Market vs JMM
Every JMM probability next to the market-implied price for the same question, once a verified comparator exists. A missing comparator is stated as a missing comparator, never estimated.
Worked example — Predictions with a captured market comparator: 0 of 11. No verified Kalshi or Polymarket checkpoint maps to a research question yet.
Why JMM doesn't show market odds next to its predictions yet
Private market observations remain outside the frontend until field-level display rights are approved and a contemporaneous quote is matched to a proof-backed JMM forecast. Rather than invent or republish prices, this board states the absence: no captured Kalshi or Polymarket checkpoint yet maps to any of the 11 registered research questions. A disagreement is calculated only after an entitled comparator passes that gate. The forecast tracker is the JMM side of this board on its own, with nothing waiting on a rights approval.
Why a market price is the right opponent
A forecast is only interesting against something that is hard to beat. A liquid prediction market has money on the answer, which makes it the cleanest available benchmark and a genuinely uncomfortable one. This board is the standing comparison: question, JMM probability, market probability, disagreement in percentage points. When JMM is wrong the loss column will be exactly as visible as the win column.
How the gap will be read
Disagreement is the absolute difference in percentage points, flagged at ten or more. A large gap is not automatically a JMM error: a market price carries liquidity, fees, and crowd noise, and a base rate carries none of those and no news either. The test is resolution. Each row settles on its official source, and calibration is judged on the settled history rather than on how bold the disagreements looked.
What Pro adds to comparisons
Captured comparators and easier side-by-side research are Pro features. The launch list sends one email at launch.