Betting Markets Can’t Predict Elections Better than Polling Averages

This is mostly about this tweet with the graphic below attached:

So the argument is that the prediction markets were accurately cautious of the polling and were able to look at the polls and correct for error from 2016. And since the “Actual Dem Win” percentage was close to the “Betting Market Projection” is proof.

This has two specifics that are wrong, so first, the betting markets aren’t predicting margins, they’re predicting who will win, it’s a binary prediction. Some states may be closer by margin, but harder to win, because of elasticity in votes. So to make this comparison isn’t a 1:1 correlation.

Then, a more substantive reasoning than my first point, is that Prediction Markets aren’t predicting polling error, they are just simply GOP biased. This is most notable in “What will be the Electoral College margin in the 2020 presidential election?” Market, where Trump winning the Electoral College by over 280 Vote Margin is at 5 cents (5% likely to happen). I am writing this on November 15th. It should’ve never been at 5 cents, much well 12 days after the election. The confirmed Electoral College Margin is Biden +74, and yet Biden +60-+90 is only at 87 Cents (87% Likely).

It isn’t just this one market, Trump is still at 17% likely chance to win the campaign, 12 DAYS AFTER THE ELECTION. It’s just pure delusion, and they got lucky that it correlated with the actual vote count.

Ignore anyone who puts Prediction Markets as a legitimate source of Election Odds.