An API-Only Launch 11 Weeks After Certification
Polymarket US processed its first multi-leg sports trade on Aug. 5, 11 weeks after certifying the product with federal regulators, according to InGame’s analysis of the exchange’s trade data. The outlet counted 16,173 trades totaling $7.4 million in volume and $928,093 in stakes from the taking side, with most of it arriving after Aug. 13. Volume on prediction exchanges counts both sides of every trade, so the stake figure is the closer measure of money actually put up.
The option does not appear in the Polymarket US app, the exchange has not widely launched a desktop site, and the combos documentation sits under the institutional section, available only to explicitly enabled API users.
The underlying product is the Combinatoric Athletic Outcome Contract: every leg must settle on the chosen side for the contract to pay $1.00; if any single leg misses, it pays nothing. The schema accepts two to 10 legs. Polymarket US self-certified it on May 20 and amended the terms on July 14, with the revised products listable no earlier than July 28. The first trade came eight days after that. Polymarket’s offshore site, which blocks U.S. traffic, had been running parlays since June 10.
Kalshi has had the category largely to itself in the meantime. Its parlay volume climbed from $4.77 billion in May to $13.78 billion in July, a stretch that coincided with record prediction-market activity around the World Cup – figures that count both sides of every trade. InGame counted $25 million in parlay taker fees over the first 16 days of August against a thin sporting calendar. Kalshi’s combos are also described as limited to professional football, professional basketball, and its “mention” markets, so the breadth is narrower by category than the leg counts suggest.
Reports that Kalshi will begin charging maker fees on parlays from Aug. 20 trace to a trader’s post on X circulated on Aug. 15, not to the exchange. Kalshi’s published fee schedule lists no combo or parlay series among the entries visible, and its upcoming-changes section states that none is scheduled. The schedule also carves out specific products that “have their own fee schedule,” so a combo need not appear in the main table at all.
What the schedule does show is how the economics fall on parlay traders. The 1.75% peak is a share of contract value, not of money staked, and it applies at 50 cents. Parlays price at long odds by construction. At one cent, 100 contracts cost $1.00 and carry a seven-cent fee, or 7% of the stake. At five cents the rate is 6.8%, at 10 cents 6.3%, and only at a coin flip does it fall to 3.5%. The cheapest fees on the curve sit where parlay traders never are.
Polymarket US has moved the opposite way. Its fee schedule, effective exchange-wide since July 1, peaks at $1.50 per 100 contracts against Kalshi’s $1.75, and pays makers a rebate of 0.0125 × C × P × (1−P) credited at the fill, worth up to $0.31 per 100 contracts. Takers clearing $250,000 in notional volume in the prior calendar month qualify for rebates paid weekly, and under a provision it calls Accelerated Tier Placement the exchange will assign a tier on verifiable proof of trailing-30-day notional volume at a competing venue. No combo carve-out appears, so a market maker quoting parlays is paid on one exchange and, unless the reported change lands, charged nothing on the other. Sportsbook operators have already taken market-maker roles on these venues, and rebates are what decide where they post size.
There is a structural difference, too, between the proposed products: Kalshi’s combos are described as priced through a request-for-quote process (RFQ) in which market makers quote the other side privately, and the trader sees only the best offer, rather than through a public order book. Polymarket’s documentation describes its RFQ as optional: combo instruments trade through the ordinary orders interface on their own book, RFQ orders enter that same book and can fill against resting liquidity, and legs can be bought or sold. Product roadmaps have been read out of API artifacts on these venues before – a reliable way to glimpse into the products’ future.
