Arithmetic, not advice
The two ends of the payout row, counted rather than argued about
The largest number in a Plinko row sits in the corner bin and the smallest sits in the centre. How many routes through the grid end in each is a counting problem with an exact answer: on a sixteen-row board, 12,870 of 65,536 routes finish in the middle bin and exactly one finishes in each corner. That is the whole of what this page claims.
Two ends of one list
The payout row is symmetrical, so it has two identical ends and a centre. The numbers printed at the ends are the largest in the row; the number printed at or beside the centre is the smallest. That is the shape of every published row, and it is described in more detail on the page about the payout row.
Under those two ends sit two very different counting problems, and both of them have exact answers.
The answers are worth having because they are the only figures anywhere near this game that do not depend on an operator publishing anything. They come from the grid.
Counting the routes into a bin
A ball crossing a board of n pin rows makes n decisions, one per row: left or right. A complete route is therefore a sequence of n letters, and there are 2 to the power of n such sequences.
Where the ball ends up depends only on how many of those decisions were left, not on the order they came in. Going left, left, right lands in the same bin as right, left, left.
So the number of routes ending in a given bin is the number of ways to choose which of the n bounces went left — the binomial coefficient. The corner bins correspond to "all left" and "all right", and each of those can be done exactly one way. The centre bin corresponds to "half left, half right", which can be done in the largest number of ways of any split.
Sixteen rows, seventeen bins, 65,536 routes
Written out for a sixteen-row board, the route counts across the seventeen bins are:
1, 16, 120, 560, 1,820, 4,368, 8,008, 11,440, 12,870, 11,440, 8,008, 4,368, 1,820, 560, 120, 16, 1.
They sum to 65,536, which is 2 to the sixteenth. The centre bin accounts for 19.64 per cent of all routes and each corner bin for 0.0015 per cent — a ratio of 12,870 to 1 between the two ends of the same row.
Shorter boards compress the spread rather than removing it. A twelve-row board has 4,096 routes, 924 of them into the centre bin and one into each corner: 22.56 per cent against 924 to 1. An eight-row board has 256 routes, 70 into the centre and one into each corner: 27.34 per cent against 70 to 1.
Fewer rows means a shorter list, a flatter distribution of routes across it, and a corner that is less isolated from the rest. It does not mean a better or worse game, and this site does not say which to pick — the settings themselves are described on the page about row count and risk.
What these counts are, and what they are not
They are counts of paths through a grid. They are exact, they are checkable with a pencil, and they are identical at every operator running a board of that size.
They are not probabilities, and the difference matters.
Turning a route count into "how often this bin is hit" requires assuming that every route is equally likely, which in turn requires each bounce to be an even split. That assumption belongs to the idealised board — the physical device the game is modelled on — and not to any particular software build. None of the eleven operators in our table publishes anything about how its build resolves a bounce, and the facts library behind this network has no field in which such a thing could be recorded. Twenty-one operators of the hundred on file carry a tag for publishing a seed-checkable round, which is a different question again: it lets a completed round be recomputed, not a distribution be audited.
So the honest form of the sentence is the long one. On a sixteen-row grid, one route in 65,536 ends in each corner. Whether the software makes those routes equally likely is not something this site has read anywhere.
Why the middle number is the smallest
The shape of the payout row and the shape of the route counts are inversions of each other, and that is not a coincidence anyone needs to prove: it is how the row was designed.
The bins with the most routes into them carry the smallest printed numbers. The bins with one route each carry the largest.
That relationship is the whole reason the row looks the way it does, and it is why a screenshot of a row with a very large corner number tells you almost nothing on its own — the corner number and the row length move together. A longer board produces both a bigger corner figure and a corner that is harder to reach, and neither half of that trade is published as a rate.
What the counting cannot reach
Everything past this point is documents rather than arithmetic, and the documents are where the eleven operators actually differ.
The figure that would let two configurations be compared as choices — a published return for a specific row count and risk setting — is not in our data for a single operator in our table. Neither is any game-level figure at all: there is no games field in the hundred-operator library, only a tag showing which operators run an in-house originals suite.
What is in the data is the plumbing. Two of the eleven print an amount above which identity documents are demanded and the rest reserve the right to ask at any point, which is set out under verification. Three print a ceiling on what leaves the account in a period, covered under withdrawal ceilings. Those are the numbers a corner bin eventually has to pass through, and unlike the route counts they are different at every site.