As with other forms of
casino gambling, the word on
video poker is
don't play. The vast majority of the machines have a
payoff of less than 100% which means that in the long run you must lose. A few rare machines, such as the
All American game (with payoffs of 800, 200, 40, 8, 8, 8, 3, 1, and 1 for the above categories), have slightly positive payoffs (in that case, 100.72%). Most, like the most common
Jacks or Better game (with payoffs of 800, 50, 25, 6, 4, 3, 2, and 1), are not (in this case 99.54%).
Even if you do play a machine with greater than 100% payoffs, the practical difference of 1% in payoffs is virtually nothing. Video poker is highly variable; and your winnings or losses will swing wildly as you keep losing and then suddenly win a sizable amount. Of course, the losing streaks will make you desperate to win more, and the wins will encourage you to play more, either way. But to keep that .72% advantage, you must play perfectly. The combinatorial and probabalistic complexity of video poker is enormous. There are 2,598,960 possible hands and 32 possible plays on each one. It takes dedicated study and practice to get your strategy down close enough to perfection to maintain an edge over the machine. Even once you've done that, it is impossible to play fast enough to make a lot more than the minimum wage with odds so close to even. Additionally, changes in the payoff table change the optimal strategy completely. If you go from one kind of machine to another without adjusting your strategy, you will lose big time.
So the moral is that it really doesn't matter whether you are playing blackjack, video poker, slot machines, or roulette. You will almost certainly lose money, and the effort it takes to be assured long term profits would be much better spent earning money honestly. If you play at all, expect to lose every dollar you bet. Generally, there are much more entertaining ways to spend money than gambling.
All that said, I find the theory of video poker and other games of chance and skill very fascinating.