Award Date


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Committee Member

Wolfgang Bein

Second Committee Member

Laxmi Gewali

Third Committee Member

Yoohwan Kim

Fourth Committee Member

John Minor

Fifth Committee Member

Shahram Latifi

Number of Pages



The majority of video poker games have been strongly solved and optimal pure strategies that maximize equity for the player are well known. In recent years many new variants that complicate analysis have been developed, most of which have been strongly solved as well. One game that is only ultra-weakly solved is called Movin’ On Up. This game is unique in that it involves multiple draws and payouts beginning with one starting hand.

The difficulty in determining a strategy for a game with multiple draws and payouts is that each additional draw increases the game complexity considerably. Methods that have been used to strongly solve other games are either not easily applicable or effective for games of this type. Additionally, optimal strategies may be too complicated to be playable by a human. Thus, our primary challenge is to develop a methodology for determining human-playable pure strategies for multi-draw/multi-pay games that are optimal or near-optimal and to do so with a reasonable amount of computation.

To effectively generate strategies a novel method was developed by both improving the prior published method of Monte Carlo simulation and incorporating the previously unused techniques of dynamic programming and lookup tables. The main result of our work is a method and implementation for the generation of decision lists describing pure strategies that have near-optimal equity for games with arbitrarily many draws and payouts. By utilizing our implementation we validate our method and protocol by producing previously unknown strategies for several multi-draw/multi-pay games.

The consequences of our work are numerous. Game developers, regulatory agencies, casinos, and players all have an interest in knowing strategies and the associated returns of games. There is also a considerable market for publications and software related to video poker strategy. Furthermore, the methodology, programming library, and decision lists described in our work both adds to the recent literature regarding Monte Carlo improvements and will assist with future research in the area of multi-draw video poker and other complex games.


Computer Sciences



Available for download on Friday, May 15, 2020