Video poker is the lowest-edge widely-available casino game after blackjack — full-pay Jacks or Better runs about 0.46% house edge, lower than baccarat or craps with full odds. Played correctly, it's nearly a break-even game.
And yet most VP players go broke. Why? Variance. Specifically, the royal flush — 1 in 40,000 hands, paying 4,000:1 — accounts for nearly all of the long-run EV. If you don't hit one in your first 40,000 hands (and 37% of players won't, by Poisson math), you'll be down roughly 2% of money wagered, even though your house edge is supposedly 0.46%.
The variance number that matters
Per-unit-bet standard deviation for full-pay JoB is about 4.42. Compare to:
- Blackjack basic strategy: σ = 1.14
- Baccarat banker: σ = 0.93
- Roulette: σ = 1.00
- Craps pass + max odds: σ = 3.16
VP variance is roughly 4× blackjack. A standard deviation captures how much a single hand's outcome differs from the mean — in VP, that's because most hands win nothing or pay 1:1, and a tiny fraction pay 800:1 (straight flush) or 4,000:1 (royal). The distribution is wildly skewed.
What this looks like in a session
Run a session: 1,000 hands at $1.25 per hand (max-coin quarter machine). Total wagered: $1,250. Expected loss at 0.46%: $5.75. Standard deviation: ~$176.
- 5th percentile: down ~$295 (a bad session)
- 50th percentile: down ~$5 (the median)
- 95th percentile: up ~$285 (a good session)
- 99th percentile: up ~$415 (a great session)
Half your sessions you'll be down more than $5. Almost a quarter of sessions you'll be down more than $100. The 'expected loss' figure is buried in the noise of any individual session.
Why royals are everything
On full-pay JoB, the EV breakdown across hand types:
- Royal flush — 1.98% of total return (despite being 0.0025% of hands)
- Straight flush — 0.55%
- 4 of a kind — 5.91%
- Full house, flush, straight — combined ~11%
- Three of a kind down to pair of jacks+ — combined ~80%
Half of the difference between full-pay and a 9/5 jacks paytable is the royal payout. If you don't hit royals, you're not playing full-pay VP — you're playing a worse-edged game until you catch up.
Bankroll for VP
Two-thirds of casual VP players underbankroll. The rule of thumb: you need about 5,000 max-coin bets to survive normal variance for a full-pay game with reasonable confidence.
At $1.25 max-coin (quarter machine), that's $6,250. At $5 max-coin (dollar machine), that's $25,000. Most casual players walk in with $300-500 and wonder why they bust.
Practical advice
- Always play max coin (5 coins). The royal payout 4,000:1 only triggers on max coin — short coin caps the royal at 1,000:1, which torpedoes your edge.
- Always play full-pay. 9/6 JoB = 99.54% return. 8/5 JoB = 97.30%. Difference = 2.24%, ~five times the full-pay house edge. Find 9/6 machines or don't play.
- Don't bankroll trip-by-trip. Track a year of VP and let variance smooth across it.
- Sign up for the players club. VP comp rates are low (15-25%), but every dollar back reduces effective edge.
- Use the VP Paytable Scanner in TableSharp to verify a machine is actually full-pay before sitting.
VP versus blackjack
If you have $5,000 to play with: would you rather play blackjack at $25 a hand for 200 hours, or VP at $1.25 max-coin for 4,000 hours?
EV-wise, similar (BJ has lower house edge, VP has lower bet). Variance-wise, VP is much rougher per hour. Skill-wise, BJ basic strategy is faster to learn but counting is much harder; VP optimal play is more complex than BJ basic strategy but doesn't require counting.
Most disciplined casino players settle on one or the other. The hybrid approach (some BJ, some VP) is fine — just don't underbankroll either.