Professional forecasters tend to change their economic predictions too much. This hides signs of both overreactions and underreactions to public information. Traditional models can't explain this, but new models suggest forecasters are too confident in their data. This, combined with how public signals are received, can explain the patterns seen in economic predictions.