Prices are compressed opinions. That is true in equities, rates, and event contracts — but the compression algorithm differs: who participates, what they can trade, when they can trade, and how resolution works.

I care about the plumbing between those venues because it is a pure systems problem dressed as finance: clocks, identifiers, missing data, and the eternal risk of storytelling.

Two markets, different contracts

A prediction market often asks a binary or categorical question with a defined resolution source. A financial instrument usually embeds a bundle of cash flows, risk premia, and constraints that never resolve to a tidy “yes/no.”

Mapping one onto the other is an ontology problem first. If the event and the instrument are only loosely related, any lead/lag study is fan fiction with a correlation coefficient.

Time is the hard part

To compare series you need:

  • A shared clock (UTC, always)
  • Explicit policies for gaps and halts
  • Awareness of venue hours and liquidity deserts

It is easy to invent a “signal” that is really just one market being closed while the other updates.

Causality is mostly out of reach

Even with clean data, “A moved, then B moved” is not causation. Common drivers — news, macro prints, social amplification — hit multiple venues at once. The honest output of early research is often a better question, not a trading rule.

What I build toward

Observation tooling: storage with provenance, simple transforms, and notes that separate hypothesis from result. See the project write-up: Prediction Markets ↔ Financial Markets.

A working stance

Be curious. Be slow to claim. Prefer infrastructure that makes the next experiment cheaper than the last.

TODO: When you publish a concrete case study, link it here and keep the claims scoped to what the data supports.