You are on page 1of 5

Case Problem: Dealer’s Absorbing State Probabilities in Black Jack

1. The first step is to create the transition probability matrix for the dealer’s hand. The states are defined
by the value of the dealer’s hand after the up card is dealt, and the values the dealer’s hand may assume
after the down card is revealed, and after taking hits. The absorbing states are 17, 18, 19, 20, 21, and
bust. According to the house rules, once the dealer’s hand takes on one of these values she quits taking
hits and either pays or collects the bets that are on the table. The states S12, S13, S14, S15, and S16
represent what are called soft hands; they include an ace that may be played as 1 or 11. There are no
S17, S18, S19, S20, and S21 states because the dealer plays these the same as hard hands of the same
value.

There are 27 states so, due to space restrictions, the transition matrix that is shown below is printed in
3 parts. All of the rows and a subset of the columns are shown in each part.

A S12 S13 S14 S15 S16 2 3 4 5 6


A 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
S12 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
S13 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
S14 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
S15 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
S16 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
2 0.0000 0.0000 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769
3 0.0000 0.0000 0.0000 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769
4 0.0000 0.0000 0.0000 0.0000 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769
5 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0000 0.0000 0.0000 0.0000 0.0000
6 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
7 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
8 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
9 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
10 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
11 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
12 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
13 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
14 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
15 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
16 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
17 0 0 0 0 0 0 0 0 0 0 0
18 0 0 0 0 0 0 0 0 0 0 0
19 0 0 0 0 0 0 0 0 0 0 0
20 0 0 0 0 0 0 0 0 0 0 0
21 0 0 0 0 0 0 0 0 0 0 0
Bust 0 0 0 0 0 0 0 0 0 0 0

CP - 1
7 8 9 10 11 12 13 14 15 16 17
A 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769
S12 0.0000 0.0000 0.0000 0.0000 0.0000 0.3077 0.0769
S13 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.3077 0.0769
S14 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.3077 0.0769
S15 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.3077 0.0769
S16 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.3077 0.0769
2 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077 0.0000 0.0000 0.0000 0.0000 0.0000
3 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077 0.0000 0.0000 0.0000 0.0000
4 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077 0.0000 0.0000 0.0000
5 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077 0.0000 0.0000
6 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077 0.0769
7 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.3077
8 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769
9 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769
10 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769
11 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769 0.0769
12 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769 0.0769
13 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769 0.0769
14 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769 0.0769
15 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769 0.0769
16 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0769
17 0 0 0 0 0 0 0 0 0 0 1
18 0 0 0 0 0 0 0 0 0 0 0
19 0 0 0 0 0 0 0 0 0 0 0
20 0 0 0 0 0 0 0 0 0 0 0
21 0 0 0 0 0 0 0 0 0 0 0
Bust 0 0 0 0 0 0 0 0 0 0 0

Chapter 16

18 19 20 21 Bust
A 0.0769 0.0769 0.0769 0.3077 0.0000
S12 0.0769 0.0769 0.0769 0.0769 0.0000
S13 0.0769 0.0769 0.0769 0.0769 0.0000
S14 0.0769 0.0769 0.0769 0.0769 0.0000
S15 0.0769 0.0769 0.0769 0.0769 0.0000
S16 0.0769 0.0769 0.0769 0.0769 0.0000

CP - 2
2 0.0000 0.0000 0.0000 0.0000 0.0000
3 0.0000 0.0000 0.0000 0.0000 0.0000
4 0.0000 0.0000 0.0000 0.0000 0.0000
5 0.0000 0.0000 0.0000 0.0000 0.0000
6 0.0000 0.0000 0.0000 0.0000 0.0000
7 0.0769 0.0000 0.0000 0.0000 0.0000
8 0.3077 0.0769 0.0000 0.0000 0.0000
9 0.0769 0.3077 0.0769 0.0000 0.0000
10 0.0769 0.0769 0.3077 0.0769 0.0000
11 0.0769 0.0769 0.0769 0.3077 0.0000
12 0.0769 0.0769 0.0769 0.0769 0.3077
13 0.0769 0.0769 0.0769 0.0769 0.3846
14 0.0769 0.0769 0.0769 0.0769 0.4615
15 0.0769 0.0769 0.0769 0.0769 0.5385
16 0.0769 0.0769 0.0769 0.0769 0.6154
17 0 0 0 0 0
18 1 0 0 0 0
19 0 1 0 0 0
20 0 0 1 0 0
21 0 0 0 1 0
Bust 0 0 0 0 1 The Q matrix is defined as the first 21 rows
(through state 16) and the first 21 columns (through state
16) of the transition matrix. We must now compute I – Q and (I – Q)-1. We did this using Excel and
Excel’s matrix inversion procedure described in the chapter appendix. To find the absorption
probabilities we multiply (I – Q)-1 times the R matrix. The R matrix is a 21 x 6 matrix. It is identified
as the first 21 rows and the last 6 columns of the transition matrix. We show (I – Q) -1R below.

Probability of Dealer’s Finishing State Given Starting State

17 18 19 20 21 Bust
A 0.1308 0.1308 0.1308 0.1308 0.3616 0.1153
S12 0.1510 0.1510 0.1510 0.1510 0.1510 0.2450
S13 0.1455 0.1455 0.1455 0.1455 0.1455 0.2725
S14 0.1400 0.1400 0.1400 0.1400 0.1400 0.3000
S15 0.1346 0.1346 0.1346 0.1346 0.1346 0.3272
S16 0.1292 0.1292 0.1292 0.1292 0.1292 0.3541
2 0.1398 0.1349 0.1297 0.1240 0.1180 0.3536
3 0.1350 0.1305 0.1256 0.1203 0.1147 0.3739
4 0.1305 0.1259 0.1214 0.1165 0.1112 0.3945
5 0.1223 0.1223 0.1177 0.1131 0.1082 0.4164
6 0.1654 0.1063 0.1063 0.1017 0.0972 0.4232
7 0.3686 0.1378 0.0786 0.0786 0.0741 0.2623

CP - 3
8 0.1286 0.3593 0.1286 0.0694 0.0694 0.2447
9 0.1200 0.1200 0.3508 0.1200 0.0608 0.2284
10 0.1114 0.1114 0.1114 0.3422 0.1114 0.2121
11 0.1114 0.1114 0.1114 0.1114 0.3422 0.2121
12 0.1035 0.1035 0.1035 0.1035 0.1035 0.4827
13 0.0961 0.0961 0.0961 0.0961 0.0961 0.5196
14 0.0892 0.0892 0.0892 0.0892 0.0892 0.5539
15 0.0828 0.0828 0.0828 0.0828 0.0828 0.5858
16 0.0769 0.0769 0.0769 0.0769 0.0769 0.6154

This matrix shows the probabilities for the ending values of the dealer’s hand given any of the starting
states identified by each row. For instance, if the dealer has a 6 as an up card, the probability of
finishing with 17 is .1654, the probability of finishing with 18 is .1063 and so on. The probability of
the dealer busting is .4232. For this reason, black jack players recommend not taking a hit with a hand
of 12 or better. Why should the player take a chance of busting when the dealer has a high probability
of busting?
Chapter 16

2. For this situation another row and column must be added to the transition matrix to account for the soft
17 state, S17. When the dealer has to stay on S17, there was no reason to differentiate soft 17 and hard
17. So only the state, 17 was used. For space reasons, we do not show the new transition matrix here.
But, after computing (I – Q)-1R, the absorbing state probabilities are given below.

Probability of Dealer's Finishing State Given Starting State

17 18 19 20 21 Bust
A 0.0575 0.1432 0.1432 0.1432 0.3740 0.1389
S12 0.0829 0.1625 0.1625 0.1625 0.1625 0.2669
S13 0.0823 0.1562 0.1562 0.1562 0.1562 0.2929
S14 0.0813 0.1500 0.1500 0.1500 0.1500 0.3189 S15 0.0801
0.1438 0.1438 0.1438 0.1438 0.3448 S16 0.0786 0.1377
0.1377 0.1377 0.1377 0.3704 S17 0.3422 0.1114 0.1114
0.1114 0.1114 0.2121
2 0.1301 0.1365 0.1313 0.1257 0.1196 0.3567 3 0.1263
0.1320 0.1271 0.1218 0.1162 0.3767 4 0.1224 0.1273
0.1228 0.1179 0.1126 0.3971
5 0.1184 0.1229 0.1184 0.1138 0.1089 0.4177
6 0.1148 0.1148 0.1148 0.1103 0.1057 0.4395 7
0.3686 0.1378 0.0786 0.0786 0.0741 0.2623 8
0.1286 0.3593 0.1286 0.0694 0.0694 0.2447
9 0.1200 0.1200 0.3508 0.1200 0.0608 0.2284
10 0.1114 0.1114 0.1114 0.3422 0.1114 0.2121

CP - 4
11 0.1114 0.1114 0.1114 0.1114 0.3422 0.2121 12 0.1035 0.1035
0.1035 0.1035 0.1035 0.4827 13 0.0961 0.0961 0.0961 0.0961
0.0961 0.5196 14 0.0892 0.0892 0.0892 0.0892 0.0892 0.5539
15 0.0828 0.0828 0.0828 0.0828 0.0828 0.5858
16 0.0769 0.0769 0.0769 0.0769 0.0769 0.6154

This matrix shows the probabilities for the ending values of the dealer’s hand given any of the starting
states identified by each row for the case when the dealer hits soft 17. For instance, if the dealer has a 6
as an up card, the probability of finishing with 17 is .1148, the probability of finishing with 18 is .1148,
and so on. The probability of the dealer busting is .4395.

3. Mathematicians have shown that when the dealer stays on soft 17 it is better for the player. We can
provide the following argument using the finishing values for the dealer’s hand shown in part 2 above.
Note that when the dealer gets S17 and hits it, the finishing values and probabilities for the dealer’s
hand are

Ending Value Probability

17 .3422
18 .1114
19 .1114
20 .1114
21 .1114
Bust .2121

If the player has 17, the probability of a tie is .3422, the probability of losing is 4(.1114) = .4456, and
the probability of winning is .2121. So, when the player has 17, she will do better (a tie) if the dealer
stays on S17. Similarly, if the player has 18, 19, 20, or 21, she will do much better if the dealer stays
on S17. She will always win.

But, if the player has 12, 13, 14, 15, or 16, she will be better off if the dealer hits soft 17. This is
because when the dealer busts she will win (if the dealer does not hit S17 she loses). And, this will
happen with probability .2121. But, the player will lose with probability 1 - .2121 = .7879 when the
dealer hits S17. This small probability of the player winning with 12, 13, 14, 15, and 16 when the
dealer hits S17 is not enough to offset the large increase in the probability of the player winning with
17, 18,19, 20, and 21 when the dealer stays on S17.

CP - 5

You might also like