Keno Strategy: Pick Count Changes Your Variance, Not Your Edge

Search for keno strategy and you will find the same advice everywhere: pick four to six numbers, because the middle of the ticket has the best returns. It is repeated so often that it reads like arithmetic rather than folklore.

We ran the numbers against our own paytables. On Suigar keno, that advice is simply false. Picking three numbers and picking ten numbers return the same expected value to the third decimal place. What changes when you move the slider is not your edge. It is your variance — and the two things get confused constantly.

The board, in one paragraph

Suigar keno uses a 40-tile board. Ten tiles are drawn each round. You choose between one and ten of them before the draw, and the draw itself comes from Sui native on-chain randomness, so the result is settled by the same transaction that took your bet. Two risk levels are playable today: Low and Classic. They share a board and a draw; they differ only in how the paytable distributes the money.

Every number below is computed directly from the paytables the contract enforces, using the hypergeometric distribution — the exact probability of hitting h of your k picks when 10 of 40 tiles are drawn. No simulation, no sampling error.

Expected value does not move

Here is the return to player at every pick count, for both playable risk levels. Read down the columns and look for the pattern.

There is no pattern, because there is no effect. Low returns 97.00% whether you pick one number or ten. Classic returns 96.50%. The tiny wobbles — 97.02% at two picks, 96.47% at three — are rounding artifacts in the published multipliers, worth two hundredths of a percent. Nobody has ever won money from a two-hundredths-of-a-percent edge.

This is not an accident or a quirk of our tables. It is what a well-built paytable looks like. The house edge is a design parameter chosen per risk level, and the multipliers at each pick count are then solved backwards to hit it. A pick count that paid better than the others would be a bug, and it would be arbitraged within a day.

What actually changes: the shape of the outcome

Pick count controls the distribution, and the distribution is dramatic. Here is what a round looks like at each pick count on Classic, alongside the floor that Low guarantees you when you hit nothing at all.

Two things in that table are worth more than the rest of this article.

The hit rate is a sawtooth, not a slope

Your chance of finishing a round ahead climbs from 25.0% at one pick to 58.9% at three picks — and then falls off a cliff to 25.6% at four. It climbs again to 47.4% at six, then collapses to 22.8% at seven.

The cause is the minimum hit threshold. At three picks, a single hit already pays 1.33x. At four picks, one hit pays nothing at all — you need two. Every time the paytable raises the bar for the smallest paying result, the hit rate resets and starts climbing again. Three, six and ten picks sit at the top of each ramp. Four and seven sit at the bottom.

So if what you want is to see the board light up often, three picks is the best ticket on the board and four picks is close to the worst. Same edge. Wildly different session.

Low and Classic have the same win rate

The "chance of finishing ahead" column applies to both risk levels, and that is not a shortcut in the table. The two paytables cross 1x at exactly the same hit count at every pick count. Low and Classic put you ahead equally often. They disagree only about what happens on either side of that line.

Low never zeroes you: even a complete miss on a ten-pick ticket returns 0.54x. In exchange, its ceiling is 4.94x. Classic pays nothing on a miss and reaches 80.21x. Picking a risk level is not picking better odds. It is picking whether you want a soft floor or a high ceiling, because you cannot have both.

The 80x is a lottery ticket, and the odds say so

The largest multiplier on the Classic board is 80.21x, paid for hitting all ten of your picks. That result comes up once in roughly 848 million rounds. For scale: if you played one round every second without stopping, you would expect to see it about once every 27 years.

Nine of ten is 1 in 27.3 million. Eight of ten is 1 in 1.7 million. The top three rows of a ten-pick ticket are, in any realistic session, decoration. The money in a ten-pick ticket comes from the 1.31x at three hits and the 2.36x at four — which together account for the 48.5% of rounds that finish ahead.

This is worth internalising for any high-multiplier casino game, not just keno. A headline multiplier tells you about the tail of the distribution. It tells you nothing about the edge, and almost nothing about what your session will feel like.

So how should you actually choose?

Since the edge is fixed, the only honest framing is: what do you want the next hour to look like?

• Frequent, small movement: three picks. 58.9% of rounds finish ahead — the highest on the board. Pair with Low if you want a floor under the losses.
• Balanced: six picks. 47.4% of rounds finish ahead, with a 15.82x reachable at 1 in 18,278 on Classic.
• Chasing the tail: ten picks on Classic. Still 48.5% of rounds ahead, but the payouts stretch to 80.21x. Expect long flat stretches.
• Avoid by accident: four and seven picks. They sit at the bottom of the sawtooth — the same edge as everything else, but the lowest hit rates on the board. There is no reason to land there unless you chose it deliberately.

What you should not do is switch pick counts mid-session hoping to find a better return. There is no better return. The only lever the game gives you is variance, and every pick count spends the same house edge to buy a different amount of it.

Check the math yourself

You do not have to take any of this on trust, which is rather the point of running a casino on-chain. The paytable and the draw both live in the contract, the randomness comes from Sui native on-chain randomness, and every round settles in a public transaction. We wrote a walkthrough of how to verify a single bet end to end if you want to trace one yourself.

If you want the wider context on where that 96.50% comes from and who it is paid to, we broke down house edge and RTP for on-chain casinos separately. The same variance-versus-edge distinction shows up in our plinko guide, where the risk tiers do the same job that pick count does here.

Keno is live now at suigar.com/keno.

Method

Probabilities are exact hypergeometric values for a 40-tile board with 10 tiles drawn. Multipliers are read from the Low and Classic paytables enforced by the Suigar keno contract on Sui mainnet. RTP is the sum of multiplier times probability across every hit count. Percentages are rounded for display; the underlying figures are not rounded before summing.