Leveraging proprietary AI to uncover investment opportunities in emerging categories.
Prediction markets are emerging as one of the most expressive instruments of the next decade - a continuous price on the future of policy, technology, science, and culture. Polylight is purpose-built to capitalize on that emergence.
Our research stack ingests structured and unstructured signals at scale - pricing, news, on-chain flows, prior probabilities - and surfaces conviction the market has yet to fully discount.
Algorithms author and size positions with institutional risk controls. Capital is deployed continuously across hundreds of contracts, never crowded into a single thesis.
Polylight is focused on prediction-market venues with deep liquidity and meaningful informational asymmetry. Each category is modeled by a dedicated research line.
Identical outcomes trade at divergent prices across venues. Polylight monitors parity continuously, enters both sides of the spread, and harvests convergence — an edge that exists precisely because markets disagree.
| Contract | Δ pp | Status |
|---|
A single research stream, continuously cycling. Each stage is auditable, replayable, and instrumented at the contract level.
Pricing, filings, news, on-chain flows and prior probabilities, normalized at sub-second cadence.
Heterogeneous signals are embedded, time-aligned, and attached to a Bayesian prior for the underlying event.
Thousands of paths are simulated per contract. The model outputs a full distribution, not a point estimate.
Positions are scaled by conviction, correlation, and capacity - with hard caps at the strategy and venue level.
Orders are split across venues with slippage and depth-of-book constraints, then settled and journaled.
A short, durable list. We revisit them at every meeting.
Engineering Wealth