Nov 15, 2026, 2:30 p.m. EST
3 min read
For at least the last half-decade, prediction markets have captured significant interest. Are they accurate? Do they offer valuable insights into electoral outcomes beyond what standard polls tell us? I remain uncertain. Should you have any perspectives to share, please reach out by emailing me.
You’re reading State of Crypto, a CoinDesk newsletter looking at the intersection of cryptocurrency and government. Click here to sign up for future editions.
Confidence over predictions?
The narrative
A sweeping “blue wave” election just took place. In its aftermath, social media platforms like X (previously Twitter) have been buzzing with debates over whether forecasting platforms like Kalshi and Polymarket outperformed traditional polling models. This conversation has surfaced persistently since at least 2020, when this publication’s preceding newsletter monitored what surveys and early prediction platforms projected for that year’s presidential race. Now, four election cycles later, it is safe to say that prediction markets have successfully secured a distinct place in the ecosystem.
Whether that space caters primarily to casual gamblers or offers genuine market intelligence remains an open question.
Why it matters
A chicken-and-egg dilemma is at play here. Did prediction markets merely track polling sentiment, or did they—true to their name—anticipate where polling sentiment was heading?
Personally, I have started viewing prediction markets less as direct reflections of polling sentiment and more as gauges of confidence in a particular result. While that does not definitively resolve the question above, I have found it a helpful perspective for my own reporting work. To be clear, I have monitored prediction markets for quite a while now. Even so, I wouldn’t say they have supplanted polling data or other traditional instruments within my personal daily workflow.
So, how did they perform during this recent election?
Breaking it down
Generally speaking, prediction markets and public opinion polls aligned across most major races tracked by CoinDesk, matching the final outcomes with only a few exceptions.
Take Wisconsin’s First House District, where Republican Rep. Bryan Steil lost to Democratic challenger Mitchell Berman. Both the polls and the prediction platforms had the contest locked in a virtual tie on the Friday prior to the vote.
In Alabama’s Second House District, incumbent Rep. Shomari Figures triumphed over challenger Rhett Marques. Both polls and forecasting markets had previously pointed to Marques holding an advantage over Figures within the newly redrawn district.
Moving to the Iowa Senate race, State Rep. Josh Turek unseated Rep. Ashley Hinson to flip the seat. While prediction markets favored Hinson, aggregated polling data pointed toward a dead-even 50/50 split.
Outside of these three exceptions, both forecasting markets and public polling successfully predicted the winners of the primary races followed by CoinDesk.
Other contests barely registered on the radar of either polls or prediction platforms.
Polymarket’s contract for the Mississippi Senate race leaned toward incumbent Republican Cindy Hyde-Smith until actual ballot counts began reporting on Tuesday evening. Kalshi did not offer any contract for that specific contest.
Even fewer details were accessible for South Carolina’s Senate race. Neither Polymarket nor Kalshi featured a contract regarding the final winner, though Polymarket did provide a parlay option covering the victors of the Montana, South Carolina, and Mississippi Senate races combined. Representatives for both businesses declined to comment when asked about these specific contests.
Both of those races turned into unexpected upsets as Democrats Scott Colom and Annie Andrews respectively secured their victories.
Nor was traditional polling any more illuminating. According to archived poll data from The New York Times, the most recent South Carolina race had Senator Graham ahead by just a single point. No polls for Mississippi were hosted on the Times website, though alternative platforms referenced numbers showing Senator Hyde-Smith holding a slim lead.
I will leave the conclusion of this newsletter as a prompt for you, my reader: How ought we to interpret prediction markets, particularly when considering the apparent lack of intelligence—polling or otherwise—surrounding these unexpected upset victories? I am eager to hear your thoughts!
Originally published at https://www.coindesk.com/policy/2026/11/15/did-prediction-markets-call-the-u-s-midterm-election-state-of-crypto.