|
|
In the coin-matching
game described above, the optimum strategy (for either player, as
this is a symmetrical game) is to stack the coins at random, half
of them heads and half tails. It is clear that if A, for example,
stacked his coins with more heads than tails, B could do better
than break even (in the long run) by stacking his coins with more
tails than heads; in fact, if B put all of his coins so as to come
up tails, he would be assured of a profit. Similarly, if A weights
his stack toward heads, B can profitably weight his toward tails.
(Of course if B guesses A's strategy wrong, A will win.)
However, if A aligns his stack
on a 50-50 basis, the game will result in a draw (in the long run),
no matter how B places his coins. (The reader should verify this
for himself.) If A (say) does
not adopt his optimal strategy (half and half) and instead weights
his stack in one direction or the other, he is, in essence, gambling
that he can outguess B. Use of the optimal strategy can guarantee
A no worse than an even split (the "fair" result) in the
long run.
Application to poker. When a
"game theory situation" arises in poker, there is similarly
an optimum strategy for each player. Just as in coin-matching a
strategy consists of taking different actions (placing a coin as
heads or tails) on a percentage basis, so in poker does a strategy
consist of making a betting decision on a percentage basis. Some
strategical questions at poker which can be analyzed on a game-theory
basis are: How often to bluff; how often to call a possible bluff;
how often to bet into a potential high-low call; how often to call
high-low without an "immortal" in each direction. |