Media Accountability

Tracking when news outlets cite prediction markets — and whether those markets were robust enough to report.

— | —
Citations (24h | 30d)
Total prediction market mentions discovered via GDELT news monitoring. Shown as last 24 hours | last 30 days.
—
Cite a Probability
Articles that quote a specific probability or price from a prediction market contract (e.g. “62% on Polymarket”).
—
Matched to Contract
Citations successfully matched to a specific Bellwether-tracked prediction market contract, enabling fragility analysis.
—
Outlets Monitored
Outlets citing prediction markets:

    Top Cited Topics

    Outlet Accountability

    Outlet CitationsNumber of times this outlet cited prediction market data. Shown as 24-hour count | 30-day count. % ReportablePercentage of this outlet's cited markets (last 30 days) that are Tier 1 (reportable) — meaning the market has sufficient liquidity and the cited probability is accurate. Higher is better. Avg $ to Move 5¢Average dollar cost required to move the price of cited markets by 5 cents (last 30 days). Higher values mean more liquid, harder-to-manipulate markets. Avg BrierAverage Brier score of resolved markets cited by this outlet over the last 30 days. Ranges from 0 (perfect accuracy) to 1 (perfectly wrong). Lower is better. Only computed for citations where the market has resolved. Tier Mix Platforms

    Citation Discovery

    We monitor news articles and TV broadcasts that mention prediction market platforms (Polymarket, Kalshi, PredictIt) using GDELT's global news index, NewsAPI, and Internet Archive TV News. Citations are collected daily and deduplicated by URL.

    Market Matching

    Each citation is matched to a specific Bellwether market contract using a three-stage pipeline: (1) direct URL extraction, (2) fuzzy text matching against our enriched market database, and (3) LLM-assisted selection from candidate shortlists. This determines which market the journalist was citing.

    Fragility Assessment

    Matched markets are scored on a 0–100 fragility scale based on four components: trading volume, orderbook depth (cost to move the price 5¢), bid-ask spread, and price volatility around the time of citation. Each market is assigned a reportability tier:

    Tier 1 — Reportable ($100K+ depth) Tier 2 — Caution ($10K–$100K depth) Tier 3 — Fragile (<$10K depth)

    Brier Score

    For resolved markets, we compute the Brier score: (probability cited − actual outcome)². A score of 0 means the cited probability perfectly predicted the outcome; 1 means it was perfectly wrong. This measures whether outlets cite markets that turn out to be accurate.

    Outlet Grading

    Outlets are ranked by the quality of markets they cite. Outlets that consistently cite deep, liquid markets (Tier 1) score higher than those citing thin, easily-manipulated markets (Tier 3). An outlet's "% Reportable" reflects what share of their citations referenced markets robust enough for responsible journalism.