A necessary first step is the selection of one or two primary poker games and betting structures to study. Poker is reknown for having an unbounded number of variations and formats, but for research purposes it is necessary to focus our attention on a particular standard. This allows certain foundations to be established and used by everyone in the field. Independent results can then be readily compared, and the body of research can become more directed.
The ``Drosophila melanogaster'' we recommend for poker research is the game of Limit Texas Hold'em with a 2-2-4-4 unit betting structure and blinds of 1 and 2 units. A more detailed description of this game can be found in any of the classic books referenced in the previous section.
There are many reasons to choose this particular variation above other candidates. First, it is the most popular poker game played in casinos and public card rooms, so it is well known to all serious players. This ensures that many human expert players are available for consultation and evaluation of programs and ideas. More importantly, there is also a good source of literature on the game, as described in the previous section.
Moreover, Hold'em has particularly simple logistics and is (arguably) the most strategically complex form of poker among popular games. It is generally accepted that among common variations, Hold'em has the smallest ratio of luck to skill (although all forms of poker have a naturally high element of chance and degree of variance). Partly for these reasons, Hold'em is the game played to determine the World Champion at the annual World Series of Poker Championship.
The recommended betting structure follows the most common pattern seen in North American casinos and card rooms, and the same basic structure is used for low, medium, and high stakes tables. For example, a ``$2-$4'' game has two blinds of $1 and $2, bets and raises of $2 for the first two betting rounds, and $4 for the last two rounds. A high stakes ``$50-$100'' game uses the same format, but with all values scaled up by a factor of 25. Since most betting strategies are dependent only on the size of the bet in relation to the size of the pot, the two games are strategically equivalent (although normally very different in character and in strength of the players).
Another advantage of studying Hold'em is that it is a member of a family of closely related games, called ``flop games''. Each variation in this family uses the same method of community board cards, and differ only in the number of private hole cards and rules for forming hands. This allows generalized strategies for Hold'em to be applied to other variations with limited modifications. This will enable us to see if those principles are indeed transferable, rather than game-specific.
The best alternative choice for poker research is the game of Seven-Card Stud. Top level players generally consider it to be the only other popular game having a strategic complexity in the same class as Hold'em. Again, there is a good supply of expert players and correct literature on the game. Seven-Card Stud is slightly more involved from a technical standpoint, since part of the basic strategy involves tracking all exposed cards and evaluating the partial information exhibited by each opponent's hand. These characteristics can lead to interesting new problems for study, but since they are not fundamental to poker in general, they seem better suited to future research.
Many of the past studies on poker have suffered from a poor choice of variation. Five-Card Draw poker was a natural choice for study in the 1960's and 1970's, since it was one of the most popular forms of poker at that time, but it has since lost much of it's appeal. Moreover, Five-Card Draw is a rather simple game from a strategic perspective, when compared to Hold'em or Seven-Card Stud. This is unfortunate for researchers because certain ad hoc rule-based approaches to the game can be fairly successful without properly addressing certain poker fundamentals. Such methods should prove to be less adequate for the more challenging games of Hold'em and Seven-Card Stud, allowing more robust techniques to emerge as the best methods for high-performance results.