Overconfidence

The tendency to state probabilities that are too extreme, saying 90% when the honest estimate is 75%. It is the most common and most costly forecasting bias.

Overconfidence quietly wrecks calibration and leads to oversized bets on things that aren't as sure as they feel. The fix is to imagine the event not happening and gauge how surprised you'd really be, and to track your calibration so the pattern becomes visible.