Read on Twitter
A quick thread about why I believe election models are better than betting markets:
The simplest answer: political prediction markets are sometimes just bad, full stop. See research:
http://bjll.org/index.php/jpm/article/view/1796
The simplest answer: political prediction markets are sometimes just bad, full stop. See research:
http://bjll.org/index.php/jpm/article/view/1796
If you read the paper above, it's pretty devastating to US markets.
US political markets can't even pass the basic test of 'don't leave huge, literally risk-free arbitrage opportunities lying around for months continuously'. They are not efficient markets or even close.
US political markets can't even pass the basic test of 'don't leave huge, literally risk-free arbitrage opportunities lying around for months continuously'. They are not efficient markets or even close.
In the case of Predictit and other US markets, the flaws are easy to spot. High fees, limited number of bettors by law, limited bet amount.
In foreign markets, the question is harder. I'm guessing it's mostly a case of an unsophisticated public + no large institutional players caring enough to correct them.
Should be noted that US $$ faces significant hurdles accessing foreign gambling that ensure 99% won't even try
Should be noted that US $$ faces significant hurdles accessing foreign gambling that ensure 99% won't even try
US markets are worse, but both US+foreign have huge flaws on state/national combos. I've seen instances where Biden is a bigger favorite in PA than to win overall, implying that Biden could somehow lose if he wins PA (and has 0% chance nationally when losing PA). Patently absurd.
Why trust the modelers? Well, Fivethirtyeight is proven to be well calibrated. They've made thousands of political predictions and you can check their body of work. When they call something a 80% chance, it does actually occur about 80% of the time. https://projects.fivethirtyeight.com/checking-our-work/
The behavioral aspects at play in 2020:
Dems have convinced themselves that the election is razor close despite a clear polling lead, as a coping mechanism.
Trump fans 'learned' in 2016 that polls are fake, shy Trump voters, MAGA MAGA MAGA.
(the polls were mostly fine in 2016)
Dems have convinced themselves that the election is razor close despite a clear polling lead, as a coping mechanism.
Trump fans 'learned' in 2016 that polls are fake, shy Trump voters, MAGA MAGA MAGA.
(the polls were mostly fine in 2016)
Political prediction is filled with
* Outright quacks
* People splashing around money based on which side they root for
* More emotion than math
* Politically motivated bettors
So if you ask me which I trust, I'm going with the modelers over the bettors and it's not close.
* Outright quacks
* People splashing around money based on which side they root for
* More emotion than math
* Politically motivated bettors
So if you ask me which I trust, I'm going with the modelers over the bettors and it's not close.
