Forecasting
Business
10 rabbit holes · 5 people exploring
Common starting questions
- How do you forecast something nobody has data on?
- If the forecast said 70% rain and it didn't rain, was it wrong?
- If I'm outside the cone am I safe?
- Why is the average of a crowd's guesses so good?
- Who made the first weather forecast?
Rabbit holes
- the Delphi method is named after an oracle and came out of RAND
RAND in the 1950s needed forecasts about things like future technology where there was no data, only expert opinion. Their fix was anonymous rounds: experts estimate, see the group's spread, revise, repeat, so the most senior person in the room doesn't anchor everyone.
Tomás Rinaldi · 6 sources
- Brier scores, or how to grade a weather forecast
Ryo Nakamura · 5 sources
- the hurricane cone doesn't mean what most people think
Tomás Rinaldi · 5 sources
- 800 people guessed an ox's weight and the median was within 1%
1906, a livestock fair in Plymouth. Galton collected the tickets from a weight-guessing contest, supposedly expecting to show how bad ordinary people were at judging. The median guess was 1,207 lb. The ox was 1,198. It only works if people guess independently though; once they can see each other's answers it falls apart.
Pete Lindgren · 5 sources
- the first numerical weather forecast was done by hand and was spectacularly wrong
During WWI, while driving ambulances in France, Richardson worked out a six-hour forecast for a patch of central Europe by hand. It took weeks and predicted an absurd pressure change. But the method was basically right; the starting data needed smoothing. He also imagined a 'forecast factory' of 64,000 human computers in one giant hall.
Tomás Rinaldi · 6 sources
- Kahneman's team estimated two years. it took eight
His team writing a school curriculum all guessed about two years. Then he asked their most experienced member how long similar teams had taken: about 7 to 10 years, and roughly 40% never finished. They kept going anyway. It took eight years, and by then nobody used it.
Arjun Pillai · 6 sources
- experts vs dart-throwing chimps, and the amateurs who beat both
Tetlock tracked tens of thousands of predictions from experts over almost two decades. On average they did little better than chance, and the more famous ones did worse. Then he ran a tournament and found a small group of amateurs who were consistently, measurably good. Mostly they broke questions into parts, started from base rates, and updated in small steps.
Tomás Rinaldi · 6 sources
- the Beagle's captain invented the weather forecast
FitzRoy coined 'forecast' in the weather sense, set up storm warnings for ports after a terrible storm in 1859, and got the first daily forecasts printed in The Times in 1861. Then he got mocked in the papers every time they were wrong.
Nikolai Volkov · 6 sources
- a university has run a real-money election market since 1988
The University of Iowa's business school, tiny stakes (accounts capped at $500), started for the 1988 presidential election. It's a research and teaching tool and has been running for decades.
Arjun Pillai · 5 sources
- Paul the Octopus went 8 for 8 and it means nothing
Germany's psychic octopus picked the winner of all seven of their 2010 games plus the final. The odds of that by pure chance are 1 in 256, which sounds impressive until you count how many animal oracles there were.
Tomás Rinaldi · 5 sources