,
5 tweets,
2 min read
Read on Twitter
The Taleb insult, a rite of passage for Twitterers
Calibration (skill at estimating how likely things are) is part of good judgment. Another part is being good at estimating value of payoffs.
Expected Value = P(success) * Value of success + P(failure) * Value of failure
Expected Value = P(success) * Value of success + P(failure) * Value of failure
Something can be worth trying if it has P(success) of only 0.01, as long as the payoff is large
But you still need calibration in estimating P(success). If you mistakenly think P is too low, you'll pass on the bet. If you mistakenly think P is too high, you'll wager too much $.
But you still need calibration in estimating P(success). If you mistakenly think P is too low, you'll pass on the bet. If you mistakenly think P is too high, you'll wager too much $.
I share Taleb's view that skin in the game filters for people who are better at estimating probabilities
But "reputation" is a form of skin in the game (if your audience cares about your accuracy). So verbal probabilities can work, if you lose reputation for being overconfident.
But "reputation" is a form of skin in the game (if your audience cares about your accuracy). So verbal probabilities can work, if you lose reputation for being overconfident.
Oh he also said this (screenshot).
Sure, you have to pay attention to the domain in which people are making predictions. A guy who's really good at predicting whether the sun will rise isn't necessarily going to be good at political forecasts!
Sure, you have to pay attention to the domain in which people are making predictions. A guy who's really good at predicting whether the sun will rise isn't necessarily going to be good at political forecasts!
