Video Poker: Where Slots Meet Skill and the Math Gets Interesting

The casino put a decision tree inside a slot machine and nobody in management seemed to worry about it. The machine looks like every other machine on the floor. Same chair. Same screen. Same coin slot. A player sits down, presses buttons, watches cards appear, holds some, draws replacements, and either wins or doesn’t. To the casual observer it is a slot machine with a poker skin. To the player who has done twenty minutes of homework, it is one of the best bets in the entire building.

The difference between those two experiences is not talent. It is not luck. It is not some mystical card sense that separates winners from losers. It is the ability to read a pay table, which is displayed on the screen before you put a single dollar in, and the willingness to memorize a strategy card that fits on an index card. That’s it. The rest is the casino betting you won’t bother.

Two Identical Machines, Six Times the Cost

Every video poker machine publishes its pay table on the screen. The numbers are right there. Most players treat them like the terms of service on a software update; technically visible, functionally invisible.

For Jacks or Better, the most common variant, the critical numbers are the payouts for a full house and a flush. A machine paying 9 coins for a full house and 6 for a flush per coin wagered is called a 9/6 machine, or “full pay.” With optimal strategy, it returns 99.54% to the player. A machine paying 8 and 5 returns 97.30%. A machine paying 7 and 5 returns 96.15%. A machine paying 6 and 5 returns 95.00%.

These percentages sound close together. They are not close together. At $1.25 per hand on a quarter machine running 600 hands per hour, the 9/6 machine costs you roughly $3.45 per hour at optimal play. The 8/5 machine costs roughly $20.25 per hour. Same game. Same buttons. Same speed of play. Same visual experience. Six times the price. The casino is not hiding this information. It is sitting on the screen in plain numbers. The casino is simply confident, and historically correct, that almost nobody will compare them.

The pay table hierarchy runs all the way from full pay down to configurations that are barely better than a standard slot machine. Every step down the ladder costs approximately one percentage point in expected return. By the time you reach a 6/5 Jacks or Better machine, the house edge is 5%, which is table-game-losing-streak territory. The player who sits at a 6/5 machine and plays video poker “strategy” is doing sophisticated work on a fundamentally bad bet. The player who walks ten feet to the 9/6 machine is playing one of the closest-to-even games in the casino.

Finding full-pay machines is its own skill. They have been disappearing from casino floors for twenty years as floor managers figure out that the math-literate players are costing them money. Full-pay machines migrate to higher denominations, off-strip properties, downtown Las Vegas, and corners of the casino where the foot traffic is thin. The website vpFREE2 maintains a database of full-pay machines by casino and denomination. Serious video poker players check it before they book a trip. This is not obsessive behavior; this is the equivalent of checking gas prices before a road trip, scaled to a context where the price difference is measured in thousands of dollars over a year of play.

The Decision Tree Most Players Get Wrong

Optimal video poker strategy is a ranked list of hold decisions. You receive five cards. You decide which to keep. You draw replacements. The correct choice depends on what you are holding, what you could draw into, and the probability-weighted payout of each possibility. Computer simulations have solved this completely for every major variant. The optimal play for every possible five-card deal is known. There are no judgment calls. There is no feel. There is a list, and you follow the list.

The strategy card for 9/6 Jacks or Better contains roughly thirty rules in priority order. You compare your hand against the list from top to bottom and hold the first combination that matches. Casinos allow players to use strategy cards at video poker machines. Unlike blackjack, where pulling out a basic strategy chart at the table draws attention from the pit, video poker players routinely consult cards and nobody cares. The machine doesn’t know. The casino doesn’t mind. You are alone at a screen.

The places where recreational players go wrong are consistent and revealing. The most common mistake is holding a “kicker” alongside a pair. You have a pair of sevens and an ace; the instinct is to hold the ace because aces feel important. The math says throw the ace away. You want three draw cards to improve the pair into trips, a full house, or four of a kind. The kicker reduces your draw by one card and helps you with nothing. Poker instinct, imported from the table game where kickers break ties between identical pairs, is actively harmful in video poker. There are no opponents. There are no ties. There is only the expected value of each hold combination against the known payout schedule.

The second common error is breaking a paying hand to chase a royal flush draw. You hold a low pair and three cards to a royal. The low pair is already money in hand. The three-card royal draw has a seductive expected value in certain configurations but a lower one in others, and the boundary between those cases is where the strategy card earns its keep. The correct play depends on the specific cards, the specific pay table, and whether a progressive jackpot has shifted the expected value of the royal high enough to justify the gamble. Without checking the card, most players guess. Most players guess wrong.

With the strategy card memorized and applied consistently, the house edge on 9/6 Jacks or Better drops to 0.46%. That is less than half a percent. The gap between how most people play video poker and how the machine should be played is one of the largest skill gaps in any casino game. And the information required to close that gap is free, legal, and fits on an index card.

The Comp System Treats You Like a Slot Player

Casino comp programs are built around slots. The formula runs: coin-in multiplied by theoretical hold percentage yields the player’s “theoretical loss,” and the casino kicks back a percentage of that theoretical loss as comps. Free rooms, free meals, free play, cash back. The percentages vary by property and loyalty tier, but the structure is universal.

Video poker machines are classified as slots in virtually every casino’s comp system. This classification is the crack in the wall. A slot player generating $750 per hour in coin-in on a machine with 8% theoretical hold has an expected loss of $60 per hour. The casino comps them accordingly. A video poker player generating the same $750 per hour in coin-in on a 9/6 Jacks or Better machine, played optimally, has an actual expected loss of about $3.45 per hour. But the comp system frequently assigns a generic theoretical hold that overestimates the player’s losses, sometimes dramatically.

The result is that video poker players receive comp value calibrated for a game that bleeds 5% to 12% while actually playing a game that bleeds 0.46%. At aggressive comp programs, the comp return can exceed the actual house edge. When a 0.2% to 0.5% comp return is layered on top of a game with a 0.46% house edge, the effective return to the player approaches or exceeds 100%. Jean Scott documented this playbook in The Frugal Gambler in the late 1990s. The fundamental math has not changed. What has changed is the availability of full-pay machines and the generosity of comp programs, both of which have tightened. The opportunity is thinner than it used to be. It has not disappeared.

Royal Flush Variance Will Test Your Bankroll

The royal flush in Jacks or Better pays 4,000 coins at max bet, which is 800 to 1. That single hand accounts for approximately 2% of the game’s total expected return. The royal arrives, on average, once every 40,000 hands. At 600 hands per hour, that is roughly 67 hours of play between royals.

This creates a structural problem that surprises players who look at the 99.54% return and expect smooth sailing. Without the royal flush contribution, the effective return drops to approximately 97.5%. You are running at a persistent deficit, session after session, waiting for a single large payout to pull the long-term average back to its theoretical number. The math works. The math just works over a timeframe longer than most people’s patience or bankroll can sustain.

Bankroll recommendations for serious video poker play run to 200 to 400 royal-flush cycles for the variant being played. For quarter Jacks or Better at max bet ($1.25 per hand), that means a bankroll of $2,000 to $4,000. For dollar machines at max bet ($5 per hand), it means $8,000 to $16,000. These numbers shock people who think of video poker as a casual Tuesday-night activity. The expected cost per hour is low. The variance around that expectation is high. Those are different things, and confusing them is how people go broke playing a game that technically favors no one.

The Variants Where the Math Gets Genuinely Strange

Jacks or Better is the foundation, but the video poker ecosystem extends into territory that surprises even experienced casino players. Deuces Wild, where all twos act as wild cards, returns 100.76% at optimal play on full-pay machines. That is a positive expectation game sitting on the casino floor, accessible to anyone who can find the machine and learn the strategy. The strategy is substantially more complex than Jacks or Better because the wild cards change the relative value of every hand; the decision tree is roughly twice as long and the penalties for deviation are steeper. Full-pay Deuces Wild machines are genuinely rare. Casinos know the math. But where they exist, a player with a memorized strategy card is playing a game that favors the player before comps enter the equation. The machine pays back more than it takes in over time. That sentence sounds like it should be illegal. It is not.

Double Bonus Poker shifts extra payout to four-of-a-kind hands, producing a full-pay return of 100.17%. The strategy tilts more aggressively toward quad draws, which means more hands where the correct play feels wrong; breaking a paying two-pair to chase four of a kind, holding a single ace over a low pair because the quad-aces payout is disproportionately large. The variance ratchets up accordingly. Sessions swing harder. The bankroll requirement deepens. Double Double Bonus pushes further into four-of-a-kind territory with kicker bonuses, creating even higher variance and even more complex strategy. These games reward the obsessive student of video poker while punishing the casual player more harshly than Jacks or Better does, because the mistakes are costlier and the swings are less forgiving.

The broader point is that video poker is not one game. It is a family of games with different return profiles, different strategies, and different variance structures, all housed inside machines that look functionally identical to the casual player. The casino floor does not advertise which of these games it is offering. The pay table is the only signal. Reading it is the only required skill. Everything follows from that.

The casino designed a game where knowledge determines whether you lose money slowly, break even, or occasionally come out ahead. It housed that game in a chassis that looks identical to machines with a 10% house edge. It priced the comp program as though every machine on the floor is a standard slot. And it bet that most players would never do the homework.

Most players don’t. That is the whole opportunity.