Creating Video Poker Strategies on Computers

A good strategy makes playing video poker accurately much easier. As someone who has been involved in trying to create better strategies since 1995, I can tell you that it is a tedious process.

Bob Dancer is one of the world's foremost video poker experts.  He is a regular columnist for Casino Player, Strictly Slots, and the Las Vegas Review-Journa land has written an autobiography and a novel about gambling.   He provides advice for tens of thousands of casino enthusiasts looking to play video poker.  Bob's website is www.bobdancer.com  Before you can create a good strategy, you first have to decide what IS a good strategy. You want it accurate, of course, but accuracy is NOT the most important criteria. Being simple enough to use is more important than accuracy, but obviously you need to have both accuracy and simplicity. Sometimes the things that make a strategy more accurate, though, make them less simple, and simplifying them to be more user friendly often comes at the cost of accuracy. There is no one-size-fits-all strategy, as beginners need it simplified and more user-friendly, and advanced players want it more accurate. One of the reasons why Liam W. Daily and I created four strategies for each of the games we tackled was that each player could decide for himself what was the right trade-off between accuracy and simplicity.

There have been efforts to automate this strategy-creation process. Tomski's "Video Poker Strategy Master," was the first widely-available program to do this. It was definitely better than figuring out everything by yourself, but it was tedious to use, and some hands were not covered.

Jim Wolf created a penalty-free strategy creation program for "Frugal Video Poker," soon to be out of print, which was an improvement over VPSM (assuming you wanted penalty-free strategies --- if you wanted a glimpse of penalties, VPSM was better), but there were some systematic errors in it. Wolf has announced that a new video poker program, called Wolf Video Poker, is coming. Whether or not this program will make significant changes to the strategy-creation component over Frugal Video Poker is something we'll have to wait and see.

I become involved with what is now called "Video Poker for Winners" after the strategy creation component was in mid-stream. The programmers and mathematicians behind this component are top-notch (they do all of the math and programming behind Action Gaming products like Triple Play, Spin Poker, Chase the Royal, Five Play Multi-Strike, etc.), but they had never created strategies before. And the project turned out to be a LOT more difficult than they anticipated, partly because they weren't expert players.

An expert player understands penalty cards, and this understanding is essential in creating good strategies --- even if you're creating penalty-free strategies. Starting with a K of one suit and a 'JT' of another, where the quote marks indicate the cards are suited with each other, in many games you sometimes hold the KJ and sometimes the 'JT', depending on the value of the other two cards. There are 1,176 different combinations of other cards that can join up with the K'JT', and many of them have to be discarded --- such as cards that make a pair, three of a kind, 4-card straight, 4-card flush, 3-card straight flush (maybe), and inside straights with two or three high cards. In order to group these, you need to have an understanding of various kinds of penalties so you can get an effective weighted average of the relevant hands. You want to end up with KJ > 'JT' or 'JT' > KJ, but you want it to be accurate.

Wolf's program did this to a much greater extent than Tomski's, so I have to assume that somewhere along the way Jim Wolf learned a lot about penalty cards.

Another thing strategies must do is to "group" like combinations in some places, and "separate" similar combinations in others. For example, 'KQ9', 'KJ9', and 'QJ8' are all 3-card straight flushes with two high cards and two insides. In many hands, they have exactly the same values, but as we saw in last week's column, in some games (like 8/5 Bonus and 9/5 Super Double Bonus) 'KQ9', 'KJ9' > AKQJ > 'QJ8' is the correct order. A strategic program that grouped these combinations together will miss this. In other games, such as 9/6 Jacks, it is perfectly acceptable to group these three combinations together. Having the flexibility to identify when to group combinations and when to separate them requires judgment. Probably human judgment.

Rightly or wrongly, I decided that the feature that would make the VPW strategies superior is this type of human judgment. Because of this, I needed to personally create the 100+ strategies necessary for this product. I couldn't evaluate whether the "canned" strategies were the best possible or not unless I knew what they SHOULD be. Since the program allows for virtually infinite flexibility in enter pay schedule changes, I obviously couldn't create the strategy for every possibility. But I could do this for all of the base games that were included, including the various levels of Multi-strike.

So I used every tool available, including both Tomski's and Wolf's programs, plus a lot of WinPoker and everything I've learned about video poker for the last twelve years. (Creating, with a co-author, my own strategy cards and Winner's Guides was an invaluable part of this process.) Each of these sources added something to the mix that the others didn't have. Once I had created the target strategies, I compared them to the strategies created by the program.

Some of the differences between my strategies and the computer-generated strategies were because of errors I had made, and some were based on error's in the computer program algorithm, but unless I had gone through this process beforehand, I couldn't be confident that the strategies were both accurate and simple enough to use. My goal was to have the most accurate penalty-free strategies possible, while making the strategies as user-friendly as possible given their level of complexity, and I'm happy with the result.

For most games, the number of hands played differently by the VPW strategy and the Wolf strategy is small. Either strategy, when used accurately, will suffice for all but the top experts. My goal was to make the VPW strategies simpler to use. Whether or not I succeeded in this will be for others to determine.

Hopefully, if and when players notice errors in the strategies they will report them to me. That way, the strategies in Version Two of VPW (whenever that comes about) will be even better.