A Plinko board puts a 1,000x multiplier on each edge and makes it look reachable. I wanted to see how often it really pays, so I dropped 1,000,000 simulated balls on a 16-row board. The fun part: you barely need a simulation, because a Plinko board is a binomial distribution you can watch.
The board is 16 coin flips
At every peg the ball goes left or right with a 50/50 chance. After 16 rows it lands in one of 17 slots, and the slot index is simply the number of "right" bounces. That is Binomial(16, 0.5).
- To hit an edge slot, every bounce has to go the same way: 1 in 2^16 = 1 in 65,536 per edge.
- The centre gets the most paths:
C(16, 8) = 12,870of the 65,536, about 19.6%.
The code
import numpy as np from math import comb ROWS = 16 p = np.array([comb(ROWS, k) for k in range(ROWS + 1)]) / 2**ROWS # binomial slot odds half = [1000, 130, 26, 9, 4, 2, 0.2, 0.2, 0.2] # high-risk table, edge -> centre pay = np.array(half + half[-2::-1]) # mirror into 17 slots print(f"edge slot: 1 in {1 / p[0]:,.0f} each, 1 in {1 / (2 * p[0]):,.0f} for either edge") print(f"middle 3 slots: {p[7:10].sum():.2%} of balls") print(f"exact RTP: {(p * pay).sum():.3%}") rng = np.random.default_rng(42) slots = rng.binomial(ROWS, 0.5, 1_000_000) # 1,000,000 balls, each = 16 coin flips won = pay[slots].sum() print(f"1,000,000 x $1 balls: returned ${won:,.0f}, board kept ${1_000_000 - won:,.0f}, " f"1,000x hits: {(slots == 0).sum() + (slots == ROWS).sum()}") Output:
edge slot: 1 in 65,536 each, 1 in 32,768 for either edge middle 3 slots: 54.55% of balls exact RTP: 98.976% 1,000,000 x $1 balls: returned $986,791, board kept $13,209, 1,000x hits: 29 What the numbers say
The 1,000x slot is real, and rare. One ball in 32,768 lands on either edge. In a million balls that is about 30 hits; this run got 29.
The middle pays for the edges. More than half of all balls land in the middle 3 slots, and on this high-risk table the five centre slots pay 0.2x. That is where most bets quietly lose 80 cents.
The edge never moves. The exact RTP is 98.98%, so the board keeps about 1 cent per $1 ball. Over 1,000,000 balls it kept $13,209, even after paying out $29,000 in 1,000x hits.
In the full write-up I compared three risk levels with the same ~99% RTP and gave 10,000 players 1,000 balls each. On high risk, 35% of players finished ahead, but the median player lost $49.80 and 1 in 3 lost $100 or more. Same RTP, very different swings.
Try it yourself
- Swap
halffor a low-risk table such as[16, 9, 2, 1.4, 1.4, 1.2, 1.1, 1, 0.5]and watch the variance collapse while the RTP stays at ~99%. - Change
ROWSto 8 or 12 and see how fast the edge odds improve (1 in 256 per edge at 8 rows). - Plot
np.bincount(slots)againstp * 1_000_000: the bell curve appears on its own.
Full write-up with charts (where every ball landed, the running total per risk level, 10,000 players): I dropped 1,000,000 Plinko balls. There is also a free Plinko calculator that computes the odds and RTP of any payout table, and all the simulation code is on GitHub.
18+. A probability exercise, not betting advice: no risk level removes the house edge.
