Here is a simplified, “stripped-down” version of poker as an instructional classroom game. Although Stripped-Down Poker is extremely simple, it nevertheless provides an excellent illustration of a number of topics: signaling, bluffing, mixed strategies, the value of information, and Bayes’s Rule. The authors begin with a description of Stripped-Down Poker: how to play it, what makes it an interesting classroom game, and how to teach its solution to students. They describe how signaling, bluffing, and so forth emerge naturally as important features of the game and then discuss possible applications of this game-theoretic model to real-world interactions, such as litigation, tax evasion, and domestic or international diplomacy. They also suggest modifications of the game either for use in class or as student exercises. For reference, they conclude with a brief history of game-theoretic treatments of poker.

Effective classroom games should meet three criteria. First, they should he simple enough to be readily understood by students. Second, they should be engaging enough to capture students’ attention and interest. Third, they should be rich enough not only to describe a particular bit of theory, but also to illustrate real-world phenomena. Our classroom game, Stripped-Down Poker, satisfies these three criteria and has served as a memorable learning experience to students in our undergraduate economics classes on game theory. From one simple game, students can obtain insights about signaling games, bluffing, converting from extensive to strategic form, mixed-strategy equilibrium, Bayes’s Theorem, and the value of private information. The game is especially timely given the recent surge of interest in poker in popular culture: we find that a growing number of our students play poker or watch it on television.

PLAYING THE GAME

We like to play Stripped-Down Poker as a game between the instructor and a student volunteer. To introduce the game to your class, emphasize that you will be playing a simplified version of poker for real money, and that the volunteer will be responsible for paying any losses out of his or her pocket. One way to heighten the students’ interest level is for you to pull out money from your own wallet.
Next, describe the nature of the game. Produce the special deck for the game, explain that it contains only four kings and four queens, and choose a student from the front row to verify the deck’s contents to the rest of the class. (This student might also serve us the dealer when you actually play the game.) Explain that only the professor receives a card during the game; the student player does not. After each player puts an ante of $1 into the pot, you (the professor) will randomly draw a card from the dealer, privately observe what it is, and decide whether to fold or to bet. If you fold, the game ends, and the student wins the pot, gaining your $1 ante. When betting, place another dollar in the pot, after which the student has the option to either fold or call. A fold by the student ends the game, this time with you winning the pot and gaining the student’s $1 ante. Calling requires the student to add another dollar to the pot and requires you to reveal your card. You win the pot with a king and lose the pot with a queen. The pot at this stage is $4, so the winner nets $2 from the loser.
After describing the rules, ask for a student volunteer. Repeat the game 10-20 times. After several rounds, you might ask for a new volunteer to play against you, either because the first volunteer no longer wants to play or because you want to give several different students the opportunity to play. We find that students get actively engaged in the game even when they are not playing: they watch attentively, shout out advice to the student player, and generally exhibit high energy levels.
At first glance, many students believe the game is fair. This na”ive belief serves two important purposes. First, it produces more willing volunteers. Second, it makes the game especially effective for teaching purposes: over the course of repeated play, the students begin to observe that the professor is winning money on average from the student volunteers. This strongly motivates the students for the subsequent analysis of the game.
For maximum effect, you should attempt to play according to the equilibrium (minimax) strategy, which is to always bet when you have a king and to bet one-third of the time when you have a queen. We find it best to use an external randomization device, such as the second hand of one’s watch. For example, if the number of seconds in the minute is between 0 and 20, you would bet with a queen, but if it is between 20 and 60, you would fold with a queen. It is crucial that you look at your watch in the same way even if you have drawn a king; otherwise, glancing at your watch could serve as a “tell” to the students that you are holding a queen and thus destroy your bluffing advantage.
When you execute this strategy correctly, the expected payoff is $1/3 to the professor (-$1/3 to the student), no matter what action the student chooses. After 10-20 repeated rounds of the game, we recommend beginning the discussion by asking the students whether the game is fair. Some students may still think the game is fair and that you were just lucky to have won a positive amount from the students after 20 rounds. Others, after seeing the professor’s profits, may now believe the game is unfair but be uncertain about the source of the unfairness. As a cliffhanger, we often choose to schedule this game for the last 20 minutes of one class period and challenge the students at the end of class to see if they can figure out whether it is a fair game before the next class period, when we do a thorough analysis.

ANALYTICAL EXPOSITION

You can lead the classroom discussion in a number of directions. What you decide to focus on depends on the level of your course and the ideas that you most wish to illustrate by using the game. The following subsections focus on separate potential topics. First, we help the students develop an intuitive understanding of the value of bluffing. Second, we show students how to represent the game in both extensive and strategic form. Third, we solve for the mixed-strategy Nash equilibrium of the game, with probabilistic bluffing and calling. Fourth, we illustrate the use of Bayes’s Rule by considering the probability of the professor having a king, given that she has bet. Finally, we use Stripped-Down Poker as a very simple signaling game to illustrate signaling equilibrium concepts. These subsections appear in a logical progression, but we have tried to make them modular. Feel free to select the combination of topics that most appropriately meets the needs of your class.

How Should the Players Play?

We prefer to start our discussion of the game by asking the students some questions to get them thinking and to gauge their understanding. First, you might take a quick survey to see how many students believe that the game is fair. You might ask one student to explain why he thinks the game is fair and ask another student why she thinks it is unfair. Then promise the students a definitive answer in the analysis to come.
Second, you can ask the students what they would do if they were playing in the professor’s position. Many students believe that the professor’s optimal strategy would be to always bet on kings and always fold on queens. When a student suggests this strategy, ask the class what the student player’s best response would be. With a little thought, they should be able to see that if the professor always folds with queens, then the only time the student gets to move is when the professor must have a king, and therefore the student should always fold. Note that with this pair of strategies, the professor earns an expected value of $0. But we have already observed through playing the game that the professor earned quite a bit of money from the students!
Next, you can ask what the professor’s optimal strategy would be if the student always folded. If there is no chance of the student calling, then there is no reason to avoid betting with queens: the professor can bluff the student into folding even if the professor holds a poor card. The professor’s optimal strategy is thus to always bet and earn $1 when the student folds. But if the professor always bets, what is the student’s optimal strategy? If the professor always bets, then the student should expect a 50 percent chance of a king and a 50 percent chance of a queen if he chooses to call. This means if the student calls, his expected payoff is $0. By contrast, if he folds, he is guaranteed to lose $1, so his optimal strategy is to always call.
If the student always calls, then what is the professor’s optimal strategy? Then the professor wins by betting with kings but loses by betting with queens, so her optimal strategy is to bet with kings and fold with queens, which takes us back to our starting point. By this intuitive discussion, we manage to demonstrate to students that there is no mutual best response, and hence no pure-strategy Nash equilibrium. This leads us to a discussion of probabilistic bluffing.
A third question to ask the students is why the professor kept looking at her watch before deciding whether to bet or fold. A few students will likely notice this behavior and ask about it during game play. This is a good time to share with the students that you did indeed use your watch as a randomizing device, although you should not say exactly what probabilities you used until you formally solve the game with the class. By now, though, the students should see intuitively that the professor cannot get away with bluffing on queens if she does it all the time, so she might want to consider bluffing randomly with a probability of less than one.
If your class is not a game-theory course, you might wish to stop with this intuitive level of analysis and skip ahead to the real-world applications of bluffing that we describe later. For an undergraduate course in game theory, we prefer to continue with a formal analysis of the game, as we describe in the next two subsections.

Extensive and Strategic Form

Stripped-Down Poker offers a good example for comparing extensive and strategic forms of the same game. This is a sequential-move game, so one can begin by drawing its extensive form (see Figure 1). You might ask the students how they would draw the game tree and complete it on the board with their help. We find that most students do not instantly recognize the initial move “by nature” to deal a card to the professor, particularly because we use this game as a very early example of a game of incomplete information. Another common problem is for students to forget to include an information set indicating that the student has to move without knowing what card the professor holds.
Because the students’ information set contains two nodes, They cannot use backward induction to solve the game. This leads us naturally to the idea of converting the game to strategic form and looking for a Bayesian Nash equilibrium using a game matrix.
This is a great opportunity to reinforce a concept that many of our students find difficult. We suggest beginning by asking the students, “How many possible strategies does each player have?” Some students mistakenly believe each player has two strategies, because students easily confuse strategies with actions. Others mistakenly believe that each player has four strategies, because they fail to notice that the student cannot condition his strategy on the card type. In fact, the professor has four possible strategies (two actions at each of two decision nodes) and the student has two strategies (two actions at a single information set).
Next, write the 4 ? 2 game matrix on the board (see Figure 2), but allow the students to help you fill in the expected payoffs. Ask the class, for example, “What is the expected payoff to the professor if she always bets and the student always calls?” Give them a minute to think about the question, and then ask for a volunteer to explain the answer. Your better students may be able to tell you the answer right away: if not, point out that for this pair of strategies, the possible payoffs are either (–2, +2) when the professor has a king or (+2, –2) when the professor holds a queen. These two cards occur with equal probability, so the expected payoff to each player is $0in the upper-left cell of the payoff matrix.
With that example under their belts, you can ask the students to calculate the rest of the entries in the payoff matrix. To engage the entire class in thought (rather than just the few bright students who always volunteer answers), you might ask the class to take a few minutes to make the calculations at their desks, in consultation with their classmates. You can stroll around the room to monitor their progress and answer questions that individual students might be more comfortable asking privately than in front of the entire class. After a few minutes, reconvene the class and ask student volunteers to help you fill out the rest of the payoff matrix on the board. The full matrix should be as shown in Figure 2.
Next, ask the students whether there are any dominated strategies. Note that the strategies FB and FF are both strictly dominated for the professor by BB but that BB and BF are not dominated. After eliminating the strictly dominated strategies , find each player’s best responses (see How Should the Players Play? section) and observe that there is no pure-strategy Nash equilibrium.

Mixed-Strategy Nash Equilibrium

Once pure-strategy Nash equilibria are ruled out, it is time to look for a mixed-strategy equilibrium.
Let p represent the probability that the professor plays BB (i.e., that she Bluffs).
Let q represent the probability that the student plays Call.
For a player to be willing to mix over two strategies, she must be indifferent between them; otherwise, she would play her preferred pure strategy. This must be true for each player, and this insight enables us to find the equilibrium mixing probabilities. To keep each opponent indifferent, we have the following two equations:
[Formula Omitted]
[Formula Omitted]
Thus the mixed-strategy Nash equilibrium of this game is
(1/3 BB + 2/3 BF; 2/3 Call+ 1/3 Fold).
Or, rewriting the professor’s strategies with a more intuitive interpretation,
(1/3 Bluffing + 2/3 Truthtelling; 2/3 Call + 1/3 Fold).
At this point, you might ask the students to guess what you were doing when looking at your watch during game play. You can then discuss how you used your external randomization device and take the opportunity to discuss the importance of keeping one’s opponent guessing. Ask the students, for example, why you looked at your watch even when you held a king (as described above).
Next, you can compute the expected value of the game to each player. The professor’s expected payoff in equilibrium is a weighted average of the four possible payoffs in Figure 3;
(2/9)(0) + (1/9)(1)+ (4/9)(1/2) +(2/9)<0) = 1/3.
Because the game is zero-sum, the student’s expected payoff is simply the opposite: -1/3.
We finally have a definitive answer for the students on whether the game is fair, and the answer is a resounding “no.” In equilibrium, the professor can expect to win $1/3 from the student for each round of the game.(FN6) The source of the unfairness is information asymmetry. Unlike the student, the professor has private information, and she can use bluffing to press this advantage.
To demonstrate that the advantage comes exclusively from the private information (e.g., not merely from the order of moves), you might ask your students to compare these results to those of a related game with no information asymmetry. There are two obvious ways to do this: either deal the card face up so that both players see it, or deal it so that neither player sees it. In the first case (Face-Up Stripped-Down Poker), the student should clearly call the bet only if the card is a queen, and given that strategy, the professor should clearly bet only if the card is a king. Thus, in equilibrium, the professor wins $1 if the card is a king and loses $1 if the card is a queen, so the expected payoff is $0. In the second case (Blind Stripped-Down Poker), the professor must bet before knowing the value of the card. In this case, the professor has a single information set when she moves, so the strategic form is a 2 ? 2 matrix instead of a 4 ? 2 matrix. This game has a pure-strategy equilibrium where the professor always bets, the student always calls, and the equilibrium payoff to each player is again $0. Thus, when both players have the same information, it is a fair game after all.

Bayes’s Rule

Stripped-Down Poker is ideal for illustrating the use of Bayes’s Rule to update probabilities. After discussing the fact that you actually played your equilibrium mixed strategy during the game, always belling with kings but betting only one-third of the time with queens, ask the students the following question: “Given that I have bet, what is the probability that I am holding a king?” Typical student responses are 1/2 or 2/3, neither of which is correct. Point out that these beliefs crucially determine the student’s optimal decision whether to bet or fold. If the student underestimates the probability that the card is really a king given that the professor has bet, he may call more often than is optimal, which the professor could then exploit (see endnote 6).
Now the students should be motivated to learn Bayes’s Rule. The rule can be illustrated quite simply for this game by using a square diagram (see Figure 4). The left half of the square represents the state of the world in which the professor gets a king (half the area of the square; probability is 1/2); the right half represents the state in which she gets a queen (the remaining half of the area and probability). The square is divided vertically into thirds to represent the possible mixing realizations: in the left half the professor bets in each third; in the right half she bets in only one of the thirds.
Each of the diagram’s six segments represents a probability of 1/6. In four of the segments (probability = 4/6 = 2/3), the professor bets, and in three of those, she has a king–that is, 3/4 of her bets are made with a king and 1/4 with a queen. This visual, geometric representation helps students understand conditional probabilities and Bayes’s Rule, concepts they generally find confusing and unintuitive.
With the diagram in hand, the corresponding algebra for the probability that the professor bets, and for the conditional probability that she has a king, is easier for students to follow: let P(K) and P(Q) represent the probabilities of being dealt a king and a queen, respectively, and let P(B) be the probability that the professor bets. Using standard notation to denote conditional probabilities, we first obtain P(B):
P(B) = P(B|K)P(K) + P(B|Q)P(Q)
= (1)(1/2) + (1/3)(1/2)
= 1/2+ 1/6
= 2/3.
Then we use Bayes’s Rule to obtain P(K|B):
P(K|B) = P(K and B)/P(B)
= P(B|K)P(K)/[P(B|K)P(K) + P(B|Q)P(Q)]
= (1)(1/2)/[(1)(1/2) + (1/3)(1/2)]
= (1/2)/(2/3)
= 3/4.
So, given that the professor has bet, the probability of a king is not 1/2, not 2/3, but actually 3/4. With this knowledge, we can check to make sure that the student really does want to play his proposed equilibrium strategy.(FN7) If the student folds, his payoff is -$1. If he calls when the professor bets, he has a probability of 3/4 of losing $2 and a probability of 1/4 of winning $2, for an expected payoff of -$1. So he is indeed indifferent between calling and folding and thus should be willing to follow his equilibrium mixture.

Poker as a Signaling Game

Because real-world poker clearly involves strategic signaling and screening, we like to use Stripped-Down Poker as a simple introduction to the topic of dynamic games of incomplete information. Most games designed to illustrate signaling equilibria, such as the well-known Beer/Quiche game (Cho and Kreps 1987) or the Attacker/Defender game used in Dixit and Skeath (2004, section 9.5), are slightly more complicated. They involve a first mover with two types and two possible actions and a second mover with two possible actions no matter what the first mover does. Our game simplifies signaling even further by eliminating the second mover’s ability to act if the first mover chooses to fold. This gives us a 4 ? 2 strategic-form game matrix, in contrast to the 4 ? 4 strategic form of these other standard signaling examples. Although this eliminates some of the richness of a full signaling model, it also yields a big pedagogical advantage. We find that students understand games like Beer/Quiche much better after they have been introduced to fundamental signaling concepts in Stripped-Down Poker.
Game-theory textbooks introduce students to a taxonomy of three types of signaling equilibria: separating equilibria, pooling equilibria, and semi-separating equilibria. Students usually do not find these concepts to be terribly intuitive, in part because the standard signaling games are conceptually difficult and complicated to solve in strategic form. Semi-separating equilibria are particularly subtle and difficult to follow because they involve only a partial separation of types. We find that most students do not really even grasp the meaning of “partial separation of types” without the benefit of a simple example, and Stripped-Down Poker provides a very good one.
Why does Stripped-Down Poker have a semi-separating equilibrium? Recall that a pooling equilibrium involves both types taking the same action. In Stripped-Down Poker, this means that the professor would always bet with both kings and queens,(FN8) so the student could not distinguish which card the professor holds. In contrast, a separating equilibrium involves each type taking a separate action. In Stripped-Down Poker, this means that the professor would bet with kings but fold with queens,(FN9) so the student could infer from a bet that the card was a king. In our discussion of how to play, we ruled out both the separating strategy and the pooling strategy, showing that neither could be part of an equilibrium. The equilibrium is in fact semi-separating, because observing a bet by the professor does give some information about the card type, but it does not completely identify the card type. In particular, observing a bet by the professor allows the student to update the probability of a king from 1/2 to 3/4, but not all the way to 1. Semi-separating equilibria typically involve mixed strategies.(FN10)
One difference between Stripped-Down Poker and standard signaling models is that in the latter, one type of first mover typically wants to signal his type truthfully, whereas one type does not. For example, in the Beer/Quiche game, the strong type wants the would-be bully to know that he really is strong, but the weak type wants the bully to think that he is really a strong type. By contrast, in Stripped-Down Poker, the professor never wants the student to know her card type. If she has a king, she would like the student to think she has a queen, so that the student will call and she can win $2 instead of $1. If she has a queen, she would like the student to think she has a king, so that the student will fold and she can win $1 instead of losing $2. When discussing signaling models, it may be worth pointing out this difference.
Real-world versions of poker, such as Texas Hold’em, have some of this flavor. A player with a weak hand may sometimes wish to bluff to get others to fold, and a player with a strong hand may sometimes wish to “slow play” to keep others in the game and win more money from them. But real-world poker is also much richer and more complicated, with multiple rounds of betting and multiple rounds of cards being dealt. Sometimes real poker players do want to signal truthfully. For instance, a player may think he likely has the current best hand with three of a kind but worries that if others stay in the betting, one of them may eventually draw a card that completes a flush and thus beat him. By betting aggressively, the player can drive out competition and win the hand before such a reversal of fortune. Poker may also involve screening, such as when an early player deliberately bets to see how a later player will respond; if the later player raises my bet, for example, I may conclude that he likely has a very strong hand, and I will therefore fold.
In a small class where many students are quite familiar with more complicated versions of poker, real-world poker examples might lead to useful discussions. Instructors who have time and interest may wish to refer to some of the wealth of anecdotal examples found in popular books on poker.(FN11) To avoid turning off students not interested in Texas Hold’em, we usually prefer to keep such discussion to a minimum and instead move on to the applications in the next section.

APPLICATIONS

Stripped-Down Poker is an extremely simple two-player game, yet it can roughly model intriguing real-life interactions. Stripped-Down Poker is isomorphic to any signaling-type game where one of the signaler’s actions could be thought of as ending the game. This terminal action can be thought of as a fold, whereas the other action(s) would represent a bet. The first two examples below are expanded in the appendix with numerical exercises suitable for student problem sets; both have semi-separating equilibria just like those of the classroom game.
Litigation is one very nice example.(FN12) A plaintiff sues for damages in civil court. The defendant has private information about whether or not he is really guilty, but the plaintiff only has an assessment of the probability of guilt. For simplicity, assume that if the case goes to trial, then the court will find out the truth, finding for the plaintiff if and only if the defendant is guilty. Before going to trial, the defendant has the option of offering a settlement to the plaintiff. If he offers a generous settlement, the plaintiff will accept and drop the case. Because the game would end here, this action may be thought of as a fold by the defendant. On the other hand, the defendant could bet by offering a stingy settlement, to which the plaintiff may respond by either accepting the stingy settlement (folding) or rejecting it (calling). If losing in court is costly to the plaintiff, then straightforward play by the defendant (making a stingy offer if and only if he is innocent) would induce the plaintiff to always accept the stingy offer. But the defendant will sometimes make the stingy offer even when he is guilty (a bluff), and therefore the plaintiff must sometimes go to court in response to the stingy offer, even though it seems to signal a strong hand for the defendant (i.e., that he is innocent).
A similar example involves tax evasion and auditing.(FN13) Consider a waitress reporting income from tips to the Internal Revenue Service (IRS). Both parties may have the same priors about the probability of earning high versus low tip income, but only the waitress knows how much was actually realized in tips. The waitress may report the truth and pay the appropriate taxes (a fold), or she may underreport high tip income and pay less in taxes (a bet). When the IRS receives a high tax payment, it does not suspect any problem and takes no action (i.e., the game ends). When it receives a low tax payment, the IRS chooses either to audit the waitress (call) or to accept the waitress’s reporting (fold). If found guilty of underreporting her tips, the waitress will be responsible for her true tax liability and perhaps an additional penalty. Both parties incur costs in the case of an audit, regardless of whether the waitress is found innocent. As both of these examples show, the game need not be zero-sum for the students to benefit from the insights of Stripped-Down Poker. Interesting efficiency issues (e.g., “What are the social costs of equilibrium auditing?”) can also arise in these games.
Other applications relate to politics at all levels, from international to domestic (household) conflicts. A dictator suspected of harboring weapons of mass destruction may comply in full with international inspections (a fold) or may obstruct such inspections to maintain his threat of power at the risk of military intervention (a bet). A schoolboy required to complete homework before playing with friends may actually do his homework first (a fold) or may simply tell his parents that it is done (a bet), possibly gaining more time to play at the risk of being caught.
These applications also present an opportunity for the professor to discuss with the students the question of choosing an appropriate economic model. Depending on one’s willingness to simplify each of the above situations, one might model the actions available to the first mover as binary, multiple yet discrete, or continuous. The binary case fits the poker model exactly; the agent literally has only two options available, which are then precisely equivalent to a fold and a bet in Stripped-Down Poker. However, real-life situations often offer a range of intermediate alternatives, and sometimes even a continuum of them. The waitress may report an intermediate level of tip income. The schoolboy may do part of his homework or do it sloppily. The dictator may come clean about certain weapons programs but remain ambiguous about the status of others. You might encourage students to extend this model to make it more realistic: when the number of actions available to the first mover is more than two but still discrete and relatively small, Stripped-Down Poker may be extended to include these possibilities. We consider this possibility in the next section.

EXTENSIONS AND EXERCISES

One of Stripped-Down Poker’s greatest virtues is its simplicity. However, if more complex examples are desired, the game lends itself very well to a host of extensions to challenge your students.
First, and perhaps most interesting, one might ask, “How would we have to change the composition of the deck to make this a fair game?” We find this to be a challenging extra-credit problem for our students to consider. The answer turns out to be a deck comprising 3/8 kings and 5/8 queens, so the “fair” game can be conveniently played by replacing one of the four kings in the original deck with a fifth queen. A sketch of the solution to this problem can be found in exercise 1 of the appendix.
Second, one might add other types of cards to the deck. For example, a version of Stripped-Down Poker could be played in which the deck contains kings, queens, and jacks. The first mover draws a card, privately views it, and either bets or folds. The second mover may then fold or call as before. The king wins for the professor, the jack wins for the student, and the queen splits the pot between the two players. This game has an 8 ? 2 strategic-form game matrix with one pure-strategy equilibrium(FN14) and is tractable enough to be an excellent out-of-class exercise to test the students’ understanding of converting imperfect-information games from extensive to strategic form.(FN13)
Third, one might change the information dynamics by having each player draw a card. For a player to win the pot in a showdown, she must hold a card of higher value than her opponent; if both players hold cards of the same value, they split the pot. This arrangement eliminates the first player’s information advantage and thereby creates a fair game. To understand this point about removing the information advantage, students may find it enlightening to solve for the two pure-strategy Nash equilibria of this game, each of which has an expected payoff of $0 to each player. There are a number of possible variations to consider. We recommend exercise 9.8 of Dixit and Skeath (2004) as one interesting and tractable example.(FN16) It is also possible to think about extending the game to have multiple rounds of betting, more than two players, and/or more than two card types to address questions of first-mover or last-mover advantage in poker and other information games (see, e.g., Karlin and Restrepo 1957). These extensions may, however, be beyond the reach of all but the best undergraduates.

HISTORICAL TREATMENTS OF POKER IN GAME THEORY
Stripped-Down Poker uses well-established game-theoretic models of poker to illustrate the importance of asymmetric information. The French mathematician ?mile Borel presented a simplified poker model in 1938 that illustrated the rationality of bluffing (Borel 1938). John von Neumann formulated two similar poker models a few years later (von Neumann and Morgenstern 1944). For simplicity, both Borel’s and von Neumann’s models feature risk-neutral players and a continuum of possible hands.(FN17,18) For a detailed analysis of the poker models of Borel and von Neumann, see Ferguson and Ferguson (2003).
Borel’s poker model works as follows: After each player antes $1 into the pot, each draws a hand represented by a draw from a uniform distribution on [0, 1]. After the players privately observe their own hands, the first player moves by either folding or raising. If the first player folds, he forfeits his ante to the second player and ends the game. If the first player raises, he adds an additional dollar to the pot. Then the second player moves by either folding or calling. If the second player folds, she forfeits her ante to the first player and ends the game. If she calls the first player’s bet, she adds an additional dollar to the pot, and both players reveal their hands in a “showdown.” At this point, the player with the highest hand wins the pot, for a net gain of $2 (the ante plus the additional bet).
Von Neumann’s second poker model is similar to Borel’s, the only difference being that the first player’s option to fold is replaced with the option to “check”–that is, to stay in the game without raising the stakes. When the first player checks, the second player does not have an opportunity to move, and the holder of the higher card wins the other player’s ante.
Borel’s model yields equilibrium bluffing, in the sense that sometimes a player bets despite knowing that she may have the weaker hand, and her opponent sometimes responds by folding when full information would have revealed that he had the stronger hand.(FN19) Von Neumann’s model results in even stronger bluffing: the first player bets in equilibrium even with the worst possible hand. Although interesting for very advanced analytical courses, the models of Borel and von Neumann would have limited pedagogical value in most undergraduate courses. First, students find it much more engaging to play a poker game with a real deck of cards, rather than a continuous distribution of hands. Second, these models are complicated to analyze: Borel’s model would probably be grasped only by mathematically mature undergraduates, and von Neumann’s model is probably best left for graduate courses.
Since the time of von Neumann, researchers have analyzed many other models of poker and other bluffing games, including the analytical models of Bellman and Blackwell (1949), Kuhn (1950), Nash and Shapley (1950), Isaacs (1955), Karlin and Restrepo (1957), Goldman and Stone (1960), and Friedman (1971). The last paper analyzed an abstract game, framed in terms of actual poker, of which our game is a special case. Burns (2005) developed a suite of four simplified poker games to study judgment and decision making in the laboratory, but even these are too complicated for in-class analytical solution. Of these, Kuhn’s poker model is the most interesting for pedagogical purposes: it is essentially von Neumann’s model with exactly three possible hands.
Stripped-Down Poker is an even simpler variation on Borel’s model, in which only the first player gets a card and there are only two card types in the deck. We were inspired to develop this classroom game by an example in the textbook by Gardner (2003), who called his version of the game “Liar’s Poker.” Thomas (1984) gave a similar presentation in his textbook, calling the game “Simplified Poker.” The only difference in rules between these two games and our classroom game is that the other authors require the first mover to bet when the high card is drawn (instead of giving him the option to bet or fold), an assumption that yields the same equilibrium behavior. Rapoport et al. (1997) conducted laboratory experiments with the game. Their game differs from ours only in the bet size and thus the bluffing and calling mixtures. Our contribution has been to transform the model from a simple textbook exercise into an engaging classroom game, vividly and concretely illustrating multiple learning objectives. We believe this is the simplest possible version of poker and, as discussed previously, also the simplest possible example of a signaling game.

CONCLUSION

We have presented a simple classroom game based on poker that provides significant instructional benefits with a minimum of materials and preparation. Stripped-Down Poker is thoroughly engaging in a classroom setting, and its analysis is accessible to students. The game lends itself to a number of interesting variations and applications which can be explored by students both in class and in out-of-class exercises. We find it a useful, effective, flexible, and memorable teaching tool. And it’s fun!

Try this sites: