Payout table
Example payout table (illustrative, not from any casino) — symmetric, multipliers shrink toward the middle. Edit any value; RTP recalculates live.
Drop a ball
Generates fresh random seeds in your browser (never sent anywhere), then replays one round through the engine's Plinko path function.
How the Plinko distribution works
A Plinko board is a triangle of pegs. A ball dropped from the top hits one peg per row and bounces left or right with equal (50/50) chance, independently each time. After r rows the ball has made r independent left/right choices, and it lands in the slot equal to how many times it went right. That is exactly the setup for a binomial distribution: the chance of landing in slot i (out of r + 1 slots, numbered 0…r) is
P(i) = C(r, i) / 2^r where C(r, i) ("r choose i") counts how many distinct left/right sequences produce exactly i right-bounces. On a 16-row board there are 216 equally likely left/right sequences in total, but they are not spread evenly across slots — they pile up in the middle.
Why the middle hits often and the edges almost never
There is only one sequence that goes right every single time (landing in the far-right slot) and only one that goes left every time (far-left slot). But there are a huge number of different orders that mix 8 rights and 8 lefts, and all of them land in the exact middle slot. On a 16-row board, slot 0 (all left) occurs 0.0015% of the time — about 1 in 65,536 drops — while the middle slot 8 occurs 19.64% of the time, about 1 in 5.09 drops. Slot 4 sits in between at 2.78%. The curve is symmetric: P(i) = P(r − i), so slot 12 has the same probability as slot 4.
RTP: turning probabilities into a payout table
Probabilities alone don't tell you what a board pays — that needs a payout table, one multiplier per slot. Once you assign a multiplier mi to every slot, the long-run return is the probability-weighted average of all of them:
RTP = Σ P(i) × m_i Using the illustrative example table loaded by the "Example table" button above (multipliers shrink from the edges toward a sub-1× middle), a 16-row board comes out to an RTP of 120.19% — a house edge of -20.19%. Negative house-edge numbers are mathematically possible too (an RTP above 100%); the calculator reports whatever the table implies, it does not clamp or sanity-check it against a "normal" range. With this same table, the chance of landing on a multiplier above 1× is 45.45%, below 1× is 54.55%, and exactly 1× is 0.00%.
This table is something we made up to demonstrate the shape many Plinko-style payout tables use — it is not copied from, and does not claim to match, any specific casino game's real paytable. Edit any cell in the tool above and the RTP, house edge and win/loss split all recompute instantly, straight from the engine.
Why big edge multipliers barely move the RTP
It's tempting to look at a large edge multiplier and assume the board is generous. But that slot might only be reached 0.0015% of the time. Its contribution to RTP is P(i) × m_i — probability times multiplier. In the example table, slot 0 pays 6.12× at a 0.0015% chance, while slot 8 pays 0.75× at a 19.64% chance; the sum of every slot's P(i) × m_i, not any single slot, is what the RTP of 120.19% is built from. A designer can make one edge payout look spectacular without it dominating the overall RTP, simply because almost nobody lands there — but that only works if the rest of the table is sized to compensate, not automatically.
What "risk level" options actually change
Some Plinko-style games offer low/medium/high "risk" presets. In our reference model the mechanism — r independent 50/50 bounces — stays the same and only the payout table changes; whether a particular game works this way is something its own published rules have to confirm. A "high risk" table typically pushes more weight into the rare edge slots (bigger multipliers there, smaller or sub-1× multipliers in the busy middle), widening variance without necessarily changing RTP. "Low risk" flattens the curve back out. In the reference model neither option changes how often each slot is reached — that's fixed by r alone.
Myths: drop timing and position don't matter
Because the board looks physical, it's easy to assume that dropping the ball a split second earlier, or releasing it slightly left of centre, changes the outcome. In our reference scheme it does not: every bounce is derived from a server seed, a client seed and a round number (nonce) fixed before the ball visibly moves, via HMAC-SHA256 — timing and release position are not inputs. Being provably fair does not by itself guarantee that a particular game works the same way; its published rules say which inputs it uses. The "Drop a ball" demo above shows this directly — generate fresh seeds, drop, then follow the verify link to recompute the same path independently. For the reasoning behind these seed-based systems, see our provably fair verifier; for how session-level losses add up around a fixed edge, see the expected loss calculator and risk of ruin calculator. Plinko's binomial curve, Mines' hypergeometric odds (see the Mines odds calculator) and a crash multiplier's distribution (see the crash simulator) are different shapes of one idea: a fixed, math-defined edge that timing or pattern-spotting cannot move.
