Overconfidence

Believing your estimates are more accurate than they are, most visibly as confidence intervals that are far too narrow.

The classic demonstration asks people for ranges they are 90% sure contain the answer; the true value falls outside the range far more than 10% of the time.

For an investor the consequences are concentrated positions, insufficient margin of safety, and forecasts stated to a precision the method cannot support.

Calibration is trainable. Recording predictions with probabilities and scoring them is the only method known to work.