Canonical events, venue-attributed listings: how W.E.T. standardizes prediction markets

Why prediction markets fragment

Section 5c(c) of the Commodity Exchange Act lets each regulated venue self-certify its own event contracts. That built an explosive market — and a fragmented one. Prediction venues can list their own contracts on the same real-world events, with different wording, prices, fees, liquidity, rules, and resolution sources, and no required consolidated tape connecting them. The same question about the world trades in silos at inconsistent prices.

No exchange can fix this: indexing a rival's listings is off-strategy for a venue, and a venue-run benchmark of its own contracts could never be independent. A neutral, cross-venue organizer has to sit above the order books. That is the layer W.E.T. builds: enabled venue adapters enter under one published, exchange-agnostic methodology, and the registry is designed to add each CFTC-regulated venue on those same terms as reliable access becomes available.

The canonical event graph

Standardization starts with identity: which venue listings actually refer to the same real-world market? Two contracts are treated as equivalent only when their outcome definition, observation window, threshold, resolution source, and material edge cases sufficiently align. Similar wording is not enough — two "Fed cuts in March" contracts that settle on different announcements are different markets.

The graph also records how non-identical markets relate: one contract can be a submarket of another, the inverse of it, conditional on it, or share a catalyst with it. Every mapping carries a confidence score and an audit state, and low-confidence automated matches never enter benchmark calculations without approval — a wrong mapping would contaminate everything downstream.

One number per market

Once listings are mapped to a canonical market, W.E.T. derives a single cross-venue probability. Each venue's price is read under a published rule — executable midpoints, a spread test against the venue's own book, and an exponential freshness decay — and the surviving venues are combined as a liquidity-weighted mean of those midpoints. Weight rises with resting depth and traded volume (both compressed, so a venue ten times larger is not ten times the reading) and falls as a book widens or goes stale. The result is one canonical read per market, with every input observation retained and dated.

That is consolidation v1, and it governs each published consolidated benchmark input today. Venue-attributed event-navigation snapshots are not silently passed through this benchmark layer. A v2 is in consultation: it combines the same midpoints in log-odds space rather than probability space, and extends the weight with venue reliability and mapping-confidence terms. The two agree closely near even odds and diverge where you would expect — on tail-priced markets where a deep book at 97c is currently dragged linearly by a thin book at 80c. It is implemented, replayed against history and unit-tested, but it is not wired to anything that publishes, and a build check enforces that separation until the consultation closes. When it is enacted the version stamp on every row changes with it, so no print is ever ambiguous about which construction produced it.

A hard rule sits under all of it: commercial relationships never affect the weights. Affiliate economics never reorder a board, weight an index, or hide a better price at a rival venue. Methodology changes go through published governance, not a growth meeting.

Divergence is preserved, not hidden

When eligible venues disagree inside the governed index engine, W.E.T. does not bury that measured disagreement inside the aggregate; the observation and divergence evidence remain separately auditable. The public event-navigation comparison board is currently withheld because its related listings do not yet carry canonical exact-match evidence. That status and its identity gate are shown at Kalshi vs Polymarket. Even a confirmed divergence is information, not an execution, equivalence, or return claim.

What standardization enables

The canonical layer is the foundation everything else stands on: the WET Indexes that combine related markets into measurable, competing views of the future; the World Event Dashboard that makes the whole board navigable; the Worldview Portfolios users build from those indexes; and the historical archive and index API that make the record citable. How the published odds themselves are computed and dated is documented in the methodology.

Everything on this page describes data reporting, not advice. Values are free to quote with attribution to W.E.T. (World Event Trading).