A quantitative data layer for prediction markets
Why prediction markets need a serious data layer — and what Oddsbase is building on top of one.
Thesis
Prediction markets are becoming liquid enough to study like financial venues, but the underlying data is still fragmented across exchanges, APIs, websocket feeds, and historical archives — often with inconsistent naming, partial history, or no coverage metadata at all.
Serious work on these markets requires normalized instruments, reproducible event timelines, complete trade and book history, and explicit metadata describing what is — and isn't — covered.
The platform
Oddsbase is building a quantitative data layer for prediction markets: normalized market instruments across venues, full live and historical trade and order-book history, and explicit coverage metadata for every dataset.
On top of that layer sit insight tools that surface microstructure, liquidity, and pricing inefficiencies — for quants, researchers, and intelligence seekers who need more than a chart.
The research lab
The research lab publishes alongside the platform. Each note connects back to the underlying data — market IDs, time ranges, endpoints, and methodology — so any finding can be reproduced, verified, and extended.
Early work will focus on Polymarket, Kalshi, and crypto coverage; live-vs-historical consistency; liquidity around resolution events; and the building blocks for reproducible studies on prediction markets.