Quote:
Originally Posted by CrazyLond
It doesn't make sense for unrestricted player to ever choose scissors since it starts off losing 40 percent of the time so the optimal strategy for him will be some combination of rock and paper.
The unrestricted player should go paper 2/3 of the time and rock 1/3 of the time. Any counterstrategy by the restricted player will be no more effective than allowing unrestricted player to win 6.6667% more often than restricted player. Any other combo of rock and paper allows a more effective counterstrategy.
The problem with this is that it assumes a nash equilibrium, and I don't see this being NE. NE is defined as neither player having a profitable deviation from the current game state. Yes, player 1's optimal state of [1,0,0] in Game state 1 and [0, 1/3, 2/3] in Game state 2 is the best adjustment
given Player 2's chosen game state, but that doesn't mean Player 2 is in an optimal state himself.
First let's discuss how player 2 is playing, which is [1/3, 2/3, 0].
In Game state 1, player 2 will win 2/3 and tie 1/3, an EV of 2/3
In Game state 2, player 2 will tie 2/9, win 2/9, and lose 5/9, an EV of -1/3.
2/3 * 2/5 - 1/3 * 3/5 = 4/15 - 3/15 = 1/15, as you calculated.
But, player 2 is smart and can figure out that player 1 is playing [1,0,0] 40% of the time, and [0, 1/3, 2/3] 60% of the time. Or:
[2/5, 0, 0] + [0, 1/5, 2/5] = [2/5, 1/5, 2/5].
Can't we maximize by playing:
[1, 0, 0]?
Now we tie w/ his rock 40%, lose to his paper 20%, and beat his scissors 40%. Our EV has jumped to 0.2! So clearly, we aren't in a NE state.
Essentially, the trick for this problem is that the mixed strategy player is trying to exploit the
forced strategy of the restricted player, while the restricted player is trying to exploit
the adjustment the mixed strategy player is making. The analytical formulation is (I think):
R1 + P1 + S1 = 1
R2 + P2 + S2 = 1
P2 = R1
(R1 - 0.4) / 0.6 = S2
P1 / 0.6 = R2
S1 / 0.6 = P2
Last edited by jdr0317; 12-04-2014 at 05:04 PM.