Market methodology
THE DATA BEHIND THE DESK
Coverage at a glance
Catalog coverage, not election odds. One race can appear at multiple sources.
Races with verified contracts
Race pages by office
567politics tracks political prediction market contracts. Prices reflect trading activity and contract rules; they are not polls or a calibrated election forecast.
What the numbers mean
Kalshi prices use the midpoint of an available, valid yes bid and ask. Polymarket prices use the outcome prices supplied by its public Gamma API. Gemini and PredictIt use valid YES bid/ask midpoints. ForecastEx uses its most recently published daily closing price, with the file date shown; inactive contracts and rows with neither open interest nor traded pairs do not become live quotes. Missing or invalid prices display as unavailable. We retain each source’s question and outcome wording.
Sources stay separate
We do not average contracts or infer that similar questions have identical resolution rules. Review the linked source before comparing. Volume, liquidity, spreads, fees and settlement rules can affect prices.
Coverage and freshness
Race maps use a verified catalog of exact event, market and outcome identifiers, with the original resolution rules. Discovery is a separate, bounded feed; a missing contract never becomes an inferred party probability. Source failures appear individually. Fetch time is distinct from a provider’s last quote update. Provider data refreshes once per day at 12:00 UTC. Visits and chart requests read stored data without contacting providers. The freshness notice shows the last successful refresh for each source; failed refreshes retain earlier observations.
Maps and charts
State maps and the full 435-seat House cartogram show the higher-priced party at the selected provider. A difference of less than two percentage points is near-even. Both Democratic and Republican quotes must be available and unambiguous; multiple competing candidate contracts for one party do not become an inferred party price; missing data stays neutral. These are market favorites, not vote shares, called seats or a forecast of chamber control. Historical charts use Kalshi closing bid/ask midpoints and Polymarket outcome-price history; a scheduled daily job stores observations in the shared Postgres database. ForecastEx daily files provide up to a year of closing history where the contract existed; their file dates use a 12:00 UTC display anchor, not a claimed trade time. Gemini and PredictIt retain actual daily observations beginning with their first successful collection when their public responses supply no older history. Previously saved provider histories are retained. Missing days remain missing. Nomination odds and general-election odds remain separate.
The newsroom
Reporting uses the same editorial platform, accounts, comments and moderation as the Frog network. Market prices do not determine editorial rankings. Read our editorial standards for our approach to evidence and fair coverage.
Common questions about election odds
Are prediction-market odds the same as a vote-share forecast?
No. A contract price reflects trading on a specific outcome under that contract’s settlement rules. A price of 60% does not mean a candidate is expected to receive 60% of the vote.
How often does 567politics update market prices?
The scheduled collector runs once a day at 12:00 UTC. Page visits read stored observations. Check each source’s freshness notice because a failed refresh can leave older observations visible.
Why do different markets show different prices?
Contracts can differ in their questions, settlement rules, liquidity, spreads, fees and trading activity. 567politics shows each source separately and does not average them into a single probability.
How should I read a politician’s issue rating?
Each issue defines all ten steps of its own policy spectrum. A higher score describes a different policy position, not a better politician. Ratings require dated evidence; unreviewed positions stay blank.