Forecaster Training

Vocabulary flashcards covering key terms, base rate examples, and foundational concepts from the Forecaster Training lecture.

Four Tips for Good Forecasting

The core methodology for effective forecasting: 1) Begin with base rates, 2) Hunt for good information, 3) Update your forecasts, and 4) Remember overconfidence.

Base Rates

Figures that refer to how often things happen or the relative frequencies of events, serving as an initial starting point for a prediction.

Good Information

Timely evidence relevant to a forecasting question that comes from a credible source and represents one or more important perspectives.

Disconfirming Information

Evidence or facts that contradict or attempt to prove wrong one's existing beliefs, which aids in making more accurate predictions.

Prediction Markets

Platforms such as Cultivate Labs, Good Judgment Open, Metaculous, and 538.com where forecasters can look for polls and related questions.

Updating Forecasts

The process of continually adjusting predictions over time (such as during a seven-day period) to converge on the truth as new events unfold.

Overconfidence

The tendency for individuals to overestimate what they know and how much they know, which is punished by scoring rules if assigning $100x$ certainty when incorrect.

Goldilocks Principle

A forecasting balance where one avoids being overconfident (too bold) or underconfident (too cautious), aiming instead for a forecast that is just right.

U.S. Hair Color Base Rates

The relative frequencies of hair color in the U.S. population: $85\%$ to $90\%$ brunette, $2\%$ to $12\%$ blonde, and $1\%$ to $2\%$ redhead.

U.S. Divorce Base Rate

The historical baseline rate showing that $45\%$ of couples in the United States end up divorced.

Certainty Accuracy Rate

The benchmark from studies indicating that people are correct only about $75\%$ of the time when they claim to be certain.