MATH 335: Roulette example

If you place a one dollar bet on red in roulette you have an 18/38 probability of netting $1 and a 20/38 probability of netting $-1. The expected value of a single play is $-2/38 = $-.053 and the variance is essentially 1 ($^2). If you make this bet n times your expected net winnings is -.053*n with a variance of n and a sd of sqrt(n).

First we use R to simulate n=100 plays:


box <- c(rep(1,18), rep(-1,20))
x <- replicate(10000,sum(sample(box,100,repl=T)))
prob <- sum(x>0)/10000 # to estimate the 
                        # probability of netting a positive amount.
hist(x)
prob


Here are the theoretical values, computed with CLT, of being ahead after n plays.
       n      -.053n      sqrt(n)       z*    P(W>0)=P(Z>z*)
 ------------------------------------------------------------
       50       -2.65        7.07       0.37      .3539
      100        -5.3       10.00       0.53      .2981
     1000       -53         31.6        1.67      .0469
   10,000       530        100          5.30      0
1,000,000    53,000       1000         53.00      0^2
-------------------------------------------------------------
Here is R code to make these computations.
n <- c(50,100,1000,10000,1000000)
sqrtn <- sqrt(n)
expval <- -.053*n
zstar <- .053*n/sqrt(n)
prob <- 1-pnorm(zstar)



cbind(n,expval,round(sqrtn,1),round(zstar,2),round(prob,4))

Here is the link to roulette information: click here