Polymarket's historical accuracy is strong: ~90% a month before events, rising to 94% hours prior, per verified research excluding extremes. Criticisms note manipulation risks in low-liquidity political markets, with past election errors. Bank economic forecasts fare worse, accurate only ~23% of the time per Haas studies, often optimistic or off by 1%+ on GDP. Overall, Polymarket outperforms banks via crowd wisdom.