An Answer Page for Exercise 15

Exercise 15

Re-do the computations from the previous section using View #2 instead of View #1. That is, find the house edge for combined don't pass + free odds bets assuming that any game can end in a tie.

 

We want to compute the house edge for players who make only don't pass bets and free odds bets.

We imagine that we are playing a series of N craps games.
Since we are assuming View #2,   a push ends the current game.

We begin each game by putting the amount B on the don't pass line. If a point is established, we bet as much as we can in free odds. Whenever a 12 is rolled on the come out, we say that the game has ended in a tie; and, the next roll of the dice begins a new game.

We use the following definitions:

N = the number of craps games we play [ some large fixed number ]
B = the dollar amount bet on the don't pass line at the start of each game
 
p = probability to win a don't pass bet
according to view #2   =   949 / 1980     [ computed in the View #1 section ]
 
L = probability to lose   =   1   -   P( tie )   -   P( win )
  = 1   -   ( 1 / 36 )   -   ( 949 / 1980 )   =   244 / 495
 
α = size of the free odds bet when the point is 4 or 10
β = size of the free odds bet when the point is 5 or 9
γ = size of the free odds bet when the point is 6 or 8

For later use, we note that :

p - L = ( 949 / 1980 )   -   ( 244 / 495 )
  = ( 949 / (20)(99) )   -   244 / ( 5 * 99 )
  = ( 949   -   976 ) / 20 * 99   =   -27 / 1980   =   -3 / 220

  Let f = the "odds factor"
    = the number preceding "X" in the casino's free odds phrase
  1X odds     f = 1,     2X odds     f = 2,     etc.

The first thing we shall do is determine the probability of establishing a point in any one game.

P( point ) = P( 4 )   +   P( 5 )   +   P( 6 )   +   P( 8 )   +   P( 9 )   +   P( 10 )
  = 24 / 36   =   2 / 3

Thus,   P( no point )   =   1   -   2 / 3   =   1 / 3

In playing N games, we'd expect to have no point N / 3 times.

In any one game, the point probabilities are:

  P( point is 4 )   =   P( point is 10 )   =   3 / 36
  P( point is 5 )   =   P( point is 9 )   =   4 / 36
  P( point is 6 )   =   P( point is 8 )   =   5 / 36

So, in N games, we would expect

  4 to be the point approximately 3N / 36 times
  5 to be the point approximately 4N / 36 times
  6 to be the point approximately 5N / 36 times
  8 to be the point approximately 5N / 36 times
  9 to be the point approximately 4N / 36 times
  10 to be the point approximately 3N / 36 times

Let's use a table to summarize what we've found so far.

Point Line Bet Odds Bet Sum of Bets Nbr of Times We
Bet That Sum
Product of
Previous Two Columns
none B 0 B N / 3 BN / 3
4 B α B + α N / 12 ( N / 12 )( B + α )
5 B β B + β N / 9 ( N / 9 )( B + β )
6 B γ B + γ 5N / 36 ( 5N / 36 )( B + γ )
8 B γ B + γ 5N / 36 ( 5N / 36 )( B + γ )
9 B β B + β N / 9 ( N / 9 )( B + β )
10 B α B + α N / 12 ( N / 12 )( B + α )

To find the total amount we'd expect to bet in N plays, we add the values in the last column of the above table.

Let   T   =   Total amount of money bet in N plays

T = ( BN / 3 ) + 2 *( N / 36 ) [ 3 ( B + α ) + 4 ( B + β ) + 5 ( B + γ ) ]
  = ( 12 BN / 36 ) + ( 2N / 36 ) [ 12B + 3α + 4β + 5γ ]
  = ( 12BN / 36 ) + ( 24BN / 36 ) + ( N / 18 ) ( 3α + 4β + 5γ )
  = BN + ( N / 18 ) ( 3α + 4β + 5γ )

The average amount bet in N plays is   T / N   =   B + ( 1 / 18 ) ( 3α + 4β + 5γ )

Since free odds bets do not change the expectation, we can compute the average loss from the expectation:

E = expectation   =   expected winnings
  = Bp   +   ( - B ) * L   +   0 * P( tie )   =   B * ( p - L )
E = B * ( - 3 / 220 )   →   expected loss   =   3B / 220

Putting the above two pieces together, we find:

H   =   house edge   =   average loss / average bet

  H  =     3B / 220
B + ( 1 / 18 ) ( 3α + 4β + 5γ )
 

To finish, we need to find α, β, and γ.

As we've seen before:

  Point   =   4 or 10 payoff odds for the don't pass free odds bet   =   1 / 2
  Point   =   5 or 9 payoff odds for the don't pass free odds bet   =   2 / 3
  Point   =   6 or 8 payoff odds for the don't pass free odds bet   =   5 / 6

Let's use   "FDO yes"   to mean that Full Double Odds are in effect   and   "FDO no"
to mean that Full Double Odds are not in effect.

  Point is 4 or 10   "FDO no"   and   y   =   payoff odds   =   1 / 2  
  α   =   fB / y   =   2fB  
 
  Point is 5 or 9   "FDO no"   and   y   =   payoff odds   =   2 / 3  
  β   =   fB / y   =   (3fB) / 2  
 
  Point is 6 or 8   and "FDO yes"   γ   =   3B  
  Point is 6 or 8   and "FDO no"   γ   =   fB / y   =   (6fB) / 5  

If full double odds do not apply then   3α + 4β + 5γ   =   6fB + 6fB + 6fB   =   18fB

Substituting this into the above equation for H gives us

 
H   =   3 / 220
1 + ( 1 / 18B ) ( 18fB )
=
3 / 220
1 + f
 

 

If full double odds do apply then

  B = line bet   =   2k   for some integer k
 
  α = size of the free odds bet when the point is 4 or 10
  = fB / ( 1/2 )   =   2fB   =   2f * 2k   =   4fk
 
  β = size of the free odds bet when the point is 5 or 9
  = fB / ( 2/3 )   =   3fB / 2   =   (3f/2) * 2k   =   3fk
 
  γ = size of the free odds bet when the point is 6 or 8
  = 3B   =   3 * 2k   =   6k
 
  and   3α + 4β + 5γ =   12fk   +   12fk   +   30k   =   24fk   +   30k   =   k * ( 30 + 24f )
 

Substituting these values into the formula for H gives us

 
H   =   3 / 220
1 + ( 1 / 18B ) [ k * ( 30 + 24f ) ]
=  
3 / 220
1 + ( 6 / 36) * [ 5 + 4f ]
 
 
 
    =   ( 3 / 220 ) * 6
6 + 5 + 4f
=
9
110 [ 11 + 4f ]
 
 

 

For case 2b in the following table,   we need   f = 2.
f = 2   and   "FDO yes"       H   =   ( 9 / 110 )/ 19     0.430622%.

 

Combined Don't Pass + Free Odds Using View #2
  case 1 odds = 1X   f = 1   H   =   ( 3 / 220 ) / ( 1 + 1 ) 0.681818%  
  case 2a:     2X   "FDO no"   f = 2   H   =   ( 3 / 220 ) / ( 1 + 2 ) 0.454545%  
  case 2b:     2X   "FDO yes"   f = 2   H   =   ( 9 / 110 ) / 19 0.430622%  
  case 3 odds = 3X   f = 3   H   =   ( 3 / 220 ) / ( 1 + 3 ) 0.340909%  
  case 4 odds = 4X   f = 4   H   =   ( 3 / 220 ) / ( 1 + 4 ) 0.272727%  
  case 5 odds = 5X   f = 5   H   =   ( 3 / 220 ) / ( 1 + 5 ) 0.227273%  
  case 10 odds = 10X   f = 10   H   =   ( 3 / 220 ) / ( 1 + 10 ) 0.123967%  
  case 20 odds = 20X   f = 20   H   =   ( 3 / 220 ) / ( 1 + 20 ) 0.06493506%  
  case 100 odds = 100X   f = 100   H   =   ( 3 / 220 ) / ( 1 + 100 ) 0.01350135%