Coinbase reportedly targets in-house prediction markets by first half of 2027
CoinDesk, Decrypt and The Defiant report that Coinbase aims to launch its own prediction market contracts in the first half of 2027, citing a Bloomberg report. The posts say Coinbase currently offers event contracts through Kalshi, and that running its own markets would reduce reliance on that partnership and give it more control and a larger share of trading revenue. CoinDesk states the Kalshi partnership brought in $200 million in annualized revenue in the third quarter, and The Defiant says the planned contracts would serve retail and institutional customers.
Why “Confirmed”
- 3 independent outlets (CoinDesk, Decrypt, The Defiant) report it
3 independent source groups across 3 posts · evidence score 92/100. How labels work
What the sources claim
-
Coinbase is targeting the first half of 2027 to launch its own prediction market contracts.
“Coinbase targets first-half 2027 for its own prediction market contracts”
CoinDesk -
Coinbase's move would reduce its reliance on its Kalshi partnership, which brought in $200 million in annualized revenue in the third quarter.
“The move would reduce Coinbase's reliance on a partnership Kalshi, which brought in $200 million in annualized revenue in the third quarter.”
CoinDesk -
Running its own event contracts would give Coinbase more control and a bigger share of trading revenue than offering them through Kalshi.
“The exchange offers event contracts through Kalshi, but running its own would give it more control and a bigger share of trading revenue.”
Decrypt -
Bloomberg reports the planned in-house contracts would serve retail and institutional customers, targeting the first half of 2027.
“Bloomberg says the planned in-house contracts would serve retail and institutional customers, targeting the first half of 2027.”
The Defiant
Sources
More in Prediction Markets
Embed this page
Summary and claims extracted by AI with prompt analyze-story@5 on 2026-11-20 11:30 UTC. Labels are computed from the sources, not by the model.