IFFA Signals · research

IFFA-EVIDENCE-0012894

An IFFA Signals evidence card: one claim, what supports it, and what it does not establish.

IFFA-EVIDENCE-0012894PARTIALLY VERIFIEDderived count

Seven days before close, India-related Polymarket binary markets (n=318 markets in 160 events, platform-reported volume ≥ $1,000) had a Brier score of 0.153 (95% event-cluster bootstrap interval 0.130–0.177), against 0.212 for always forecasting the sample base rate.

Source
ARGUS dataset calibration_india
Stored bytes (sha256)
9a34396f1417f6e0237b2e8f9d787803cd43a35b3bae3bc00d5fc646309716b1
Interpretation
Describes market-implied probabilities against platform-reported resolutions for this sample and period. It is not a property of prediction markets in general.

This does not establish

  • that market prices are objective probabilities
  • how Polymarket markets outside this India-related sample are calibrated
  • who moved any price, or why

Extracted facts

  • n=318
  • events=160
  • base rate=0.305
  • ECE=0.056
  • log loss=0.441

Evidence ids are permanent. A corrected claim keeps its id and changes status; see the corrections policy in the repository (signals/CORRECTIONS.md). Methodology