Polymarket Favorites Won 87% Of Primaries, But Lost 4%

Candidates with the highest reference prices on Polymarket won 238 of 273 U.S. Senate, House and governor primaries during the 2026 election cycle, according to Bitquery’s scorecard published Oct. 9. The candidates won 87% of the races.

The hypothetical strategy placed $1 on the favorite in every race and used Bitquery’s reference price to calculate the number of shares purchased. It lost 4 cents per dollar across the sample.

On Polymarket, a winning outcome share pays $1, while a losing share pays nothing. A share purchased for 90 cents produces a 10-cent gain before fees if it wins, while a loss removes the full 90-cent stake. The strategy also bought more shares when prices were lower, increasing the payout from a winning contract at a lower price.

Candidates priced at 90 cents or more won 177 of 182 races. Favorites priced between 50 and 90 cents won 71% of the time, with an average reference price of 77 cents.

Bitquery calculated reference prices from trades made during the 24 hours before 12 p.m. UTC on election day. When no trade was recorded in that period, it used the latest trade from the previous 30 days. For races decided in runoffs, it used the runoff date.

The prices were study references rather than guaranteed purchase prices at the cutoff. The methodology did not provide a complete adjustment for bid-ask spreads, slippage or trading fees. Polymarket’s current politics fee schedule lists fees from 0% to 1%, but it does not establish the fees charged on the historical trades in the sample.

The scorecard excluded nine markets. Some had settled before voting, some lacked usable prices and one had no settlement record. The analysis used Polygon trade data and excluded Polymarket’s U.S. app and other prediction markets, including Kalshi.

Bitquery cautioned that the primary results may not apply to November general-election markets, which typically have more trading activity and polling coverage than smaller local contests.