Video Poker Strategy

Key Takeaways

• Video poker is solved: for any dealt hand and paytable there is exactly one expected-value-maximizing hold, and the full strategy fits on a chart.
• The paytable is the RTP: 9/6 Jacks or Better = 99.54% with perfect play; the same game at 8/5 = 97.30%, at 6/5 = 95.00%. Machine selection outweighs every in-game decision.
• Always bet five coins — the royal flush pays 800:1 only at max coins (250:1 below), and that single cell is worth ~1.5% of return.
• Strategy errors are quantifiable: casual "common-sense" play typically gives up 2–4% versus the chart, converting the best machine in the house into a mediocre one.
• The skill is real but bounded: even perfect play on 9/6 JoB leaves a 0.46% house edge — video poker is cheap entertainment, not employment (outside rare 100%+ paytables).

How Video Poker Works

One hand of five-card draw against a paytable:

1. Bet 1–5 coins (five activates the full royal payout — non-negotiable for correct play).
2. Deal: five cards from a standard shuffled 52-card virtual deck (regulated machines must deal from a fair deck — no weighting, unlike slot reels).
3. Hold: choose any subset of the five to keep. This is the game's entire decision space — 32 possible holds per hand.
4. Draw: replacements dealt for discards; the final hand pays per the table.

Jacks or Better — the base variant — pays from a pair of jacks upward:

The bolded cells name the variant: "9/6" machines return 99.54% at optimal play; every step down (9/5, 8/5, 7/5, 6/5) costs roughly 1.1% each. Two glances at a paytable price the machine exactly — a transparency no slot offers.

The Strategy: expected value per hold

Optimal play evaluates all 32 holds and keeps the best. Software solved this decades ago; humans use ranked charts. The core of the 9/6 JoB chart, from strongest to weakest (hold the highest-ranked pattern your hand contains):

1. Royal flush / straight flush / four of a kind (pat)
2. Four to a royal — even breaking a pat flush or straight
3. Full house / flush / straight / three of a kind (pat)
4. Four to a straight flush
5. Two pair
6. High pair (jacks+)
7. Three to a royal
8. Four to a flush
9. Low pair (tens or worse)
10. Four to an outside straight
11. Two suited high cards
12. Three to a straight flush
13. Two unsuited high cards (lowest two if three)
14. One high card
15. Discard everything

(A condensed chart; the full version adds a handful of edge-case lines and costs only ~0.08% versus computer-perfect play.)

The chart's counterintuitive lessons are where casual players leak money:

• Low pair beats four to a straight (line 9 vs 10): the pair's quad/full-house upside outweighs the straight's completion odds (~17%).
• Never keep a "kicker" with a pair — the ace next to your jacks is pure EV poison.
• Break a flush for four to a royal: 800:1 at ~1-in-47 beats the sure 6.
• Never draw five when you hold a high card — but never hold three random "good" cards either; the chart, not hope, ranks draws.

What perfect play is worth

9/6 JoB, optimal:        99.54%  (house edge 0.46%)
Typical intuitive play:  ~96–98% (2–4% strategy loss)

At 600 hands/hour and $1.25 per hand ($750/hour turnover), the difference between chart play and guesswork is roughly $15–30/hour of real money — the best-paying learnable skill per minute of study in the casino.

The royal's role

With optimal play the royal arrives about once per 40,400 hands (chasing draws makes it more frequent than the 1-in-649,740 raw deal odds) and contributes ~2% of total return. Practical implications: short sessions almost never see one — sub-royal return runs near 97.5% — so bankroll variance is dominated by that missing 800:1 event, and the certified return only materializes across tens of thousands of hands.

Hand frequencies and bankroll reality

What optimal 9/6 Jacks or Better play actually produces, per final hand (standard combinatorial results, rounded):

Read the two columns together and the game's temperament emerges: ~45% of hands return something, but nearly half of those are the 1:1 pair — effectively a push. The bread-and-butter value lives in two pair through full house; the royal is a rare 2% annuity you prepay every session.

The variance number that matters: 9/6 JoB has a standard deviation of about 4.4 bets per hand. Over a 4,000-hand weekend (roughly 7 hours), one standard deviation is ±280 bets — against an expected loss of just 18 bets at 0.46%. Swings of hundreds of bets in either direction are therefore the normal experience, and the often-quoted bankroll guidance of 3–4 royal cycles (several thousand bets) for serious play is variance arithmetic, not superstition. Casual players can scale down: the same math says a 400-bet stake survives most single sessions of correct quarter play.

The variant map

Every variant is the same skeleton with a re-solved chart:

Multi-hand formats (Triple Play, Five Play, up to Hundred Play) deal one starting hand, let you hold once, then draw each hand's replacements from its own independent deck remainder. Per-hand return is unchanged; variance per round rises with correlated results — a dealt four of a kind pays on every line, a dead deal dies everywhere at once.

The 100%+ schedules are historical artifacts and competitive rarities — where they exist, they assume flawless play of noticeably harder charts (Double Bonus strategy is far less forgiving than JoB). For online play, the SlotCodex catalog tracks 117 video poker titles from 28 providers (July 2026 snapshot); online paytables cluster tighter than land-based floors but still vary per title — the two-cell check works everywhere.

For Players

• Learn one variant deeply. A 9/6 JoB chart mastered beats five variants half-learned; charts are variant-specific, and cross-applying them leaks EV.
• Price machines in two glances: full house and flush cells (JoB) or the quad/five-kind cells (Deuces). Walk past anything shaved twice.
• Five coins, smaller denomination. A five-coin quarter bet ($1.25) outperforms a one-coin dollar bet ($1.00) in return terms purely through the royal cell.
• Use a trainer. Free strategy trainers flag your chart errors; an evening of drills typically eliminates most of the 2–4% casual leak.
• Respect the variance. A 99.5% game still swings hard between royals; a bankroll of several thousand bets is normal for surviving the cycle without the royal's subsidy.
• Slot-style myths don't apply: the deck is fair, hot/cold machines don't exist, and your hold decisions are the only variable you control — see hit frequency for how often you'll see any return (~45% of hands, nearly half of them the pair-back push-like win).

For the Industry

• Video poker is the transparency benchmark. Its visible paytable + solvable strategy model is regularly cited in fairness debates as what slots don't offer; catalog products (including SlotCodex) can compute and display exact per-title returns — a data quality bar no other category matches.
• Paytable shaving is instantly legible to the target audience. The category's players are the floor's most math-literate; 8/5 conversions harvest short-term hold at measurable reputational cost within exactly the segment most likely to publicize it.
• Online variant depth is an underexploited niche. Land-based floors monetize habit; online lobbies carry thin video-poker selections relative to demonstrated demand — 117 tracked titles versus thousands of slots — leaving a differentiation opening for operators courting analytical players.
• Skill-adjacent compliance nuance. Return figures assume optimal play; several jurisdictions require disclosure language distinguishing theoretical maximum return from expected casual-play return. Marketing "99.5% RTP" without the strategy caveat is a recurring advertising-standards trap.