Treating prediction‑market prices as inputs to public forecasting models can create feedback loops: a prominent forecast influences market prices, which then get re‑ingested into the same or other forecasts, eroding independence and complicating statistical inference. High correlation between market signals and model outputs also makes it hard to estimate which source adds predictive value and risks overfitting to moving targets.
— If forecasters, journalists, and platforms start blending market prices into models without guarding against recursivity, public forecasts could become self‑reinforcing and distort political information flows.
John Masko
2026.04.12
85% relevant
The Polymarket episode is a direct example of prediction‑market feedback: a large financial pool ($14M) created incentives for actors to try to change or coerce the factual record (threats to journalist Emanuel Fabian) so that the market outcome would pay out, demonstrating how markets can alter events or reporting they are trying to forecast.
Nate Silver
2026.03.23
100% relevant
Nate Silver’s explicit policy — he consults for Polymarket but refuses to incorporate market prices into Silver Bulletin models citing recursivity and high correlation — is the concrete example raising this problem.
← Back to all ideas