant by “flop”) is the first three cards can see the front end of a round all players can use. There is another betting round followed by two final rounds (“Turn” and “River”) where one card is seen in each round. If there are several players in the final match featuring the original hidden card and the winner is the player who can assemble the best five hand poker cards using a combination of five community cards and two hidden cards.

In certain cases, bowling and his colleagues are at work, “head-up” only that, two players representing “limits” are fixed bets and are limited in the number of possible salary increases in each round. That means. This limitation greatly simplifies the game. Make players more even more complicated strategic decisions Leaving unlimited variable bets leads to more potential in each round of betting.

Despite this simplification, there are still around 319 billion things that players can still find. Symmetry games can make use of (to replace all spade clubs, or vice versa, to replace) around 13.8 trillion matches remaining weapons after simplifying them to consider the scenario.

Complete the game

Bowling and his team created a computer program that was cleverly divided into manageable sections of 13.8 trillion better scenarios. This program is designed to gradually increase in stages as a strategy according to a recurring strategy gradually closer to the ideal strategy.

For each game scenario, the program tries to make possible strategies. After playing to the end, we will compare the strategy with other scenarios. The program will then update the strategy based on this comparison. Avon’s Future to choose to act in a scenario that might be better chosen than time would avoid greater likelihood of deteriorating behavior. Repeating this will eventually lead to a perfect strategy.

The team ran this program in a large computing group of 200 sophisticated computers for more than two months. Meanwhile, the process of trying and adjusting the strategy has been repeated 1,579 times and the last strategy has become very strong.

This strategy is strong enough to meet the standard of bowling “finishing” the game at its core. If you are a poker player for 200 hours per hour for 12 hours a day for a computer that runs the last strategy, you are at least 95% likely that you cannot distinguish between a real perfect strategy and a computer strategy.

Although Bowling and his team and other computer poker researchers have surpassed the simpler “synthetic” version of poker in the past, the head-up limit, Texas Hold ‘Em, is actually the first game that humans play competitively. This must be solved this way.

Surprising strategy

The paper concludes by showing some interesting aspects of the initial strategy of the pre-flop game from the algorithm. Heads up Texas Holdem starts with “blind” bets. The dealer must enter a small amount of money, and the other player must double the amount before processing the card. The dealer then handles two hidden pocket cards.

At this point, the dealer will start the first betting round in one of three ways. There is no risk of throwing your hands away and losing a small blind person but throwing away money (folding). Put another half of the big bet and submit the decision making process to another player (“Call”). You can add more chips to double the big blind bet (“increase”) after adjusting to the big curtain.

The final strategy of this program is that dealers tend to be quite aggressive in their initial games and others are aggressive. This diagram shows the possibility of drawing (green), calling (blue) or folding (red) the program based on two pocket cards held by the player. The lower left corner of the box represents the card in another setting. Half the upper right card of the same type. On the left is the dealer’s first step. The response to the dealer increases with other players on the right.


Dealers almost always raise their hands, make very few calls and fold their weakest hands. Assuming you are increasing the dealer, the second player will respond to a raise with the right hand and will call most of the other hand to make the next flop and betting round.

Some instructions for poker research include expanding the analysis of games with more players and more complex gambling structures. Bowling and his colleagues also noticed that theoretical analysis of the game following the lines used to learn poker was very helpful in other areas such as airport security and medical diagnostics.