Overall, I was clearly overconfident:
- 60% Predictions: 50% (Right 7, Wrong 7)
- 80% Predictions: 78% (Right 15, Wrong 4)
- 90% Predictions: 66% (Right 2, Wrong 1)
- 99% Predictions: 91% (Right 10, Wrong 1)
Most of the error came from two sources.
First, vaccines had just come out and looked great at the time that I was making predictions. I definitely failed to put sufficient weight on a variant coming out and making them worthless as far as preventing spread, which is what happened with Omnicron. So I failed pretty hard there -- the one 99% failure, as well as some errors in the 80% / 90% buckets, come from this failure to consider the hypothesis.
Second, when predicting crypto prices, I generally predicted a correction, because prices had recently surged. This correction did not occur. The correlated error here accounts for a lot of my other errors.
All in all, I think I generally need to spend more time trying to think of ways I can be wrong, and be aware of heuristics that I'm using.
Percentage buckets: 60%, 80%, 90%, 99%
Economic, Macro
- (Wrong) 60%: Bitcoin ends the year lower than it starts (28,590 USD).
- (Wrong) 60%: Ether ends year lower than it starts (730 USD).
- (Wrong) 60%: Gold ends year higher than it starts (1,907 USD).
- (Right) 60%: Dollar ends lower against yuan than started (6.53).
- (Wrong) 60%: US Covid-19 infection rates < 0.1% current (260k per day) at end of year.
- (Right) 80%: Unemployment rate ends year lower than starts (6.7%).
- (Right) 80%: Bitcoin ends year above 8k USD.
- (Right) 80%: Bitcoin ends year below 56k USD.
- (Right) 80%: Ethereum ends year above 200 USD.
- (Wrong) 80%: Ethereum end year below 1400 USD.
- (Wrong) 90%: US Covid-19 infection rates < 1% current (260k per day) at end of year.
- (OMG So Wrong) 99%: US Covid-19 infection rates < 10% current (260k per day) at end of year.
Technical + Misc
- (Right) 80%: SLS is not canceled.
- (Right) 80%: Tesla does not complete coast-to-coast autonomous drive.
- (Right) 80%: Tesla opens up FSD beta to everyone in US.
- (Wrong) 80%: Starship superheavy booster does test flight.
- (Wrong) 80%: Starship does test flight to > 20k.
- (Right) 80%: OpenAI publishes or opens new interesting stuff about cross-modal learning.
- (Right) 90%: More hardware blows up while working on Starship.
- (Right) 99%: No plausibly AGI system goes public.
- (Right) 99%: MIRI releases nothing of immediate technical use.
Political
- (Right) 80%: Trump is not in jail.
- (Right) 99%: Trump is not president / dictator.
Health
- (Right) 60%: My hand is better at EOY, no typing problems
- (Wrong) 80%: My knee is better at EOY, able to exercise bike sometimes.
- (Right) 60%: I will not have learned to code by voice
- (Right) 80%: My plantar fascitis is NOT 100% better.
- (Wrong) 60%: Have gotten heartrate to 60 BPM or below for a day at least.
Meta
- (Right) 80%: No post on site has > 1000 views.
- (Right) 99%: This post has < 1000 views.
- (Right) 99%: No post on site has > 100000 views.
Goals / Personal
- (Right) 60%: ██████████ █████ ██ ███████ ██ ██ ██ ██
- (Right) 60%: Blog is still updated in > 6 months.
- (Right) 60%: Have not finished book.
- (Right) 90%: ██ ████████ ███ █████████████████ █████████
- (Right) 99%: ████ ████ ███ █████████████████████ ██ █
- (Right) 99%: ██████████████████████ ██ ██████
- (Right) 99%: ██████████
- (Right) 80%: ███████████ █ █████ ████████ █
- (Right) 99%: ████ ███████ ██ █ ██ █████
- (Right) 80%: ██████████ ███████████████████████ ███████████ ██
- (Wrong) 60%: ███████ █████████ ███ ██ ██ ███ █████████ █████
- (Right) 80%: ███ ███████ █ █ █████
- (Right) 80%: █ ████ ████ █ █████████ ██ █████ ██ ██ ██████ ██
- (Right) 99%: █████ █████████████ ████ ██ ██ ████ █
- (Right) 60%: ███ ████ ███████ ████ ██████ ███████ ███
- (Wrong) 60%: ████ ██████ █ █ █ ████████