Expected Value of a Hold

The expected value of a video poker hold answers one specific question: If I keep these cards and draw the rest, what is the average return of that decision across every possible draw?

That definition is the foundation of correct strategy. The best hold is not necessarily the hand that looks strongest now, produces the most frequent wins, or creates the most exciting jackpot possibility. It is the hold with the highest probability-weighted payout under the exact game and paytable.

Start by treating every legal hold as a separate decision

From a five-card deal, a player can hold none of the cards, all five, or many combinations in between. Each hold creates a different set of possible draws from the remaining deck.

A strategy analyzer conceptually performs the same process for each candidate:

1. Freeze the chosen cards.
2. Enumerate every possible draw for the discarded positions.
3. Classify each resulting five-card hand.
4. Apply the machine’s paytable.
5. Weight each payout by how often it occurs.
6. Average the results.

The hold with the largest average return ranks first.

This is why video poker strategy is mathematical rather than stylistic. A hand can “look good” and still have a lower EV than a less obvious alternative.

The calculation is a weighted average, not a prediction

The core relationship is:

EV of hold = Σ (probability of final result × payout of final result)

If an analyzer works directly from draw combinations, the same idea can be written as:

EV of hold = total payout across all possible draws ÷ number of possible draws

The two forms are equivalent when every possible draw is counted correctly.

EV is not a forecast of what the next draw will pay. A hold with an EV of 1.10 credits per credit wagered can still lose immediately. The number describes the long-run average of repeating that exact decision under the same conditions.

A pair can outrank prettier high-card combinations

Consider a simple Jacks or Better-type situation where the player has a low pair plus two unsuited high cards. A casual player may prefer the high cards because they offer visible straight or royal possibilities. Strategy can prefer the pair because it already has made-hand value and can improve to trips, a full house, or four of a kind.

The decision is settled by the return distribution, not by the visual appeal of the cards.

This is why low pair vs high cards is a useful companion page. It illustrates how a modest-looking made hand can carry more average value than attractive but weakly connected high cards.

Breaking a paying hand can sometimes be correct

Expected value also explains one of video poker’s most uncomfortable plays: discarding part of a completed paying hand.

Suppose a player has a flush but also four cards to a royal flush in a game where the royal award is large enough that the four-card royal draw has greater EV than keeping the flush. The player gives up a guaranteed current payout to pursue a higher average return.

That does not mean players should routinely break made hands. It means every hold is compared on average value.

The four to a royal and breaking a made hand pages show why “never throw away a winner” is not a reliable rule.

The paytable can change the ranking of two close holds

Expected value belongs to a specific payout schedule. Change the rewards and the weighted averages change.

If one version pays more for a full house, draws that produce full houses gain value. If another game heavily rewards premium four-of-a-kind outcomes, some pair or kicker decisions can move. If a progressive royal grows, royal-producing holds may become more valuable.

This is why a strategy chart copied from the correct game name but wrong paytable can still be inaccurate.

The Wizard of Odds video poker analyzer is a practical example of paytable-specific hand analysis, and the video poker return summary shows how different schedules produce different overall returns.

Compare close decisions by EV gap, not by confidence

Some holds are separated by a large amount of expected value. Others are extremely close. That distinction matters for learning.

Imagine two candidate holds with these hypothetical returns per five-credit wager:

Hold A is still correct, but the penalty for choosing Hold B is only 0.05 credits on average for that decision.

Now compare:

The second mistake is much more expensive.

A good training plan therefore prioritizes high-cost errors first rather than treating every strategy miss as equally serious.

One lucky draw cannot validate a bad hold

Suppose the analyzer says Hold A is superior, but the player chooses Hold B and hits a royal. The outcome was excellent; the decision can still have been inferior.

Likewise, the player can make the top-EV hold and draw nothing. That does not make the strategy wrong.

This separation between decision quality and result quality is central to gambling analysis. Video poker gives immediate, emotionally powerful feedback, so players can easily build false rules from memorable outcomes.

If you want to improve strategy, record whether the hold was correct before looking at whether the draw won.

Strategy error cost can be expressed in dollars

The EV gap between the best hold and the chosen hold represents the average cost of that mistake.

Suppose the correct hold is worth 4.10 credits on average and the selected hold is worth 3.90 credits.

EV error = 4.10 - 3.90 = 0.20 credits

At 25¢ denomination:

0.20 × $0.25 = $0.05 average cost for that decision

Five cents sounds trivial. Repeated hundreds or thousands of times, recurring mistakes reduce the effective return of the game.

This is why common video poker strategy mistakes focuses on repeated decision leaks rather than isolated bad luck.

The casino’s theoretical return assumes a strategy model

Published video poker RTP figures are generally tied to a particular paytable and an assumed strategy, often optimal play. A casino’s actual observed hold over a short period can differ dramatically because of jackpots and normal variance. Over larger volumes, player mistakes can also contribute to a gap between theoretical and realized results.

For operations, this distinction matters when interpreting game performance. A machine with a strong published return can still produce positive casino revenue because not every player uses optimal strategy and because short-run results fluctuate widely. Conversely, a large player win does not prove the paytable is misconfigured.

Device integrity and strategy are separate layers. Technical standards such as GLI standards address approved game-device behavior. They do not determine which cards the player should hold.

How to use an analyzer without becoming dependent on it

An analyzer is most useful as a learning tool after the decision, not merely as an answer machine.

For each mistake, ask:

• What were the two strongest candidate holds?
• How large was the EV difference?
• Which final outcomes created that difference?
• Was the error caused by the paytable, a wild card, a kicker, a penalty card, or a general hierarchy rule?

Over time, this turns isolated answers into pattern recognition.

The goal is to understand why a low pair beats certain high-card combinations, why four to a royal can outrank a made hand, or why a kicker matters in one bonus-game situation but not another.

Expected value is the language that connects hand strategy to overall RTP

A game’s total return emerges from making these hand-level choices correctly across the full distribution of deals. Strategy charts are therefore compressed EV rankings. Each line is a shortcut for a large amount of draw enumeration.

That perspective resolves many video poker myths at once. The best hold is not the safest hold, the most exciting hold, or the hold that won last time. It is the one with the largest average return under the actual rules.

Continue with hold or draw decisions, video poker math basics, and how to read a strategy chart to turn the EV concept into practical play.