Oddbean

I think prediction markets are pretty neat. They were the primary way of predicting the outcome of elections prior to the invention of modern opinion polling. In recent weeks the polymarket for the US presidential election has noticeably diverged from the polling. The polls have stayed flat, it’s essentially a coin flip as to who will win. The prediction markets have broke to Trump with a 60% chance of winning. Is this a real shift that bettors are seeing vs the polls or is someone manipulating the market. You could make the argument that placing bets on your preferred candidate is a capital efficient way of shifting the outcome. Journalists use the markets to shape their coverage and will start writing articles about momentum when they see the prediction markets shift. Elections are very messy so it’s easy enough to find evidence of momentum in any direction if you’re looking for examples to fit the narrative. And the narrative about a candidate pulling ahead will lead to more press and focus and is likely to help the candidate get more votes on Election Day. All of this to say it appears one person is attempting to do this in favor of Trump. https://www.cryptopolitan.com/bettor-trump-harris-betting-pair-polymarket/

Thank you for this! I have had these same thoughts. I know people who act like prediction markets are some kind of collective super intelligence and I’m like, bro, lots of ways to manipulate them or for them to be distorted (participation bias, down bad cope, people in the jurisdiction in question generally being restricted form most only prediction markets etc etc)

It’s certainly true that it can be manipulated and I agree that it could be a good use of campaign money (assuming it’s possible for this money to be funneled into prediction markets) but traditional polls are no better indicators. The tools to ensure the integrity of the US elections have been long gone.

While I do think the polling is distorted as well, often for similar reason, I do think the integrity of our elections is sound. Our system is doing what it’s supposed to do, make the maximal number of people unhappy at the alter to gridlock and the maintenance of the status quo unless change is PAINFULLY, ABSOLUTELY and UNDENIABLY necessary.

I strongly disagree on the US having sound integrity for its elections. It’s probably one of the worst among Western democracies. The widespread use of voting machines running opaque proprietary algorithms is absolutely contrary to the principle of vote integrity. Exit polls were recently suppressed because they were being used to prove the fraud from the electronic counting of the vote. The lack of integrity of the voter rolls data which when combined with the recent generalization of mail-in voting and/or the possibility in many states to vote without presenting an ID, is another way to manipulate the outcome of the elections. US elections are nothing but secured. In the vast majority of Western democracies: - Voting machines are not allowed - Voting by mail is not allowed - Identity verification is required And this is just on the integrity of the voting process. The elections are rigged in many more ways including through the polls, the media, the two party system and so on. #ElectionIntegrity #USPolitics #USPol #Elections2024

More accurate than the current polling methods imo but yeah its a toss up. I think its more interesting that this is the first election in my lifetime that no matter who wins, the other side Will truly believe the other side didn’t steal it.

I can see what you mean Rabble. I studied a bit of data analysis recently and the central limit theorem instead uses the law of large numbers for a normal distribution bell curve in a hypothesis test of say an example - voting results. This requires a greater variety and numbers in sample size to avoid errors. The larger the sample size and use of confidence intervals, the more accurate the prediction. For example, in 2016, the polling sizes where just to small for the accuracy of those predictions. So the polling and this poly market could be diverging based on the law of large numbers. Which seems to be what the article discovered. Though, now you got me thinking about the rabbit hole 🕳️ of the data the media is using and their giant megaphone and the influence of changing those votes too🤦‍♀️

Interesting perspective. Totally plausible. The way I see it: they’re two totally different numbers. It’s almost like the poly market is a derivative of the actual polls. In the sense that supposed one candidate polled at 52% but had huge leads in most swing states, the betting markets would weight that heavily and could conceivably go to 70%+

It was also considerably skewed towards Harris before. And the centralized UK bookies e.g. give Trump even better chances than current Polymarket odds. Paddy Power has them at 8/13 and 11/8, which translates to 61.90% and 42.11% respectively.