Thread by @RealCarlAllen

Just a reminder for my friends in forecasting - @jhkersting @gelliottmorris @NateSilver538 @ReedForecasts and I'm really sorry if I missed anyone -

That 2012 is the best comp for 2020.

2016 being most recent, most memorable, makes it *feel* more comparable, but that's a bias.

https://twitter.com/RealCarlAllen/status/1305624347110002688

And while I'm here, I need to issue an apology to @gelliottmorris.

At the beginning of my deep, deep dive into politics/polling, I had a big problem with @TheEconomist Forecast.

He may or may not even remember (and I wasn't *that* big of an a-hole..) but I should elaborate

My beginning of analysis into polling & politics data came from sportsbook odds. I know how sportsbook odds work - and how people normally misunderstand how they work - and based on this I concluded it's more likely pollster/forecasters were wrong than the sportsbooks.

Why? Sportsbooks take money on this stuff. They have literally millions on the line. Bad odds can mean millions in losses. It didn't seem remotely possible that oddsmakers were wrong with $$$ on the line, and internet forecasters were right.

So I decided to run numbers myself.

And, to spare you all the long, boring analysis:

I think the sportsbooks are largely wrong. Less so today (-200 odds) than 30, 45 days ago (-110) but nonetheless, still pretty wrong. Sportsbooks are never this wrong - but this election is an exception for many reasons.

So thanks to @gelliottmorris for releasing your model early, enraging me with what I thought was an unreasonable and indefensible forecast, and to whatever it was about my upbringing that makes me want to crunch numbers on everything, only to realize..I was the one who was wrong.

So to all 9 people who are reading this:

Question your own assumptions. Challenge those who are considered experts. And don't forget, even when you are probably right, there's still a chance you're wrong, and don't stop learning until you know for sure.