His 2016 prediction was 70% chance for Clinton, and 30% chance for Trump. That’s substantially higher odds for Trump than any other notable prediction.
Yup. Disliking the guy is no reason for academic dishonesty though. He gave 70/30 odds. The key part about odds though is that they’re basically confidence ratings.
He was wrong and surprised, along with basically everyone else.
Guy came up with the method basically everyone uses to combine and aggregate polling data now, which is far more accurate than previous methods. It’s weird to say he’s an idiot.
Yes. He gave Trump like a 15% chance of winning which in a well calibrated model happens well 15% percent of the time. Which was larger than most other aggregators.
Good probabilistic functions are calibrated like that where the odds should be correct which includes the “loser” winning the amount of times its predicted. Before being bought out FiveThirtyEight did great retros on their sports bets to ensure that the winner did not win greater then predicted chance and the loser won at the correct rate. Its called model calibration
Also didn’t he predict Hilary Clinton would win in 2016?
His 2016 prediction was 70% chance for Clinton, and 30% chance for Trump. That’s substantially higher odds for Trump than any other notable prediction.
NateHate and not knowing what you’re talking about, name a more iconic duo XD
Ahaha, well put.
Yup. Disliking the guy is no reason for academic dishonesty though. He gave 70/30 odds. The key part about odds though is that they’re basically confidence ratings.
https://fivethirtyeight.com/features/why-fivethirtyeight-gave-trump-a-better-chance-than-almost-anyone-else/
He was wrong and surprised, along with basically everyone else.
Guy came up with the method basically everyone uses to combine and aggregate polling data now, which is far more accurate than previous methods. It’s weird to say he’s an idiot.
Yes. He gave Trump like a 15% chance of winning which in a well calibrated model happens well 15% percent of the time. Which was larger than most other aggregators.
Good probabilistic functions are calibrated like that where the odds should be correct which includes the “loser” winning the amount of times its predicted. Before being bought out FiveThirtyEight did great retros on their sports bets to ensure that the winner did not win greater then predicted chance and the loser won at the correct rate. Its called model calibration