Forecasters vs. prediction markets

Of all US elections, betting has probably never taken up so much of the debate as it is this year. Polymarket, in particular, has been hitting the headlines during the campaign, with a major gap in favor of Trump emerging around the start of October. A significant divergence between traditional election forecasts and prediction markets appeared before bookies followed.

Those inflows betting on Trump were, apparently, mostly the work of a solitary French trader, who claimed to simply be working from his own model with no political bias or preference for who won. We’ll probably not figure out whether that’s true, but it remains the case that whoever it was shifted the dial in the markets considerably, at a time when polls remained largely static.

The betting thesis for Harris, then is a simple one: forecasters and polls have the election at a 50-50, and prediction markets at the time of writing have Harris at 35%. Getting offered 13/7 on a coin flip is a no-brainer, and so you should put your money on Harris.

Essentially, it boils down to one question: who’s right, the forecasters or the prediction markets? 

Of course, the thesis hinges on the assumption that the forecasters are correct in their analysis, and the prediction markets are wrong, which creates the inefficiency that allows for profit. Essentially, it boils down to one question: who’s right, the forecasters or the prediction markets? 

Reliability

We’ll take FiveThirtyEight as the gold standard here, which is as good as you’re going to find for independent election modeling. Nate Silver’s old site has been remarkably consistent in predicting US presidential contests, and even Trump’s shocking victory in 2016 was given far more likelihood of happening on there than with other forecasters.

Accordingly, FiveThirtyEight has the election at 53-47 in favor of Harris, which is essentially a coin flip, given margins of error. Silver himself, who uses a slightly adapted version of FiveThirtyEight’s model, has the race very slightly more in favor of Trump, by less than a percentage point, at the time of writing.

polling companies and forecasters have far more skin in the game than random punters

The view that the prediction markets are inherently more reliable seems difficult to argue. Some have argued that they’re more accurate because people have their own money on it, but this is a deeply silly argument – polling companies and forecasters have far more skin in the game than do random punters.

Essentially, pollsters have solid methodology. Prediction markets don’t and aren’t trying to. Pollsters also have a solid track record, and prediction markets don’t, even over their more limited sample size. There just isn’t a convincing reason to back the wisdom of crowds over professional forecasters.

Market manipulation

The reasons behind the massive inflows we’ve seen backing Trump have been hotly debated. We’ll probably never know the percentages, but it’s likely a mix of market manipulation, betting syndicates, and sincere Trump voters believing their man has it locked down.

the problem is that there’s not really been any proof of manipulation

If there’s a lot of market manipulation going on, then it’s a no-brainer. The odds are being distorted and the probability isn’t, so Harris is clear value. The problem is that there’s not really been any proof of manipulation, and it seems a strange move in any case. 

The millions of dollars required to tip the scales of a market as liquid as Polymarket would surely have been put to better use being spent on campaigning in swing states if a Trump win was your aim. The money, of course, may come from sources who for various reasons can’t make traditional donations, but shifting the odds on Polymarket, which is unknown to the vast majority of Americans, seems like a questionable use of resources. 

There have been recent claims of wash trading by analysts, which does add a little bit more credibility to this idea. Overall, while it seems an unsound basis upon which to hang the entire thesis, we can probably factor it in to a limited degree.

Dumb money

Trump’s supporters have regularly been accused of behaving like a cult, and while some Democratic voters certainly have their own issues, they are far from flat-earth territory.

We should also remember that Polymarket is not a very old institution, if you’d like to call it that. There are even cultural issues here – America has a strong betting culture in general, despite the regulatory landscape, but there was a fair bit of surprise expressed by many Americans at the thought you might wager on an election. Doing so was seen as far more unusual compared to how it would appear in the UK, for example.

Polymarket’s bettors are still overwhelmingly going to be crypto natives

What this adds up to is that Polymarket’s bettors are still overwhelmingly going to be crypto natives, and that’s a market that is far more pro-Trump than a random sample of Americans. It’s also a space that, as you’ll know if you’ve spent any time there, is full of echo chambers, inexperienced gamblers, groupthink, and borderline cultish behavior.

Silver himself, appearing on “The Lock-In,” hosted by VSO’s own David Lappin and Dara O’Kearney, compared the event to Conor McGregor’s fight with Floyd Mayweather, where a tsunami of ill-informed money on McGregor made Mayweather a clear value bet. The thesis here would be the same – sometimes there’s so much dumb money being put on the table that smart hands can’t pick it up fast enough.

Smart money

This is perhaps the best thesis for the prediction markets being correct. The idea is that the money flowing in has been coming from betting syndicates, whose own modeling is showing that Trump is a significant favorite in the race.

This is the only realistic scenario in which Harris isn’t a smart bet, but it requires a few assumptions. First, that the data favoring a Trump win exists and is correct. Also, that it isn’t equally balanced out by syndicates who have come to a similar conclusion as this article. Even if they purely ran their own models, it seems unlikely that few of them would agree with Silver, or perhaps even consider Harris a favorite, thereby balancing out some of the Trump money.

At the end of the day, we know somebody is wrong, and it’s probably not the analysts.

Potential arbitrages have been appearing constantly between Polymarket and other bookmakers, proving that the market is anything other than efficient. The evidence we have available heavily favors the notion that it’s the prediction markets that are the source of it. At the end of the day, we know somebody is wrong, and it’s probably not the analysts. Harris seems like value to us.