Quantitative#084

Kalman-Filter Pairs on Cointegrated Outcome Tokens

Two outcome tokens that move together (e.g. two candidates in one primary, or BTC-above-X vs ETH-above-Y) usually trade at a stable price relationship. A Kalman filter is a running estimator that keeps re-fitting that relationship as new ticks arrive, so its ratio adapts faster than a fixed formula. You trade the spread between the two tokens whenever it drifts too far from the filter's current estimate, betting it snaps back. The edge: the on-chain order book is slow to reflect the true co-movement that the recursive estimate captures first.

What you need to run it

Tick history per token to estimate state-space params
Kalman/Bayesian filter implementation with online beta update
Cointegration screening (Johansen/Engle-Granger) across token pairs
Capital to hold both legs plus inventory buffer

Where this applies

Markets on Polymarket where kalman-filter pairs on cointegrated outcome tokens is the natural play:

Will Candidate A win the primary? vs Will Candidate B win the same primary?
Will BTC close above $150k by Dec 31, 2026? vs Will ETH close above $7k by Dec 31, 2026?
Will the Lakers make the playoffs? vs Will the Clippers make the playoffs?

Capabilities this demands

Model / quantCustom code / APIData ingestionSignificant capital

At a glance

CategoryQuantitative

Requirements4

CapabilitiesModel / quant, Custom code / API, Data ingestion, Significant capital

VenuePolymarket (CLOB, Polygon)

Build it

Related quantitative strategies

This is documentation, not advice. Poly Research & Robotics publishes how these strategies work because the method should be checkable — not as a recommendation to trade them. See the full strategy database (297 strategies) or the data feeds directory.