The betting markets currently have Donald Trump at about a 62 percent chance of winning Tuesday’s election. This has been a significant shift over the past few months since Kamala Harris became the presumptive nominee.
Now, you will hear from pundits and seasoned election experts that the betting markets are never wrong. And usually they aren’t. But anyone who has ever gambled in a casino, or knows the statistics behind it, understands that the house always wins.
I’ve heard the everyday bettor, the pundit and the highly knowledgeable news anchor all say “what do the betting markets know that we don’t? It must be something.”
Well, I’ve got news for you — it’s a load of hooey.
Do you really think the betting markets have special proprietary polls that the candidates can’t access themselves? I mean, why are the betting markets showing something like a 5 or 6 percent chance that the winner will not be inaugurated on Jan. 20? (Don’t get too excited, Tim and JD!) While that is pretty darn crazy, it isn’t based on any sort of special inside information.
The same goes for predictions that Trump is going to win. This is because you have to understand what drives these betting markets — the bettors. They are betting on nothing more substantial than the “vibes” that once made Harris’s victory seem inevitable earlier this summer.
The groups that are putting hundreds of millions in bets behind Trump, which in turn is pushing the markets toward his win, are using their own models that really don’t rely that much on polling, but rather the way social media is treating the candidates. They track things like comments, engagement around posts, and reach, especially in battleground states. But there’s a problem with these types of social media vibes — someone controls them.
When you create a new model that predicts who will win in the future, the best way to test it is to take all the same social media information from a previous election (2020), feed all that same data into and see whom it predicts as the winner. Ostensibly, when these modelers did that with their model, it correctly predicted that President Biden would win. Now, when they are feeding all the new data on social media (the engagements, likes, comments, etc.) into the model to predict who is going to win in 2024, it predicts Donald Trump…by a lot. Otherwise, they wouldn’t be betting $100 million on it. But there is a major flaw in their system.
We know that Twitter and Facebook interfered in the election by suppressing documents and information that would have hurt Joe Biden — think Hunter Biden’s laptop, the lack of scientific basis for COVID-19 restrictions and everything in between. It isn’t really a stretch to think that Republican speech was being suppressed in other ways across social media platforms.
But there has been a profound change in social media since the last presidential election. Elon Musk, a strong proponent of free speech, took control of Twitter/X in 2022, and as for Facebook, the discovery of its cover-ups most certainly has put a much harsher spotlight on it this time around.
What does this all mean? It means that those betting models predicting that Trump will win based on of the “vibes” of social media engagement may not actually be seeing an increase in Republican-Trump sentiment as much as a change in the social media ecosystem. The model thinks that social media is exploding for Trump, but it is actually only seeing more of the engagement that was being suppressed last time around, when Trump lost.
To add the cherry on top, would anyone really put it past Musk to juice the Twitter algorithm to show more pro-Trump materials in the lead-up to the election?
If that isn’t enough to dissuade you from betting, I would just add this: In 2016, Hillary Clinton was the 88 percent favorite to win on the Sunday before the election. If you had the bright idea of putting your life savings on the easy favorite in hopes of making a quick buck, you lost it all.
My gut still tells me that Trump will make one of the greatest comebacks in political history and win. But the betting markets, and those gamblers putting hundreds of millions of dollars on Trump, might just lose their shirts.
Liberty Vittert is a professor of data science at Washington University in St. Louis and the resident on-air statistician for NewsNation, a sister company of The Hill.
