The informational content of prediction markets: evidence from Polymarket Bitcoin contracts
Author
Olsen, Oliver Pegani
Term
4. semester
Publication year
2026
Submitted on
2026-05-31
Pages
56
Abstract
Prediction markets have become a large industry, but research on newer contract types has lagged. This thesis studies Polymarket price-target (touch-probability) contracts, where traders bet on whether Bitcoin will reach a specified price within a fixed window. Because these contracts reference a continuously traded asset with an active options market, their quoted probabilities may already be embedded in existing market prices. The core question is whether these contracts add to price discovery or mainly repackage information from the spot and options markets. The analysis covers 258 monthly barrier contracts traded between October 2024 and March 2026 and proceeds in three steps: (1) a Mincer-Zarnowitz calibration test to assess whether market prices are well-calibrated probability forecasts; (2) per-contract Granger causality tests at an hourly frequency to check whether Polymarket prices predict next-hour Bitcoin returns; and (3) a cross-sectional comparison of Polymarket prices with a closed-form one-touch barrier probability computed from the spot price, target (strike), time to maturity, and a volatility measure, using both implied volatility (from options) and realized volatility (observed price moves). The calibration test cannot reject that Polymarket prices are unbiased forecasts of outcomes. The Granger tests find no detectable predictive content for next-hour Bitcoin returns. A simple barrier-probability model explains over 95% of the cross-sectional variation in Polymarket prices under both implied and realized volatility, although the hypothesis of exact equality is rejected. Polymarket prices track the realized-volatility benchmark and sit below the implied-volatility benchmark, a gap consistent with the variance risk premium embedded in options (the extra compensation investors require for volatility) rather than independent information about Bitcoin. All three results replicate out of sample on 252 Ethereum contracts. Overall, prediction markets on financial assets appear to add little information beyond what is already reflected in the underlying spot and options markets.
Forudsigelsesmarkeder er blevet en stor industri, men den akademiske viden om nye kontrakttyper halter efter. Denne afhandling undersøger Polymarkets pris-mål (touch-sandsynligheds) kontrakter, hvor man satser på, om Bitcoin når et bestemt prisniveau inden for en fast tidsramme. Fordi kontrakterne bygger på et aktiv, der handles kontinuerligt og har et aktivt optionsmarked, kan de angivne sandsynligheder allerede være indlejret i eksisterende markedspriser. Afhandlingen spørger derfor, om disse kontrakter bidrager til prisdannelse, eller om de primært genpakker information fra spot- og optionsmarkederne. Analysen dækker 258 månedlige barrierekontrakter handlet mellem oktober 2024 og marts 2026 og gennemføres i tre trin: (1) en Mincer-Zarnowitz kalibreringstest for at vurdere, om markedspriser er velkalibrerede sandsynlighedsprognoser; (2) Granger-kausalitetstests pr. kontrakt på timebasis for at se, om Polymarket-priser forudsiger Bitcoin-afkast den næste time; og (3) en tværsnits-sammenligning af Polymarket-priser med en lukket-form one-touch barriere-sandsynlighed beregnet ud fra spotpris, mål (strike), tid til udløb og et volatilitetsmål, både med impliceret volatilitet (fra optioner) og realiseret volatilitet (observerede prisbevægelser). Kalibreringstesten kan ikke forkaste, at Polymarket-priser er ubiasede prognoser for udfald. Granger-testene finder ingen påviselig forudsigelsesevne for Bitcoin-afkast den følgende time. En simpel barriere-model forklarer over 95% af variationen på tværs af kontrakter i Polymarket-priser under både impliceret og realiseret volatilitet, selv om hypotesen om fuldstændig lighed forkastes. Priserne følger benchmarken baseret på realiseret volatilitet og ligger under benchmarken baseret på impliceret volatilitet, hvilket er i tråd med en variansrisikopræmie i optioner (den ekstra kompensation investorer kræver for volatilitet) snarere end selvstændig information om Bitcoin. Alle tre resultater replikerer out of sample på 252 Ethereum-kontrakter. Samlet set tyder fundene på, at forudsigelsesmarkeder for finansielle aktiver kun tilføjer begrænset ny information ud over det, der allerede er priset ind i spot- og optionsmarkederne.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords