I have a statistics question. Let's say i have a game in which there are three possible…
- User
I have a statistics question. Let's say i have a game in which there are three possible outcomes: Win, Lose and Draw. Whenever i win i add one token, when i lose i remove one and when i Draw nothing happens. The number of token cannot go below 0 and when they reach 5 i win. let's say the probabilty for Win, Lose and Draw are p, q, and r. What is the probability to win the game?
- Assistant
To compute the probability of winning the game given the described conditions, we can use a recursive approach. We need to find the probability of reaching 5 tokens before reaching 0 tokens. Let's denote ( P_n ) as the probability of winning the game starting with ( n ) tokens. Specifically, we need to find ( P_1 ).
Here are the details:
-
Boundary conditions:
- ( P_0 = 0 ) (if you have 0 tokens, you can't win since the next loss will keep it at 0).
- ( P_5 = 1 ) (if you have 5 tokens, you already won the game).
-
Recursive formulation:
- For ( 1 \leq n < 5 ), the probability ( P_n ) can be determined using the probabilities of moving to adjacent states: [ P_n = p \cdot P_{n+1} + q \cdot P_{n-1} + r \cdot P_n ]
Since ( r ) is the probability of a draw which leaves the state unchanged, we can simplify the equation:
[ P_n = p \cdot P_{n+1} + q \cdot P_{n-1} + r \cdot P_n ]
By rearranging the terms, we get:
[ P_n - r \cdot P_n = p \cdot P_{n+1} + q \cdot P_{n-1} ]
Simplification:
[ (1 - r) \cdot P_n = p \cdot P_{n+1} + q \cdot P_{n-1} ]
[ P_n = \frac{p \cdot P_{n+1} + q \cdot P_{n-1}}{1 - r} ]
To solve this, we will work our way from the boundary conditions ( P_0 ) and ( P_5 ):
- Given ( P_0 = 0 )
- Given ( P_5 = 1 )
Now, we can recursively calculate ( P_n ) for ( n = 4, 3, 2, 1 ):
First, starting with ( P_4 ):
[ P_4 = \frac{p \cdot P_5 + q \cdot P_3}{1 - r} = \frac{p \cdot 1 + q \cdot P_3}{1 - r} ]
Next, for ( P_3 ):
[ P_3 = \frac{p \cdot P_4 + q \cdot P_2}{1 - r} ]
Now for ( P_2 ):
[ P_2 = \frac{p \cdot P_3 + q \cdot P_1}{1 - r} ]
Finally for ( P_1 ):
[ P_1 = \frac{p \cdot P_2 + q \cdot P_0}{1 - r} = \frac{p \cdot P_2 + q \cdot 0}{1 - r} = \frac{p \cdot P_2}{1 - r} ]
This gives us a system of equations:
- ( P_4 = \frac{p + q \cdot P_3}{1 - r} )
- ( P_3 = \frac{p \cdot P_4 + q \cdot P_2}{1 - r} )
- ( P_2 = \frac{p \cdot P_3 + q \cdot P_1}{1 - r} )
- ( P_1 = \frac{p \cdot P_2}{1 - r} )
These equations can be solved iteratively or using matrix algebra to find the probabilities of each state until we get the value for ( P_1 ).
Given these probabilities, ( P_1 ) will be the probability of reaching 5 tokens (winning the game) starting with 1 token. In a practical scenario, if you know the values of ( p, q, ) and ( r ), you can carry out these computations numerically.
-