[aeternum]

Did FiveThirtyEight really get that much right other than the 2016 election?

I remember thinking they were the best data journalists out there, and they had some nice visualizations but did their other predictions actually hold up?

[akio]

Yes, they forecast thousands of U.S. elections and thousands of sports games, and their forecasts had excellent calibration—e.g., events they said would happen 30% of the time actually happened about 30% of the time.

https://archive.is/sId82

https://github.com/fivethirtyeight/checking-our-work-data