Keno Probability Explained: The Actual Math Behind the Odds

Why this is worth understanding

Our Keno Odds Explained guide covers the shape of keno's odds curve without the underlying formula. This guide is for anyone who wants the actual math — not because you need it to play keno well, but because understanding exactly where the numbers come from removes any doubt about whether keno's odds are as advertised, and makes it obvious why certain beliefs about the game (hot numbers, patterns, "due" numbers) don't hold up.

The type of probability keno uses

A keno draw is what's called sampling without replacement from a fixed population — the game draws 20 numbers from a pool of 80, and once a number is drawn, it isn't eligible to be drawn again in that round. This is formally known as a hypergeometric distribution, the same category of math used to calculate odds in card games (where drawing a card removes it from the deck) rather than dice or roulette (where every roll or spin is independent and unaffected by previous ones).

The simplest case: a 1-spot ticket

Start with the easiest possible example. If you pick a single number, your probability of catching it is simply the number of winning numbers divided by the total pool:

P(catch) = 20 / 80 = 25%

This one is intuitive and easy to verify: 20 of the 80 numbers will be drawn, so any single number you pick has a 1-in-4 chance of being among them.

A 2-spot ticket, worked through step by step

Once you pick more than one number, you need to account for the fact that the pool shrinks as numbers are drawn. The cleanest way to calculate this is sequentially:

• Probability your first picked number is drawn: 20/80
• Given that it was drawn, probability your second picked number is also drawn: 19/79 (one winning number and one total number have already been "used up" by the first check)

Multiply these together for the probability of catching both:

P(catch both) = (20/80) × (19/79) = 380/6,320 ≈ 6.01%

This sequential method works because each additional number you need to match further shrinks both the remaining winning-number pool and the remaining total pool — which is exactly the "sampling without replacement" behavior that defines a hypergeometric distribution.

The general formula for catching all your numbers

For a ticket with n picked numbers, the probability of catching every single one follows the same pattern, extended across all n picks:

P(catch all n) = (20/80) × (19/79) × (18/78) × ... × [(20−n+1) / (80−n+1)]

Each term in the sequence accounts for one more number already having been "removed" from both the winning pool and the total pool by the previous checks. This is exactly why the odds of a full catch get dramatically longer as pick count increases — you're multiplying together more and more fractions, each one further reducing the running total, rather than adding a fixed amount of difficulty per extra number. This compounding effect is the mathematical reason behind the pattern described in Keno Odds Explained: odds for a full catch don't get proportionally longer as pick count rises, they get exponentially longer.

The full formula: catching exactly k of n picks

The examples above cover catching all of your numbers. The more general formula — for catching exactly k numbers out of n picked, when 20 are drawn from 80 — uses combinatorics (counting combinations) rather than simple sequential multiplication:

P(X = k) = [C(n, k) × C(80−n, 20−k)] / C(80, 20)

Where C(a, b) means "a choose b" — the number of ways to choose b items from a items, regardless of order. In plain terms: this formula counts every possible way to catch exactly k of your n numbers while missing the rest, divides by every possible way the full draw of 20 could come out, and the result is your probability. This is the formula behind every keno paytable calculation and every RTP figure a game discloses — Keno RTP Explained covers how this feeds into that larger calculation.

Why this confirms number selection doesn't matter

Notice that nowhere in either formula does it matter which specific numbers you picked — only how many. The math treats every number in the 80-number pool identically, which is the formal, provable basis for why hot numbers, cold numbers, birthdays, and Quick Pick are all mathematically equivalent choices. Keno Number Selection Explained covers the practical implications of this directly.

Extending the worked example to a partial catch

The 2-spot full-catch example earlier showed how to calculate catching all your numbers. It's worth also seeing how a partial catch is calculated, since most real paytable tiers pay for partial catches, not just full ones. Take a 3-spot ticket and calculate the probability of catching exactly 2 of your 3 numbers (missing one).

Using the general formula, P(X = k) = [C(n, k) × C(80−n, 20−k)] / C(80, 20), with n = 3 and k = 2:

• C(3, 2) = 3 — the number of ways to choose which 2 of your 3 numbers are the ones that get caught
• C(77, 18) — the number of ways the remaining 18 drawn numbers can come from the 77 numbers you didn't pick
• C(80, 20) — the total number of ways any 20 numbers can be drawn from 80

The numerator (3 × C(77,18)) counts every way to catch exactly 2 of your 3 numbers while missing the third; dividing by the denominator (C(80,20)) converts that count into a probability. This is the same structural logic as the full-catch example, just accounting for the extra ways a partial catch can happen (since it matters which 2 of your 3 numbers were the ones caught, not just that 2 were caught) — which is exactly what the C(n, k) term captures.

Why the combinatorial formula and the sequential formula agree

It's worth noting that the sequential method used earlier for full catches (multiplying fractions like 20/80 × 19/79) and the combinatorial formula here are mathematically equivalent ways of describing the same hypergeometric distribution — they'll always produce identical results for the same question, just via different calculation paths. The sequential method is more intuitive for full-catch calculations specifically; the combinatorial formula is the general-purpose version that handles partial catches (and full catches) alike. Neither is more "correct" than the other — they're the same underlying math, viewed two different ways.

Expected value: combining probability and payout

Once you have the probability of each catch tier, multiplying each by that tier's payout and summing the results gives you the game's expected value per unit staked — which, expressed as a percentage, is exactly RTP. This is the same calculation illustrated with simplified numbers in Keno RTP Explained, just connected here directly to the formal probability formula that feeds into it. Understanding this connection is really the payoff of working through the math: RTP isn't a number operators simply choose, it's a direct, calculable consequence of the paytable and the fixed probabilities of the 80-number, 20-draw structure.

Independence between draws

One more property worth stating precisely: each keno draw is fully independent of every other draw. The formulas above describe a single, self-contained draw of 20 from 80 — nothing about a previous round's results feeds into or adjusts the next round's probabilities in any way. This is the formal basis for why a number that hasn't been drawn recently isn't "due," a common misunderstanding known as the gambler's fallacy.

A note on variance and standard deviation

Probability tells you the likelihood of each individual outcome; variance (and its square root, standard deviation) describes how spread out your actual results are likely to be around the expected average across a session. Two pick counts can share a similar RTP while having very different variance — one producing a tighter cluster of outcomes close to the average, the other producing wider swings with more extreme highs and lows. This is a formal way of describing what How Many Numbers Should You Pick in Keno? calls "session feel" — higher pick counts generally carry higher variance, meaning your actual results are more likely to land far from the theoretical average in any given session, in either direction.

Why large sample sizes matter for verifying fairness

A single session, or even a few hundred rounds, isn't enough data to statistically distinguish a fair game from a subtly unfair one just by observing results — the natural variance in a random process is large enough to mask small deviations over a limited sample. This is exactly why independent RNG certification relies on enormous sample sizes (often millions of simulated rounds) run directly against the software, rather than relying on aggregated player-reported outcomes. It's also why, as a player, personally trying to verify a game's fairness by tracking your own results isn't a practical approach — the sample size available to any one player is far too small to draw a statistically meaningful conclusion either way.

What this math is useful for in practice

You don't need to run these calculations yourself to play keno — legitimate games disclose paytables and often RTP directly, already accounting for this math. What understanding the formula actually buys you is confidence: knowing that keno's odds are a fixed, calculable property of the 80-number/20-draw structure means there's no hidden pattern to find, no number that's secretly better, and no way to game the system through selection method. The only genuine levers you have are pick count (which this math shows directly shapes your odds) and choosing a game with a better paytable — both covered in our Keno Strategy Guide

Frequently asked questions

Do I need to understand this formula to play keno well? No — it's useful for understanding why the odds behave the way they do, but playing well comes down to pick count, paytable comparison, and bankroll management, none of which require calculating probabilities yourself.

Is this the same math used for lottery odds? Very similar — lottery number-draw games typically use the same hypergeometric structure, though with different pool sizes and draw counts depending on the specific lottery format. See Keno vs Lottery

Why does the probability of a full catch drop so fast as pick count rises? Because each additional number you need to match multiplies another shrinking fraction into the total probability, rather than adding a fixed amount of difficulty — a compounding effect rather than a linear one.

Can I verify a specific game's paytable using this math? Yes, in principle — if you know the full paytable and calculate the probability of each catch tier using the formula above, you can derive the game's RTP yourself. In practice, checking the disclosed RTP on the game's info screen is far more practical than doing this by hand.

Why does the formula use combinations (C) instead of simpler multiplication? Because catching exactly k of n numbers can happen in multiple different ways (different subsets of your numbers could be the ones caught), and combinations count exactly how many of those ways exist without regard to order — which is precisely what's needed here, since the order numbers are drawn in doesn't affect your payout, only which numbers end up drawn.

Does this math change for way tickets or multi-race tickets? No — the same formula applies to each individual way or race within those ticket types, since each is still an independent application of the same 80-number, 20-draw structure. What changes is how many times you're applying it within one ticket, not the formula itself. See Keno Betting Options Explained