Market Psychology

Ten price ranges, each one given with “90% confidence”. Only 5 of 10 contained the actual value. A calibrated forecaster would hit about nine; hitting half means the ranges should have been much wider. This is overprecision, the most robust form of overconfidence, and you only see it by scoring a batch of forecasts.

• Barber & Odean (2000) found the 20% of households that traded most earned 11.4% a year net, against 17.9% for the market — but trading costs, not bad picks alone, did much of the damage.
• “93% think they are above-average drivers” comes from a small U.S. student sample (Svenson 1981); the Swedish sample was 69%. The effect is real, the number is not universal.
• Confidence is useful when it is earned and calibrated; the fix is scoring, not self-doubt.

fc = pd.DataFrame({                              # ten ranges you were "90% sure" of, made up
    'low':    [95, 40, 1.8, 210, 70, 12, 3.1, 150, 88, 25],
    'high':   [105, 48, 2.2, 240, 80, 15, 3.5, 170, 96, 30],
    'actual': [108, 44, 2.5, 236, 64, 13, 3.4, 181, 90, 31]})
hit_rate = fc.actual.between(fc.low, fc.high).mean()

Assumes import numpy as np and import pandas as pd. Each snippet carries its own example numbers; the comments say which are made up.