VIP & VIP+
Exclusive Content   Join Now

A new breed ofvideo poker guru

Sep 23, 2003 3:09 AM

Allow me to introduce myself. My name is Elliot Frome. That name may sound familiar to many of you. I am the son of Lenny Frome, the original Guru of video poker. Since my father’s passing more than five years ago, I have continued to sell his books, video tapes and software. Over that time, I have made minor modifications to his works, but it was about 10 months ago that I decided to spend more time devoted to this mission. For the past 15 years, I have toiled designing computer systems for some of America’s largest corporations. It was time to toil for myself.

My dad and I had spent a lot of time discussing his work, so I wasn’t jumping in cold. We had often discussed building a video poker engine that could analyze virtually any pay table and any variation of the game. A few similar programs had been created and were out there for our using, but we wanted one of our very own, one that we knew the inner workings of.

The idea behind this "engine" was that it would play each of the roughly 2.6 million hands and analyze the optimum way to play each hand based on a particular pay table. Even as PC’s were getting faster, the number of hands that had to be analyzed was staggering. How could I get a program to analyze all possibilities accurately and take less than weeks to run?

A few months ago, I was pondering how to finally accomplish this, and it finally hit me. Theoretically, each five-card hand on initial draw can be played 32 different ways (you can hold 0 cards 1 way, 1 card 5 ways, 2 cards 10 ways, 3 cards 10 ways, 4 cards 5 ways and 5 cards 1 way for a total of 32 ways).

So now, we had 32 times 2.6 million hands for more than 80 million hands to analyze. Of course many of the 32 ways had hundreds of thousands of ways to be played, so again, the numbers grew to staggering sizes. Instead of playing each of the post draw hands, I devised a way of categorizing each hand initially dealt and keeping track of what was held and what was discarded. In order to get a 100% accurate analysis, the discards are very important. In this way, I was able to keep the number of hands to be analyzed to the roughly 80 million.

It took me about two weeks to code the program. I started with the basic full pay VP (9/6 Jacks or Better). I was very happy when the first full run of the program executed in only 10 hours. I had some ideas for how to improve this, but 10 hours wasn’t bad. The real question was, was it accurate?

I began to pour through the results and compared them to my dad’s numbers. In most cases, the results were identical. In a few cases, I had some minor differences. I don’t like minor discrepancies. They are very difficult to find the root cause.

I dove deeper into the details and I discovered the reason for the discrepancies. Sometimes hands overlap. For example, a 2-card royal that has a KQ, KJ or JQ may also be a four-card straight with four high cards with a JQKA.

The expected value (EV) of these hands are very similar .597 vs .596. EV’s however are not absolute for a given type of hand, but are averages of all the variations of a hand (we’ll save the details of this for a later column).

So, some four card straights with four high cards actually have an EV higher than the two card royal. Usually, the difference only applies to a few hundred variations and the EV difference is less than .01.

My dad would frequently ignore these relatively infrequent occurrences in order to keep the strategy table easier to memorize. The overall impact was always less than .1% and who would want a strategy table that said a two card royal with a JQ, JK or QK should be played over a four card straight except when the non-honor card that is to be discarded is of the same suit as the royal. The chance for errors would go up considerably and easily eliminate the miniscule advantage of playing 100% perfect.

In the end, my program never differed from my dad’s results by more than 0.1% in total payback and the strategies remained identical. These were results I wanted to see. It wasn’t as important that I could ”˜prove’ my dad’s results as that I was confident I could now analyze any new pay tables or new version of games that were on the market. The wild card version of my engine is in progress.

Of course, what does this all mean to you, the video poker player? It means there’s a new guru in town. Okay, it may take some time for me to earn that title. At least, now I have the tools, and my goal is to be the players pal that my dad was to all of you.

Over the next several months, I expect to begin work on an updated Winning Strategies for Video Poker. Expert Video Poker for Las Vegas was recently updated to include strategies for Bonus Poker. In the meantime, feel free to drop me a line at [email protected] or visit us on our website at http://pages.prodigy.net/kilroydq.