I was reading some older Iowa market research and the result looks interesting. Across 964 election polls, the average polling error was 3.37 percentage points, while the market forecast error was 1.82 points. But newer research on the 2024 election also found big price differences between platforms and possible inefficiencies. So maybe crowd forecasts are useful, but not automatically reliable. What would you test before using this data in a real forecasting model? Calibration, liquidity, trade
CryptoAlerta — análise de criptomoedas e mercado em tempo real