Cheapest: high risk, 12 rows
Returns 99.487%, an edge of 0.513%. Wagering $10,000 here costs about $51.27 in expectation. It also has 85.4% of drops coming back under the stake, so it feels far worse than it costs.
Duel Originals
27 Payout Tables · Measured 8 August 2026
Duel Plinko Every Payout Table, Measured
Nobody publishes what Plinko actually pays. So we stepped the risk tabs and the row slider through all 27 settings, copied every multiplier off the board, and worked out the return of each one. Here is the whole set.
27
Payout tables measured
0.88%
Mean house edge
1000x
Top multiplier
The 12-row board on medium risk, drawn from the measured strip: 33x at the edges, 0.3x in the centre.
How We Got These Numbers
Read Off the Board, Not Off a Press Release
Plinko has two controls: a risk selector with three positions and a slider for the number of rows. The slider steps through every integer from 8 to 16, which is nine settings rather than the four that most write-ups list. Three risk levels times nine row counts gives 27 separate payout tables, and the board redraws the bucket strip every time you move either control.
On 8 August 2026 we walked all 27 combinations and copied the multiplier in every bucket. Then we computed each table's return the only way it can be computed: a ball crossing a board of n rows makes n independent left-or-right decisions, so the chance of finishing in bucket k is the binomial term C(n,k) divided by 2n. Multiply each bucket's probability by its multiplier, add them up, and you have the return to player exactly. No simulation, no sample size, no margin of error.
Three of the 27 tables were re-read in a second pass to catch any misread, and all three matched. Every figure on this page, including the ones quoted in the FAQ, is generated from that one set of measurements, so the prose cannot drift away from the data.
The Result
What Each Setting Returns
Return to player for every risk level and row count, with the expected cost of wagering $10,000 at that setting. Green marks the cheapest table on the board, red the most expensive.
| Setting | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
|---|---|---|---|---|---|---|---|---|---|
| Low risk | 99.063% | 99.023% | 99.023% | 99.014% | 98.979% | 99.001% | 99.001% | 99.010% | 99.004% |
| Cost per $10,000 | $93.75 | $97.66 | $97.66 | $98.63 | $102.05 | $99.85 | $99.85 | $99.00 | $99.64 |
| Medium risk | 98.906% | 99.141% | 99.102% | 99.131% | 98.989% | 99.185% | 99.117% | 99.005% | 98.991% |
| Cost per $10,000 | $109.38 | $85.94 | $89.84 | $86.91 | $101.07 | $81.54 | $88.26 | $99.55 | $100.86 |
| High risk | 99.063% | 99.414% | 99.258% | 99.258% | 99.487% | 99.136% | 99.198% | 99.426% | 99.196% |
| Cost per $10,000 | $93.75 | $58.59 | $74.22 | $74.22 | $51.27 | $86.43 | $80.20 | $57.37 | $80.38 |
Costliest: medium risk, 8 rows
Returns 98.906%, an edge of 1.094%. The same $10,000 costs $109.38. Identical game, 2.13 times the price, and the only difference is a tab and a slider position.
Nothing reaches 100%
Across all 27 tables the return runs from 98.906% to 99.487%, averaging a 0.884% house edge. Not one setting is break-even, and the spread between the best and worst is wider than the edge on many blackjack tables.
Full Reference
All 27 Plinko Multiplier Tables
Filter by risk level. Because every table mirrors around the middle, the sequence from the centre bucket outward is the complete table.
| Setting | Buckets | Multipliers, centre outward | Top | Return | Cost per $10k |
|---|---|---|---|---|---|
| Low, 8 rows | 9 | 0.5 · 1 · 1.1 · 2.1 · 5.7 | 5.7x | 99.063% | $93.75 |
| Low, 9 rows | 10 | 0.7 · 1 · 1.6 · 2 · 5.7 | 5.7x | 99.023% | $97.66 |
| Low, 10 rows | 11 | 0.5 · 1 · 1.1 · 1.4 · 3 · 9 | 9x | 99.023% | $97.66 |
| Low, 11 rows | 12 | 0.7 · 1 · 1.3 · 1.9 · 3 · 8.5 | 8.5x | 99.014% | $98.63 |
| Low, 12 rows | 13 | 0.5 · 1 · 1.1 · 1.4 · 1.6 · 3 · 10 | 10x | 98.979% | $102.05 |
| Low, 13 rows | 14 | 0.7 · 0.9 · 1.2 · 1.9 · 3 · 4 · 8.2 | 8.2x | 99.001% | $99.85 |
| Low, 14 rows | 15 | 0.5 · 1 · 1.1 · 1.3 · 1.4 · 1.9 · 4 · 7.2 | 7.2x | 99.001% | $99.85 |
| Low, 15 rows | 16 | 0.7 · 1 · 1.1 · 1.5 · 2 · 3 · 8.1 · 15 | 15x | 99.010% | $99.00 |
| Low, 16 rows | 17 | 0.5 · 1 · 1.1 · 1.2 · 1.4 · 1.4 · 2 · 9.1 · 16 | 16x | 99.004% | $99.64 |
| Medium, 8 rows | 9 | 0.4 · 0.7 · 1.3 · 3 · 13 | 13x | 98.906% | $109.38 |
| Medium, 9 rows | 10 | 0.5 · 0.9 · 1.7 · 4 · 18 | 18x | 99.141% | $85.94 |
| Medium, 10 rows | 11 | 0.4 · 0.6 · 1.4 · 2 · 5.1 · 22 | 22x | 99.102% | $89.84 |
| Medium, 11 rows | 12 | 0.5 · 0.7 · 1.8 · 3 · 6.1 · 24 | 24x | 99.131% | $86.91 |
| Medium, 12 rows | 13 | 0.3 · 0.6 · 1.1 · 2 · 4 · 11 · 33 | 33x | 98.989% | $101.07 |
| Medium, 13 rows | 14 | 0.4 · 0.7 · 1.3 · 3 · 6.1 · 13 · 43 | 43x | 99.185% | $81.54 |
| Medium, 14 rows | 15 | 0.2 · 0.5 · 1 · 1.9 · 4 · 7.1 · 15 · 59 | 59x | 99.117% | $88.26 |
| Medium, 15 rows | 16 | 0.3 · 0.5 · 1.3 · 3 · 5 · 11 · 18 · 89 | 89x | 99.005% | $99.55 |
| Medium, 16 rows | 17 | 0.3 · 0.5 · 1 · 1.5 · 3 · 5 · 10 · 41 · 111 | 111x | 98.991% | $100.86 |
| High, 8 rows | 9 | 0.2 · 0.3 · 1.5 · 4 · 29 | 29x | 99.063% | $93.75 |
| High, 9 rows | 10 | 0.2 · 0.6 · 2 · 7.1 · 43 | 43x | 99.414% | $58.59 |
| High, 10 rows | 11 | 0.2 · 0.3 · 0.9 · 3 · 10 · 77 | 77x | 99.258% | $74.22 |
| High, 11 rows | 12 | 0.2 · 0.4 · 1.4 · 5.2 · 14 · 121 | 121x | 99.258% | $74.22 |
| High, 12 rows | 13 | 0.2 · 0.2 · 0.7 · 2 · 8.2 · 24 · 171 | 171x | 99.487% | $51.27 |
| High, 13 rows | 14 | 0.2 · 0.2 · 1 · 4 · 11 · 37 · 262 | 262x | 99.136% | $86.43 |
| High, 14 rows | 15 | 0.2 · 0.2 · 0.3 · 1.9 · 5 · 18 · 57 · 424 | 424x | 99.198% | $80.20 |
| High, 15 rows | 16 | 0.2 · 0.2 · 0.5 · 3 · 8.1 · 27 · 84 · 625 | 625x | 99.426% | $57.37 |
| High, 16 rows | 17 | 0.2 · 0.2 · 0.2 · 2 · 4 · 9.1 · 26 · 131 · 1000 | 1000x | 99.196% | $80.38 |
Each table is mirror-symmetric, so the sequence from the centre bucket outward is the whole table. Return is the sum of bucket probability times bucket multiplier across every bucket. Read off duel.com/plinko on 8 August 2026.
Reading the Board
Why the Middle Buckets Punish You
A ball on a 16-row board makes 16 coin flips on the way down. Finishing at an edge means calling all 16 the same way, which happens once in 65,536 drops. Finishing near the middle means splitting them roughly evenly, which is overwhelmingly the most likely outcome. That is why the centre buckets pay under 1x on every table and the edges carry the headline numbers.
- • Buckets always equal rows plus one, so 9 through 17
- • Odd row counts have no single centre bucket, just two joint-lowest
- • Centre pays 0.5x on low risk, as little as 0.2x on high
- • Row count moves variance, not price: the return barely shifts
The 1000x
What the Headline Multiplier Is Worth
High risk at every row count: the top multiplier, how often either edge bucket lands, and how much of the table's total return it actually supplies.
| Row count | Top multiplier | Odds of either edge | Share of total return | Drops under 1x |
|---|---|---|---|---|
| 8 rows | 29x | 1 in 128 | 22.66% | 71.1% |
| 9 rows | 43x | 1 in 256 | 16.80% | 82.0% |
| 10 rows | 77x | 1 in 512 | 15.04% | 89.1% |
| 11 rows | 121x | 1 in 1,024 | 11.82% | 77.3% |
| 12 rows | 171x | 1 in 2,048 | 8.35% | 85.4% |
| 13 rows | 262x | 1 in 4,096 | 6.40% | 73.3% |
| 14 rows | 424x | 1 in 8,192 | 5.18% | 82.0% |
| 15 rows | 625x | 1 in 16,384 | 3.81% | 88.2% |
| 16 rows | 1000x | 1 in 32,768 | 3.05% | 79.0% |
The pattern is the point. Moving from 8 rows to 16 multiplies the jackpot by roughly 34, divides its odds by 256, and cuts its contribution to your return from 22.7% down to 3.1%. The bigger the number on the edge, the more the table depends on its 0.2x centre.
The Zero Edge Badge
What That Label Does and Does Not Cover
Duel prints a Zero Edge badge directly beneath the Plinko board, and its own explainer makes a genuinely good argument for why a 0% edge is not merely a small improvement on 1%: the edge is charged on every bet, so it compounds, and only at exactly zero do a player's chances of a big win and a big loss become symmetric.
What that page does not do is name Plinko, publish a per-game figure, or describe a rebate that returns the edge. It closes by conceding that a casino with no house edge is expected to go bankrupt in the long run. The measured tables sit on the other side of that argument: every one of the 27 keeps between 0.513% and 1.094% of stake. Nor is it given back elsewhere, because the instant rakeback schedule covers slots at 50% and live blackjack at 60% and leaves the originals out precisely on the grounds that they already return everything.
None of which makes Plinko a bad game to sit at. An edge under one percent is better than almost anything in a land-based casino and better than the slots in the same lobby. It is simply not zero, and the gap between the badge and the bucket strip is worth knowing before you set the slider.
For comparison, the one original that does come close is Mines. Its largest recorded win, 2,040,932.03x, works out to exactly 0.999 of a full clear at nine mines, which fixes its edge at 0.1%, between a fifth and a tenth of what any Plinko table keeps. The full working is in the Duel Mines payout schedule.
Provably Fair
Rebuilding a Drop From Its Seeds
Plinko is unusually satisfying to audit, because the seeds do not merely fix the bucket the ball lands in. They fix every bounce on the way down, so a verified drop can be replayed peg by peg against the animation you watched.
Before the round the server generates a seed, hashes it, and shows you the hash. You supply a client seed of your own, and a nonce counts each drop. Those three values are hashed together, and the digest is sliced into one sub-window per row, each resolving to a single bit: left or right at that peg. Rotate the seed pair and the server reveals the plaintext seed it committed to. Hash the plaintext, check it against the hash you were shown, and you have proof nothing was swapped after you bet. Then re-derive the bit sequence and confirm the ball went exactly where the maths says it went.
The full verification walkthrough covers the seed rotation flow and the same commit-and-reveal scheme as it applies to Dice, Crash, and the rest of the originals.
Plinko FAQ
Questions About the Numbers
Not according to its own payout tables. We read all 27 risk and row settings off the live board on 8 August 2026 and computed the return of each from the binomial bucket distribution. They span 98.906% to 99.487%, averaging a 0.884% house edge. The Zero Edge badge sits directly under the board, and Duel's own Zero Edge explainer argues the case for a 0% edge in general, but it never names Plinko and never publishes a per-game figure.
High risk at 12 rows, which returns 99.487% and costs $51.27 per $10,000 wagered. The most expensive is medium risk at 8 rows, returning 98.906% and costing $109.38. That is 2.13 times as much for the same game, and nothing in the interface flags the difference.
It is the opposite of what most players assume. High risk carries the lowest house edge at eight of the nine row counts. What it does carry is brutal variance: on high risk at 10 rows, 89.1% of drops return less than the stake. The setting that feels most punishing is the one that costs least per dollar wagered.
Every integer from 8 to 16, which is nine settings rather than the four that most write-ups list. Combined with three risk levels that gives 27 distinct payout tables. The odd row counts, 9, 11, 13 and 15, produce an even number of buckets and therefore have no single centre bucket, just two joint-lowest ones.
The 1000x bucket exists only at 16 rows on high risk, and there are two of them, one at each edge. The chance of landing in either is 2 in 65,536, or 1 in 32,768 per drop. Despite the headline size it supplies just 3.05% of that table's total return, so the setting lives or dies on its 0.2x centre buckets.
No, because originals are excluded from it. The instant rakeback schedule pays 50% of the edge back on slots and 60% on live blackjack, and leaves the in-house originals out on the grounds that they already return 100%. Since the Plinko tables do not return 100%, that leaves roughly nine tenths of a percent of every stake uncompensated.
Yes, and the path is fully reconstructible rather than just the final bucket. Duel commits to a hashed server seed before the round, mixes in a client seed you set and a nonce that increments once per drop, then derives one left-or-right bit per row from the hash. Rotating the seed pair reveals the plaintext server seed so you can re-hash it, confirm the commitment held, and replay the exact bounce sequence.
There is no in-round decision at all: once the ball is released, every bounce is already determined by the seed pair and the nonce. The only choice that changes your expected cost is which of the 27 tables you sit down at, and that choice is worth a factor of 2.13 in house edge. Everything else, including autobet patterns and bet sizing, changes the shape of your variance and not the price.
Set the Slider Knowing What It Costs
Under one percent at every setting, but a factor of 2.13 between the cheapest table and the most expensive. Pick deliberately.
18+ · Gamble responsibly · Tables measured at duel.com/plinko on 8 August 2026 · Affiliate link disclosed