Video Poker Tutorial: Hold Decisions by Expected Value

Video poker is five-card draw against a fixed paytable: the machine deals five cards, the player holds any of them, the rest are replaced once, and the final hand is paid by the table on the screen. There is no opponent to read, so the only decision is the hold.

That makes the game a measurable problem. Every hand has a limited set of possible holds (keep nothing, keep everything, or any combination in between), and each hold has an expected value: the average payout it returns across all possible draws. The correct play is the hold with the highest expected value, whether or not it looks like good poker.

Jacks or Better is the standard teaching game because its paytable is simple. Any pair of jacks or higher returns the bet, and everything above it scales up to the royal flush. Variants such as Deuces Wild change which hands pay and how wild cards count, which changes the right holds; the video Deuces Wild poker report covers how that game's paytable sets its return.

A tutorial built on expected value compares each candidate hold by what it returns on average, not by how strong it looks. That comparison is finite. Research in the UNLV Gaming Research & Review Journal counts 134,459 distinct starting hands in Jacks or Better once suits that play the same are treated as one, so every starting position can be worked out exactly rather than guessed.

This is why published strategies read as rankings. A strategy does not say "play aggressively" or "trust a hunch". It lists hold types in order, and the player keeps whichever qualifying hold sits highest on the list.

The holds that cost the most are the ones that follow instinct from table poker: keeping a made hand that is too small, chasing a draw that looks exciting, or throwing everything away in hope of a better deal. The published figures settle each case.

Two patterns run through the table. First, a paying hand in Jacks or Better is valuable because it is already a winner: the high pair outranks four to a flush in the published 9/6 strategy even though a flush pays more, because the pair still pays when the draw misses. Second, the royal flush is the one hand large enough to change the order, which is why three to a royal beats a low pair on that machine.

The J-T-9 suited case is the one that surprises most readers. It looks like a strong straight flush draw, yet DataGenetics puts holding J-T-9 suited and drawing two at an expected value of roughly 0.65. Looking strong and returning well are different things.

Very little, if it is a good one. The Wizard of Odds puts the optimal return of 9/6 Jacks or Better at 99.54%, and states that the mistakes in its simple strategy cost 0.03%, or one total bet every 3,805 hands (as of August 2026, the page's last update).

That figure is the practical case for learning a ranked list instead of computing every hand. A short list of hold rules recovers almost all of the available return. The bigger losses come from the instinctive holds in the table above, not from the fine distinctions a simplified list leaves out. Detailed strategy charts are a subject for a separate report; the point here is that the ranking exists and is short enough to learn.

The paytable sets the maximum return before any hold is made, and the law where the machine sits can cap it lower. In South Dakota, Rule 48:02:08:03 (effective August 1989) requires that video draw poker pay out between 83 and 96 percent of the amount wagered.

Set against the 99.54% optimal return of full-pay 9/6 Jacks or Better, that cap means a player in that state cannot meet a full-pay machine at all, however perfect the holds. The same hold rules apply, but they work toward a lower ceiling. Reading the paytable first is therefore step one of any tutorial: the free Jacks or Better pay table practice report walks through how each line on that table contributes to the return, and the Video Poker game reports collect the other variants.

Is video poker a game of skill or luck?

Both. The deal is random, but the hold is a choice, and the published math shows some holds return more than others on every hand. Choosing the higher expected value hold each time is the only part of the result a player controls.

Should

I keep a low pair or three cards to a royal flush?

In 9/6 Jacks or Better, the Wizard of Odds verdict is to go for the royal. The pair feels safer, but the royal draw carries more expected value on that machine.

Is a high pair better than four to a flush?

Yes. The published strategy keeps the high pair, because it already pays and still has room to improve, while the flush draw pays nothing when it misses.

Does every video poker machine return the same amount?

No. The return depends on the paytable and on local rules. South Dakota's rule caps video draw poker at 96 percent of the amount wagered, while full-pay 9/6 Jacks or Better plays to 99.54% with optimal holds.

• UNLV Gaming Research & Review Journal, Optimal Conditional Expectation at the Video Poker Game Jacks or Better, 2019-01-01
• South Dakota Legislature, Rule 48:02:08:03 Software requirements for percentage payout, 1989-08-18
• Wizard of Odds, Jacks or Better 9-6 intermediate strategy, 2026-08-03
• DataGenetics, Video Poker Strategy, 2019-07-03