VPK 302: Video Poker Expected Value — From One Hold to a Full Session

In video poker, expected value (EV) is the average mathematical value of a decision if the same situation could be repeated many times. It can describe one hold, one dealt hand, an entire paytable played with a specified strategy, or a session’s theoretical result.

EV is useful because it judges the decision before the draw card appears. A correct hold can lose. A poor hold can hit a jackpot. The result tells you what happened once; expected value tells you which choice was mathematically better before the random draw.

That distinction is the core of video poker strategy.

Four different things players call “EV”

The phrase gets used at several levels:

These are connected, but they are not interchangeable.

A game advertised or calculated at 99.54% RTP does not mean every hand is worth 99.54% of its wager. Some dealt hands have very high conditional value; many have low value. The overall return is the weighted average across the enormous distribution of possible initial hands and optimal decisions.

For the narrower combinatorics of a single decision, use expected value of a hold.

The basic formula

For any decision with multiple possible outcomes:

EV = Σ [P(outcome) × net value(outcome)]

where:

• P(outcome) is the probability of that result after the chosen hold;
• net value(outcome) is the payout or profit assigned to that result;
• Σ means add the probability-weighted value of every possible result.

If one hold has EV 1.42 units and another has EV 1.31 units, the first hold is better mathematically even if the second one happens to win on the next draw.

The gap between those two EVs is the cost of the strategy error for that decision.

A worked hold example: four to a royal in 9/6 Jacks or Better

Suppose a five-credit 9/6 Jacks or Better hand is:

10♠ J♠ Q♠ K♠ 2♦

Hold the four suited royal cards and discard 2♦. There are 47 possible replacement cards because the five originally dealt cards are no longer in the draw deck.

Under the standard full-pay max-credit schedule, the draw outcomes are:

The conditional expected payout is:

EV payout = [(1×4000) + (1×250) + (7×30) + (6×20) + (9×5)] / 47
          = 4625 / 47
          ≈ 98.40 credits

That number is not the RTP of the machine. It is the expected payout from this unusually strong specific draw position after the initial hand has already been dealt.

This illustrates why video poker strategy is a conditional-EV problem. The player is not asking “Which hand looks strongest right now?” The player is asking “Which set of cards produces the highest average value across all legal draws from here?”

Paytable changes alter game EV even when the rules look identical

“Jacks or Better” is not one fixed return percentage. The full house and flush rows alone can materially change long-run return.

For example, 9/6 Jacks or Better pays 9 for a full house and 6 for a flush per unit on the standard schedule. 8/5 Jacks or Better reduces those two rows.

The dealing rules may look identical, but the expected value changes because the probability of each hand is multiplied by a different payout.

This is why quoting an RTP without naming the exact paytable is incomplete.

The paytable also affects strategy. A draw that is marginally better under one schedule can be marginally worse under another because the value of flushes, full houses, quads, kickers, or wild-card outcomes has changed.

Strategy is part of the published return assumption

A theoretical return for video poker normally assumes a particular standard of play, often optimal play for the exact paytable.

If the player makes mistakes, the cards are not changed and the paytable is not changed. The player simply selects holds with lower conditional EV than the best available choice. Over time, those decision losses reduce the achieved return.

That gives video poker an important distinction from a conventional chance-only slot:

Configured game return under optimal strategy
-
value lost to strategy errors
=
approximate achieved return

This is a conceptual relationship rather than a simple fixed subtraction, because the cost depends on which errors are made and how often they occur.

A strategy chart is therefore not magic. It is a compressed way of selecting the highest-EV hold for recurring hand patterns.

The research problem is larger than a short strategy chart suggests

For Jacks or Better, the number of strategically distinct starting situations is large even after suit symmetry is used to combine equivalent hands. Ethier, Kim and Lee’s paper on optimal conditional expectation in Jacks or Better describes how exact conditional expected returns can be used to derive a complete hand-rank strategy.

The practical lesson is not that a player should calculate thousands of cases at the machine. It is that a simple strategy list rests on a much larger EV comparison underneath it.

That is also why a strategy for one variant should not be copied blindly to another.

Game EV becomes RTP

At the whole-game level, expected return is usually expressed as RTP:

RTP = expected amount returned / amount wagered

If a game has an optimal theoretical return of 99.5439%, then its corresponding house edge is:

House edge = 1 - 0.995439
           = 0.004561
           = 0.4561%

That figure assumes the exact paytable, wager conditions, and strategy used to derive the 99.5439% return.

It does not mean the player receives 99.5439% back during a short session. Rare hands, especially the royal flush, contribute to long-run return but may not appear for a very long time in an individual sample.

For that reason, video poker house edge should always be read alongside video poker variance.

Session EV: converting percentage edge into dollars

Once the game-level edge is known, theoretical session cost can be estimated from coin-in:

Total amount wagered = wager per hand × number of hands
Expected loss = total amount wagered × house edge

Using the 99.5439% example above, suppose a player wagers $5 per hand for 1,000 hands:

Coin-in = $5 × 1,000 = $5,000
House edge = 0.4561%
Expected loss = $5,000 × 0.004561 ≈ $22.81

The $22.81 is a long-run average cost under the stated assumptions. It is entirely possible to lose hundreds of dollars in that sample, or to finish far ahead, because the variance of video poker is much larger than the small average edge.

This is a good example of why “low house edge” and “low session risk” are not synonyms.

EV can rise without the base paytable changing

Progressive video poker can add value to one or more top awards as a meter grows. If the royal-flush prize increases while other payouts stay fixed, the expected value of holds that can reach the royal also increases.

At a sufficiently large progressive value, the optimal strategy itself can change because chasing the royal is worth more than it was under the base schedule.

That does not mean every progressive is positive EV. The full calculation must include:

• the exact base paytable;
• the current progressive amount;
• the bet required to qualify;
• the probability of reaching each affected award;
• the strategy appropriate to that meter value.

A jackpot amount by itself is not enough information.

Max credits can change EV if the royal is nonlinear

Many traditional video poker schedules pay proportionally for one through four credits but give a disproportionate royal-flush award at five credits. A classic pattern is 250 credits for a one-credit royal but 4,000 credits for a five-credit royal rather than 1,250.

In that structure, playing fewer than the qualifying maximum changes the paytable and therefore changes expected value.

This is not an argument to increase a wager beyond a comfortable budget. It means the player must compare the actual paytable at the intended wager size, not quote the five-credit RTP while playing a different schedule.

See max coins in video poker for that specific issue.

Regulators treat skill contribution as part of game mathematics

Video poker is still a regulated gaming device. Nevada’s current gaming-device submission guidance specifically asks, when player skill is involved, whether its effect on game payback percentage can be calculated or estimated.

That is exactly the issue EV analysis handles: the machine’s random deal and draw are combined with a player decision that can change the expected return.

The device does not need to know whether a particular player understands EV. The approved mathematics define what the game can return under specified play; the player’s decisions determine how closely actual strategy follows the optimal conditional choices.

EV does not tell you which session will win

Expected value is powerful precisely because it is limited.

It can tell you:

• which hold is better on average;
• which paytable is better under comparable strategy;
• how much a strategy error costs in expectation;
• what long-run return a game offers under stated assumptions;
• what theoretical loss corresponds to a given amount of action.

It cannot tell you:

• whether the next draw will complete a flush;
• when the next royal will appear;
• whether a 500-hand session will finish ahead;
• how severe the largest drawdown will be;
• whether a losing streak means the game is “due.”

Those are outcome and variance questions, not expected-value questions.

The decision rule that matters

For every dealt hand, the mathematically correct principle is:

Choose the legal hold with the highest conditional expected value for the exact paytable and wager condition being played.

You do not need to calculate that from scratch at the machine. A correct strategy chart or analyzer has already done the comparison. But understanding the rule prevents a common mistake: judging a hold by what happened after the draw.

A bad result does not make the highest-EV hold wrong. A lucky result does not make a lower-EV hold right.

For detailed hand-level calculation, continue to expected value of a hold. For dollar-cost estimates across a session, use the expected-loss calculator or compare swings with the variance simulator.