Methodology
If our EVs were wrong, this site would be worse than useless. Here is exactly how the numbers are computed, how they were verified, and where the limits are.
How the solver works
Video poker is fully solvable. For any dealt 5-card hand there are exactly 2⁵ = 32 possible holds. For each hold, the solver enumerates every possible draw from the remaining 47 cards — no sampling, no simulation, no lookup tables — evaluates each final hand against the selected paytable, and averages. The hold with the highest expected value (EV) is the best play, by definition rather than by opinion. A full 32-hold solve takes a few milliseconds and runs entirely in your browser (a Web Worker keeps the interface smooth); our server never sees your hands.
How it was verified
The RTP check is the strong one: computing a game’s total return requires solving every possible deal. We collapse the 2,598,960 deals into 134,459 suit-equivalence classes, solve each exactly, and weight by multiplicity. Landing on the published figures to seven decimal places is an end-to-end proof that the evaluator, the enumeration, and the paytables are all correct simultaneously.
How the strategy charts were built
The charts we teach are not copied from books — they are optimized against the solver and the claims are measured. The Jacks or Better study evaluated the classic simple strategy over every possible deal (exactly 99.4593%) and found a two-line upgrade to 99.5003%. The Double Bonus study found a royal-draw line that published charts miss, verified across all 3,456 relevant hands. Where a tweak measured worse, we rejected it and said so.
Known limitations
EVs assume max bet (five coins), where the royal pays 800 per coin. The trainer teaches ordered strategy lists, which trade a small, exactly-measured amount of return (0.036% on 9/6 JoB) for memorability — the solver itself is always optimal, and the analyzer shows the true ranking for any hand. Charts are paytable-specific: a 9/6 chart on an 8/5 machine plays almost correctly but cannot recover the paytable’s ~2.2% penalty. And no strategy changes variance: even perfect play on a 100.17% game loses on most sessions.
Check us
Every number on this site can be audited from your own browser: open the hand analyzer, enter any hand, and compare against any other solver. If you find a disagreement, we want to know: [email protected].