Research
Thinking on information markets, mechanism design, and the primitives that make them work.
Are Polymarket Traders Hedging on Trade.xyz?
Matching 338 million Trade.xyz fills against 63,264 Polymarket finance bettors. Of 1,413 wallets active on both venues, 31 held offsetting positions and only 4 did it twice.
Surprise Events and Tail Probabilities
540 Polymarket markets traded past 95% confidence and then resolved the other way. Sorting them into four kinds, and finding the one a different contract shape would reach.
Hedging That Needs Continuous Probability
Polymarket's March 2026 oil markets case study: how binary architecture fragments liquidity across strikes, why staircase hedges leave basis risk, and how a continuous market collapses 21 binaries into one position.
Binary Events: Does Liquidity Trade The Tails?
Splitting 18,863 Polymarket events into continuous and categorical slices and re-running the v1 analysis. Concentration is architecture-wide; ghost markets are a categorical phenomenon.
Binary Events: What Happens When You Split One Market Into Twenty
A metadata study of 36,777 Polymarket events: splitting one question into many binary contracts concentrates liquidity in ~5 markets, leaves ghost markets in the tails, and taxes the low-priced contracts.
The Yes Bias Might Not Exist
Analysing 7,292 resolved events and 28,793 Polymarket trades: the apparent YES bias looks like longshot preference channelled through how questions are framed.
Information as Supply
Prediction market TAM should include the supply side: as the cost of producing real-time probability estimates collapses, the addressable market widens far past trading volume.
Noisy Traders Are Not Dumb Money
The smart-money-vs-dumb-money framing misreads prediction markets. Drawing on NBER, the BIN model, and the Kapoor Quadrant, noise is a structural requirement, not a bug.
Prediction Market Resolution: A State Confirmation Problem
Prediction markets must agree on what actually happened. functionSPACE compares delegated (Kalshi) and collective (UMA, Reality.eth, Chainlink) confirmation.
The Prediction Market Economy
Prediction markets are under 0.5% of DeFi volume, but reframed as information finance they look early, not small. functionSPACE maps the four-layer prediction economy.
Information Vectors: an intro to composable beliefs
Beliefs aren't binary — they're directional, bracketed, distributional. Binary contracts force 1-bit expression and fragment capital up to 256×; composable beliefs trade on a shared surface.