Every Polymarket combo ever traded

Combos went live on May 29, 2026: one position across several markets that pays only if every pick wins. Polymarket publishes no feed of them. Every combo route in its Data API takes a wallet address and answers for that wallet alone, the one combo endpoint that needs no login returns the catalog of markets that can be combined, and the ordinary trades route never returns a combo fill. You can look up your own combos, but not everyone's.

So we read Polygon instead. The CombinatorialModule contract registers each combo and its picks, Exchange v3 records the fills, and the backend has read both every 5 seconds since the launch block. One match emits a fill for the taker order and one for each maker order it crossed, so we count takers only: a $20,000 order that swept 15 resting orders is one trade, not 16.

At 23:02 UTC on September 6, 2026, that held 1,187,808 taker fills from 82,894 wallets and $54,462,121.68 staked on buys. We have not found another public index of it.

The weekly money in the chart below followed the World Cup. In the week after the July 19 final, wallets buying combos fell from 21,010 to 5,753, but the stake fell by less than half, from $4.31 million to $2.51 million, and was back to $3.72 million a week later. The casual players left with the tournament. The money mostly stayed.

The 20 most-traded competitions of the last 30 days are all soccer. The English Premier League carries 13.32% of every pick bought, La Liga 11.28%, Serie A 5.61%, the Bundesliga 4.49%, Ligue 1 3.63%, and MLS 3.25%. Then come the Primeira Liga, the Champions League, the Saudi Pro League, the Leagues Cup, the Conference League, the Eredivisie, and 8 more, all of them football.

The first entry that is not soccer is the ATP tour at 1.41%. Baseball is behind tennis at 1.16%. The NFL, in September, is 0.14%. American leagues barely register.

The picks are the clubs you would guess. Over 30 days, FC Barcelona appeared in 5,955 combos bought, Real Madrid 5,243, Manchester City 4,155, Arsenal 3,508, and Liverpool 3,375. Bayern, Chelsea, Manchester United, and Paris Saint-Germain follow, with the over/under lines and both teams to score between them. Wallets are stacking favorites and totals, the same shape as a sportsbook parlay slip.

The last 7 days give the current rate: 52,792 combo buys from 6,255 wallets, $3,631,181.62 staked, an average ticket of $68.78, and an average of 3.40 picks per combo. On a busy day that is around 12,000 tickets and about $750,000.

On June 22 at 21:50 UTC a wallet then called Rash-Copying bought a two-pick combo at 31.5 cents: Algeria to win and Norway to win, both World Cup fixtures on the same day. The stake was $19,999.96, which at that price bought 63,535.23 shares.

Algeria beat Jordan, and Norway beat Senegal. A winning combo share redeems for $1, so the position paid $63,535.23, a gain of $43,535.27 on two results in one afternoon. It was the largest combo payout in the September census, and every fill of it is in the index.

Polymarket's own profit curve credits the wallet with $45,654 realized on combos, almost all from this one $43,535.27 win, and puts it at -$2,194,412 across its whole history on October 11, 2026. One ticket does not establish repeatable combo skill. It carries an F grade, graded on booked profit rather than on one afternoon.

On July 17 a wallet then called YarushaKing put $57.62 on 9 picks and collected $6,720.32: 116.6 times the stake, the largest multiple any winning combo of $50 or more had returned by September 6. Its ticket was one of 99 buys of $50 or more priced to pay 100 times or better, out of 144,900.

The wallet now goes by Auramaxxking. On October 11, 2026, Polymarket's curve had it $2,943 up on combos and $3,351 down in all, and Polymarket's leaderboard put it at -$27,064. Both of the census's record holders are underwater on Polymarket's own numbers. One winning ticket is not a record.

The average ticket falls with every pick from two to six, and is $13.71 at 19 picks. The longest combo anyone had bought by September 6 had 38 picks. The dollars follow the same slope: two-pick combos held $32,009,741.55 of the $54.46 million staked, and three-pick combos another $12,273,757.17. Everything with 5 picks or more is small by value and large by count.

Read that as two products sharing one page. Short combos take the size: a wallet with a view on two matches, sizing it like a real position. Long combos are mostly $20 tickets. The combo index shows both, and its size filter separates them fastest.

82,404 wallets have at least one combo that has settled. 19,219 of them, 23.32%, got back more than they put in. 43,897, or 53.27%, have never had a single combo hit, so more than half of everyone who has tried combos has nothing to show for it.

The median wallet staked $28.37 in total. The best net result in the ledger is $190,171.68 and the worst is -$117,641.56. Across all 1,008,641 settled buys, $53,435,156.57 went in and $49,651,141.81 came back: 92.9¢ on the dollar, held to settlement and before fees. The per-pick version of that curve, which gets steeper with every pick added, is in the combos guide.

That does not make combos a bad bet. It says the average combo buyer pays roughly 7¢ on the dollar to hold one, and that the distribution is lopsided: a small group takes the payouts and a much larger group takes the zeros. You can see which group a wallet has belonged to before you copy its ticket, which is why the grade sits on every row.

The census counts taker buys only, treats each one as held to settlement, pays winners $1 a share and leaves fees out. By October 10, 22,955 wallets had also sold combo positions 200,225 times as takers, for $15.8 million. A wallet that sold half a ticket at a profit shows up here as whatever the ticket eventually did, and the resolution status describes the combined picks, not anyone's realized profit.

On the other side of the buys sit the makers: 1.95 million maker fills, nearly all of them buys from 634 wallets. The census does not measure them, so it says nothing about how the people quoting combos did. The index is also a chain read with the chain's limits: a fill can land before its combo is registered, and a combo with a voided pick reads Unknown, because Polymarket has published no rule for what a void does to a combo.

The wallet totals above come from Polymarket's own profit curve, not from summing our stored rows; only the provider's figure is an account total. Everything else in the census was read from production at 23:02 UTC on September 6, 2026, and the market has moved since.

In the 30 days to October 10, 2026, 170,467 of 207,881 combo buys staked under $50, and 28 staked $10,000 or more. The small tickets are the crowd. The money sits in the $100-to-$10,000 tickets from a few thousand wallets, and that is where a combo says something about a trader's view.

The combo index lists every combo with its stake, its payout, its picks and the wallet behind it, sortable by newest, largest or payout and filterable by sport, result, grade, pick count, size, odds and window. Pro sees each combo as it fills; without Pro the index runs 24 hours behind. Set the size floor at $1,000 and sort by largest to skip the crowd. The combos guide covers how a combo is priced and settled.