Discrepancy of the Day
I’ve been saying for a while that the US presidential election on November 5th is likely to be decided by tens of thousands of votes across a handful of US states. The main piece of evidence for this is Five Thirty Eight’s poll of polls.
As of writing, Five Thirty Eight has Trump on 46% and Harris on 48%. The margin of error in polls is usually considered to be 3%.
Recently, the prediction market Kalshi won a legal battle to operate unhindered in the US. Kalshi today has a 14% gap between the two presidential candidates:
It’s often said that people betting with their own money will be more accurate than polling. This election will be a fascinating test of this proposition.
It’s worth saying that betting markets aren’t a new thing, and like polling, there are plenty of examples of them getting the result wrong.
The two most obvious explanations are:
Hidden preferences. Voters not wanting to admit to pollsters that they want to vote for Trump.
Exuberrance and overconfidence on the part of Trump supporters.
In any outcome, it will be interesting to understand why this discrepancy has appeared.