Video Poker Playing Strategies

(Courtesy of Steve Jacobs)


Max-EV

Min-Cost

Min-Cost-Royal

Min-Risk

Best-Shot-Royal

Kelly Optimal

Which Strategy ?

Virtual Payoffs


* * * * * * * * * * * * * * * * * * * * * *


Max-EV (aka Max-ER)

The 'Max-EV' strategy always chooses the highest EV option when making hold and discard decisions in the play of a video poker hand. This strategy maximizes EV per hour.


* * * * * * * * * * * * * * * * * * * * * *


Min-Cost (aka Min-Loss)

The Min-Cost (aka Min-Loss) strategy seeks to maximize dollars-won for a fixed number of dollars-lost, or minimize dollars-lost (cost) for a fixed number of dollar-won.


* * * * * * * * * * * * * * * * * * * * * *


Min-Cost-Royal (aka Min-Loss)

The 'Min-Cost-Royal' strategy gives the smallest average loss between royal flushes. Equivalently, this strategy allows the player to 'pay' the least total dollar amount per royal flush. The Min-Cost-Royal strategy tends to be very close to the strategy which minimizes risk (or RoR).

The Min-Cost-Royal strategy is also very close, but not identical to the best-shot strategy and the strategy that I call Min-Risk. The Min-Risk strategy maximizes the probability of surviving until you reach a specific dollar target. With the best-shot strategy, you care only about hitting a royal and you don't care (directly) about how many dollars you end up with as a result.

The 'Min-Cost-Royal' strategy, which minimizes the average loss between royals, is a static strategy. The strategy maximizes EV for the royal payoff that gives a breakeven game. It wins the most money per royal jackpot. In contrast, the Max-EV strategy wins most quickly rather than extracting the most dollars from the jackpot.


9/6 JoB Min-Cost-Royal Strategy

99.5103% ER (with 800 Unit Royal)

Final Hand Payoff % Hit Cycle % Return
Royal Flush  4879.97  0.0028  35939.23  2.7157
Straight Flush  250  0.0111  8986.44  0.5564
Quads  125  0.2355  424.61  5.8878
Full House  45  1.1485  87.07  10.3367
Flush  30  1.1122  89.92  6.6729
Straight  20  1.1308  88.43  4.5234/b>
Trips  15  7.4165  13.48  22.2495
Two Pair  10  12.8919  7.76  25.7839
High Pair  5  21.2737  4.70  21.2737
Garbage    54.7769    


1    975.9932 Royal Flush
2    50.0000 Straight Flush
3    25.0000 4/Kind
4    22.4487 4/royal
5    9.0000 Full House
6    6.0000 Flush
7    4.3025 trips
8    4.0000 Straight
9    3.5551 4/str-flush (0 holes)
10  2.5957 two pair
11  2.3825 4/str-flush (1 hole)
12  1.6582 suited KQJ
13  1.6545 suited QJT
14  1.5694 suited AQJ/AKJ/AKQ
15  1.5664 suited KJT/KQT
16  1.5365 pair (AKQJ)
17  1.4761 suited AJT/AQT/AKT
18  1.2170 4/flush
19  0.8723 unsuited KQJT
20  0.8237 pair (T98765432)
21  0.8085 unsuited QJT9
22  0.7447 unsuited JT98
23  0.7313 3/str-flush (1 hole, 2 high)
24  0.7266 3/str-flush (0 holes, 1 high)
25  0.6809 4/straight (0 holes, 0 high)
26  0.6366 3/str-flush (2 holes, 2 high)
27  0.6312 3/str-flush (1 hole, 1 high)
28  0.6254 3/str-flush (0 holes, 0 high)
29  0.6188 suited QJ
30  0.6028 suited KJ/KQ
31  0.5957 unsuited AKQJ
32  0.5864 suited AJ/AQ/AK
33  0.5362 3/str-flush (2 holes, 1 high)
34  0.5319 4/straight (1 hole, 3 high)
35  0.5304 3/str-flush (1 hole, 0 high)
36  0.5153 unsuited KQJ
37  0.5079 suited JT
38  0.5024 unsuited QJ
39  0.4941 suited QT
40  0.4893 unsuited KJ/KQ
41  0.4810 Jack
42  0.4800 suited KT
43  0.4771 unsuited AJ/AQ/AK
44  0.4770 Queen
45  0.4728 King
46  0.4714 suited AT
47  0.4712 Ace
48  0.4349 3/str-flush (2 holes, 0 high)
49  0.3602 (draw 5 cards)


* * * * * * * * * * * * * * * * * * * * * *


Min-Risk (aka Min-Bankroll)

Min-Risk - For any starting bankroll size and any 'target' bankroll, this strategy maximizes the probability of reaching the target before going broke. This is equivalent to having the smallest bankroll requirement for a given probability of going broke before reaching the target.

The Min-Risk (aka Min-Bankroll) strategy gives the smallest bankroll requirement for a given risk of ruin, or equivalently gives the least risk of ruin for a given bankroll size.

The Min-Risk strategy is found by pretending that the payoffs are all 'stretched' according to the formula which represents an equal risk parameter for all payoffs. If the actual payoffs were set to the values of the virtual payoffs, we would get a breakeven game. What this boils down to is that minimizing risk is just like pretending that the payoffs were 'risk adjusted' and the strategy optimized for the virtual payoffs, resulting in a breakeven game. For neg-EV games, this process results in virtual payoffs that are larger than the real payoffs, causing the Min-Risk strategy to 'try harder' to get big payoffs. In contrast, pos-EV games result in virtual payoffs that are smaller than the real payoffs, causing the Min-Risk strategy to 'avoid' big payoffs. So, to reduce risk in pos-EV games, we play a less agressive strategy as if the royal was worth significantly less than its true value. In neg-EV games, minimizing risk requires us to do just the opposite -- pretend the royal payoff is larger than it really is, and play more agressively.

Once one becomes used to thinking in terms of virtual payoffs, it becomes obvious that minimizing risk means 'play an overly cautious strategy if you have the advantage, play agressively if the house has the advantage'.


9/6 JoB Min-Risk Strategy

99.5352% ER (with 800 Unit Royal)

Final Hand Payoff % Hit Cycle % Return
Royal Flush  4739.34  0.0026  38059.21  2.4905
Straight Flush  252.55  0.0111  9019.16  0.5600
Quads  125.62  0.2360  423.79  5.9286
Full House  45.07  1.1502  86.94  10.3689
Flush  30.03  1.1051  90.49  6.6373
Straight  20.01  1.1280  88.66  4.5146
Trips  15.01  7.4342  13.45  22.3119
Two Pair  10  12.9150  7.74  25.8353
High Pair  5  21.3528  4.68  21.3528
Garbage    54.6650    


1    947.8674 Royal Flush
2    50.5096 Straight Flush
3    25.1244 4/Kind
4    21.8523 4/royal
5    9.0149 Full House
6    6.0062 Flush
7    4.3098 trips
8    4.0025 Straight
9    3.5764 4/str-flush (0 holes)
10  2.5974 two pair
11  2.3932 4/str-flush (1 hole)
12  1.6326 suited KQJ
13  1.6295 suited QJT
14  1.5505 suited AQJ/AKJ/AKQ
15  1.5440 suited KJT/KQT
16  1.5372 pair (AKQJ)
17  1.4501 suited AJT/AQT/AKT
18  1.2170 4/flush
19  0.8723 unsuited KQJT
20  0.8244 pair (T98765432)
21  0.8085 unsuited QJT9
22  0.7447 unsuited JT98
23  0.7323 3/str-flush (1 hole, 2 high)
24  0.7281 3/str-flush (0 holes, 1 high)
25  0.6809 4/straight (0 holes, 0 high)
26  0.6371 3/str-flush (2 holes, 2 high)
27  0.6322 3/str-flush (1 hole, 1 high)
28  0.6269 3/str-flush (0 holes, 0 high)
29  0.6172 suited QJ
30  0.6012 suited KJ/KQ
31  0.5957 unsuited AKQJ
32  0.5847 suited AJ/AQ/AK
33  0.5367 3/str-flush (2 holes, 1 high)
34  0.5319 4/straight (1 hole, 3 high)
35  0.5313 3/str-flush (1 hole, 0 high)
36  0.5153 unsuited KQJ
37  0.5067 suited JT
38  0.5025 unsuited QJ
39  0.4925 suited QT
40  0.4894 unsuited KJ/KQ
41  0.4809 Jack
42  0.4786 suited KT
43  0.4772 unsuited AJ/AQ/AK
44  0.4770 Queen
45  0.4727 King
46  0.4712 Ace
47  0.4698 suited AT
48  0.4354 3/str-flush (2 holes, 0 high)
49  0.3602 (draw 5 cards)


* * * * * * * * * * * * * * * * * * * * * *


Best-Shot-Royal

For a given bankroll size, the 'Best-Shot-Royal' strategy maximizes the probability of hitting a royal before going broke. This strategy does not change with the payoff for the royal. For JoB 9/6, if a single bet is played and replayed until the player either hits a royal or runs out of credits, this strategy gives a 1 in 1038.68 chance of eventually hitting a royal. This is equivalent to getting a 50/50 chance of royal from a starting bankroll of 720 units. Each additional 720 units added to the initial bankroll will reduce the probability of failure by another factor of two.

If you have a limited bankroll and want to have the best chance of hitting a royal with your limited funds, then the Min-Risk strategy comes very close to achieving that goal. The Best-Shot-Royal gives a 1 in 1038.68 chance, while the Min-Risk strategy gives a 1 in 1043.82 chance. The Max-EV strategy isn't terribly far behind, giving a 1 in 1054.47 chance for play from a single wager to survive until you hit a royal (730.6 units for a 50/50 shot at a royal).

For JoB 9/6, whether favorable or unfavorable, whether full coin play or short coin play, the Best-Shot-Royal strategy is the same. The Min-Cost-Royal has this same property. These represent two different defintions of 'best way to play for royal flushes'. One maximizes probability of success, while the other minimizes average cost (assuming you have unlimited funds so that you never have to stop play before hitting a royal).

A player who starts with a bankroll of one unit will have a one in 1038.6796 chance of surviving to hit a royal flush, so the probability of reaching the royal is 1 / 1038.6796. The values in the strategy list are all proportional to the probability of surviving to hit a royal flush. To convert one of these values to an exact probability, divide the value by 1038.6796. For example, holding a high pair has a virtual EV of 1.5349 units, so drawing to a high pair gives the player a probability of 1.5349 / 1038.6796 = 0.00147774 of eventually hitting a royal before the bankroll drops below its current level.

Another way to interpret the virtual EV numbers is to think of them as describing how good the draw is, compared to a final payoff of one unit. Drawing to a high pair is 1.5349 times as good as a one unit payoff, and a 4/royal draw is 23.7749 times as good as a one unit payoff, where the phrase 'X times as good' means that the probability of success is X times as much as the probability of success provided by a one unit payoff. This probability is in terms of surviving to hit a royal before the bankroll drops below its current level.


9/6 JoB Best-Shot-Royal Strategy

99.4972% ER (with 800 Unit Royal)

Final Hand Payoff % Hit Cycle % Return
Royal Flush  5193.4  0.00285  35136.30  2.9561
Straight Flush  244.19  0.01116  8960.207  0.5451
Quads  123.57  0.23534  424.9138  5.8161
Full House  44.83  1.14795  87.11155  10.2919
Flush  29.93  1.11502  89.68411  6.6741
Straight  19.97  1.12680  88.74684  4.5007
Trips  14.99  7.41080  13.49382  22.2110
Two Pair  10  12.8859  7.760411  25.7594
High Pair  5  21.2457  4.706845  21.2457
Garbage    54.8185    


1    1038.6796 Royal Flush
2    48.8386 Straight Flush
3    24.7133 4/Kind
4    23.7749 4/royal
5    8.9654 Full House
6    5.9856 Flush
7    4.2856 trips
8    3.9942 Straight
9    3.5028 4/str-flush (0 holes)
10  2.5919 two pair
11  2.3550 4/str-flush (1 hole)
12  1.7144 suited KQJ
13  1.7096 suited QJT
14  1.6221 suited AQJ/AKJ/AKQ
15  1.6173 suited KJT/KQT
16  1.5349 pair (AKQJ)
17  1.5334 suited AJT/AQT/AKT
18  1.2142 4/flush
19  0.8714 unsuited KQJT
20  0.8221 pair (T98765432)
21  0.8075 unsuited QJT9
22  0.7437 unsuited JT98
23  0.7284 3/str-flush (1 hole, 2 high)
24  0.7226 3/str-flush (0 holes, 1 high)
25  0.6799 4/straight (0 holes, 0 high)
26  0.6348 3/str-flush (2 holes, 2 high)
27  0.6283 3/str-flush (1 hole, 1 high)
28  0.6219 suited QJ
29  0.6214 3/str-flush (0 holes, 0 high)
30  0.6060 suited KJ/KQ
31  0.5953 unsuited AKQJ
32  0.5899 suited AJ/AQ/AK
33  0.5344 3/str-flush (2 holes, 1 high)
34  0.5314 4/straight (1 hole, 3 high)
35  0.5274 3/str-flush (1 hole, 0 high)
36  0.5150 unsuited KQJ
37  0.5112 suited JT
38  0.5022 unsuited QJ
39  0.4972 suited QT
40  0.4891 unsuited KJ/KQ
41  0.4834 suited KT
42  0.4810 Jack
43  0.4770 Queen
44  0.4769 unsuited AJ/AQ/AK
45  0.4749 suited AT
46  0.4728 King
47  0.4712 Ace
48  0.4331 3/str-flush (2 holes, 0 high)
49  0.3600 (draw 5 cards)

* * * * * * * * * * * * * * * * * * * * * *


Kelly Optimal

The Kelly Optimal VP strategy differs from both Max-EV and Min-Risk. In some sense, the Kelly strategy is 'between' these two -- less agressive than the Max-EV strategy, but more risky than the Min-Risk strategy. If a Kelly player is constrained to bet much less than the optimal Kelly fraction, then the strategy which maximizes log(bankroll) (and equivalently, geometric mean of the outcome) will approach the Max-EV strategy. The means that Max-EV is appropriate for well-bankrolled players who have bankrolls so large that they cannot find machines that allow them to play as 'large' as they would like. But, if the Kelly player is constrained to bet more than the optimal Kelly fraction, the best 'constrained' strategy appears to approach the Min-Risk strategy as the bet fraction increases from 1X toward 2X Kelly. This means that a Kelly player with an insufficient bankroll should probably choose to use a Min-Risk strategy.


* * * * * * * * * * * * * * * * * * * * * *


Virtual Payoffs

I've developed a method called 'virtual payoffs' that I use to compute optimal strategies. The math behind it is a bit complicated, but the idea is really quite simple. What I do is pretend that we're playing a game that is slightly different than the actual game. This is done by pretending that the payoffs are different than their actual values, so that 'virtual' values are used in place of each real payoff. Then, I use a VP program to find the best strategy for a game that uses these virtual payoffs, by maximizing 'virtual EV' instead of maximizing 'actual EV.' The big trick here is to find a formula that turns real payoffs into virtual values that correctly represent the desired objective.

It would require a lot of highly technical posts to give a detailed description of how this works, so I won't try to do that right now, but I can give several examples of virtual payoff tables that I've used to find some optimal strategies. For these examples, I'll use a 9/6 JoB game with a royal jackpot of 1300 units. This gives the game a healthy EV for the player so that RoR can be used for one of the examples.

The following table shows virtual payoffs that apply to the 9/6 JoB game. For the Max-EV strategy, the actual payoffs for the game are used. I've rounded the virtual payoffs for this table, but when I compute optimal strategies the virtual payoffs are specified to 14 digit accuracy. These numbers are all very precisely defined.

9/6 JoB Virtual Payoffs

Hand Max-EV Min-RoR bs_RF mc_RF Min-cost Max-Royal
Royal Flush  1300  1003.5  1038.7  975.99  1277.9       1
Straight Flush  50  49.492  48.839  50  49.166       0
Quads  25  24.875  24.713  25  24.591       0
Full House  9  8.9850  8.9654  9  8.8638       0
Flush  6  5.9937  5.9856  6  5.9149       0
Straight  4  3.9975  3.9942  4  3.9489       0
Trips  3  2.9987  2.9971  3  2.9660       0
Two Pair  2  1.9996  1.9990  2  1.9830       0
High Pair  1  1  1  1  1       0


The optimal strategy for each of these cases is found by plugging the virtual payoffs into any VP program that computes a max-EV strategy. I don't know if any commercial VP programs will allow non-integer payoffs to be entered, but my own program supports this, and it is very useful for studying alternate strategies. If the commercial programs don't allow this, you might try multiplying all payoffs by 100 or 1000 and rounding to approximate the virtual payoffs.

For 9/6 JoB there is a simpler trick for finding the best-shot(royal) strategy. Changing the royal payoff to 1038, while leaving all other payoffs at the actual values, just happens to give the correct RoRBR strategy. The same strategy results from any royal payoff between 1022 and 1043. However, this kind of trick doesn't work for 8/5 JoB or for 10/7 DB, and I haven't tried it for other games.

Here a quick summary of what each of these strategies is trying to do:

Min-RoR minimizes overall risk-of-ruin, which is equivalent to maximizing the probability that the player will play forever without ever going broke. The 50/50 bankroll for this strategy is 1661 units, which gives the player a 50% chance of playing forever without going broke. The max-EV strategy gives a 50/50 bankroll of 1670 units.

The bs_RF column is best-shot(royal) strategy, which maximizes the probability of hitting a royal before going broke. This strategy has a 50/50 bankroll of 720 units, which gives the player a 50% chance of hitting a royal before going broke. This bankroll is smaller than the 50/50 bankroll for Min-RoR strategy, because the task of hitting a royal is easier than the task of playing forever. The max-EV strategy gives a 50/50 bankroll 731 units when trying to hit a royal before going broke.

The mc_RF column is min-cost(royal) strategy, which minimizes the average loss that occurs between royal flushes. When the royal payoff if 975.99 units, royals occur just often enough to pay their own cost, making the game exactly breakeven. By comparison, the average loss between royals when playing the Max-EV strategy is 984.299 units.

The Min-cost column is the strategy that maximizes the average number of units returned per unit consumed by losses. This views the game as if you were playing at a casino on the border between two countries and using the VP game to exchange currency. If the VP machine accepted only U.S. dollars but made all payoffs in Canadian dollars, then this strategy would give the player the best possible exchange rate. For this game, each dollar paid back to the player costs only 98.297651 dollars that are paid to the machine in lost wagers, on average. This strategy can also be viewed as a 'best winning streak' strategy. If you start with one unit and collect your net winnings, but re-play only the original unit and keep playing until you eventually lose, then this strategy maximizes your average payoff from these winning streaks. By comparison, the Max-EV strategy has an average cost of 98.297656 units lost per unit returned to the player. This strategy is almost identical to Max-EV for this game, and this tends to hold true for most games.

The Max-Royal strategy tries to hit royals at all costs. This strategy would discard a dealt straight-flush in order to try for a royal instead. This would be the right play to make on the last few hands of a tournament, if your only hope for winning a prize is to hit a royal flush. Otherwise, this strategy is a great way to burn through a bankroll in a hurry.


* * * * * * * * * * * * * * * * * * * * * *


Which Strategy ?

Max-EV is best for cases where 'time' (or the number of hands played) is used to measure progress. Risk based strategies tend to work best when time isn't relavent, and the player measures progress solely on the basis of current bankroll. Cost based strategies are different than either of these, and measure progress according to dollars lost in exchange for dollars won.


Some 9/6 JoB Numbers

Royal Flush = 1300 Units

Strategy ER % Royal Cost Royal Shot 10% ROR
Bankroll
Max-EV  100.953    984.30  1042.58  5644.60
Min-Cost  100.953    984.31  1042.59  5644.77
True-Log-Opt.  100.947    980.94  1040.30  5583.20
Best-Shot-Royal  100.920    976.66  1038.68  5518.57
Min-Risk  100.907    976.06  1039.10  5515.68
Min-Cost-Royal  100.902    975.99  1039.40  5517.75


vpFREE Home Page

vpFREE Acronyms

vpFREE Links


All rights reserved - vpFREE since 2001