The Wisdom of the Market: Why Polymarket Outperforms Traditional Polls in Election Predictions

Original author: Haseeb ＞|＜, Dragonfly Managing Partner
Original translation: TechFlow

As the dust settles on the election, there is a story that the Wall Street Journal and the New York Times are not reporting. While the mainstream media was busy with the spectacle on TV and hesitant to predict the results of key swing states, Polymarket, the world's largest prediction market, had already made a judgment before midnight Eastern Time, saying that the probability of Trump's victory was 97%. This was even before the media announced the results of any swing state.

1. Throughout the election, Polymarket has always been one step ahead

I want to explain why this is the case, because from the feedback I received on Twitter last night, most people have a deep misunderstanding of this.

Polymarket does better than the media in two basic ways.

First, Polymarket was more accurate before the election. Let’s look at the pollsters and analysts. The election model based on the polls claimed that the election was a 50-50 situation, while Polymarket gave Trump a clear advantage - his chance of winning was set at about 62% before the election.

If you remember, the mainstream media ridiculed Polymarket for its different opinions. They thought that Polymarket should agree with the opinions of those who made the model! Obviously, this difference meant that Polymarket could not be trusted. They thought that Polymarket’s pricing was different because its user base was full of Trump-supporting crypto enthusiasts. It was funded by Peter Thiel and only foreigners traded on it. Since it was unregulated, they thought it must be rigged and big money was driving up the price of Trump. Such criticisms are endless.

Implicit in this criticism is a deep distrust of the market. As if the market cannot be trusted unless there is clear evidence of their reliability. Of course, if you really trust the market, you may no longer trust the media. The media’s business model is based on making you distrust other sources of information — otherwise, why would you keep clicking on their endless stream of attention-grabbing articles?

But anyone with experience in markets knows that it doesn’t matter what the markets are made up of, whether they’re made up of Republicans, Democrats, foreigners, or someone else. In fact, we know that JP Morgan uses Polymarket, as do some of the world’s largest hedge funds (most with non-US subsidiaries). It’s integrated into the Bloomberg Terminal and has been referenced on CNN. Yet the media talks about Polymarket as if it were just a platform similar to 4chan.

You need to know that Polymarket did $3.6 billion in trading in the presidential election. It was the largest election betting market in history, with an order of magnitude more volume than any other election market. This is far more important than the career prospects of any single prediction modeler. Markets work because there are so many interests riding on getting the answer right.

Those alleged biases—like crypto enthusiasts or non-Americans who support Trump—did not affect the market’s accuracy. (In hindsight, non-Americans may have been more sobering in predicting the election.)

But the identities of the participants don’t matter. Prediction markets aggregate information from so many different participants to come up with prices that transcend bias. Markets don’t care about ideology, they only care about whether the results are right.

The fact is that Polymarket is more accurate than any pollster or modeler.

Now, let me be clear: the difference between 60/40 and 50/50 sounds big, but it isn’t. Elections are inherently uncertain. According to high school statistics, if you want to tell whether a coin is rigged to be 60/40 instead of 50/50, you need to flip the coin more than 100 times to be 90% sure. The result of “Trump won this election” does not tell you whether the coin is 60/40 or 50/50.

The point I’m making is not that Polymarket is completely right and the forecasting models are completely wrong. In fact, the difference between them is not that big. What I want to emphasize is that the market has always priced Trump’s chances higher than the polls. Remember, the market knows what the polls and analysts are saying. The market incorporates all the available information, but Polymarket’s pricing is different from the pollsters. The only explanation the analysts can think of is that Polymarket is biased.

They don’t have the humility to consider that, perhaps, Polymarket might have something that the polls don’t capture.

Polls are not as accurate as they used to be. That’s now abundantly clear. Before the internet, polls were much more accurate. Back then, landline poll response rates routinely exceeded 60%. Today, they’re around 5%. That means pollsters are subject to huge sampling biases that can’t be corrected with simple statistical corrections. (Also, pollsters—as sellers of products with reputations to protect—often converge their forecasts to avoid being outliers, which affects the synthesis of poll results.)

In addition, Trump is a unique and divisive figure in American politics. Therefore, in three consecutive elections, we have seen the polls make a serious underestimate of his support, which is the so-called "shy Trump voter" effect.

Polymarket may think that the polls are missing some important information. Pollsters said that they have updated their models and made adjustments. Polymarket's response was: I don't believe it. And it turns out that Polymarket is right.

Again! Polymarket did not assert that Trump has a 90% chance of winning. 62% is not an absolute number because the election itself is full of uncertainty. What puzzles me is that the media has not shown the slightest curiosity about this discrepancy. Maybe Polymarket knows something that we don't know? Or, there is some information that we have ignored and this information is not reflected in the polls?

Remember, Trump was far ahead of polls nationwide, in both Republican and Democratic states. He won every single swing state and even the popular vote, which seemed incredible to most people.

Are you really confident that there is no other way to discover the true sentiment of tens of millions of Americans than relying on those traditional pollsters and good old internet surveys?

That’s what the markets teach us. Markets are smart, but they don’t explain why—they only show what happened.

2. This brings us to the second way Polymarket outperformed the media

Polymarket predicted the election results in real time, before the media did. The unpredictability of the markets was on full display on election night. Polymarket reacted fast and furiously before the results were announced in any of the swing states. According to Polymarket, the election was decided at midnight, but the mainstream media did not officially announce the results until 6 am the next morning. Why is this?

First, Polymarket discovered an important correlation that the mainstream media did not want to explain to its audience. Poll errors are rarely random; they are often correlated across states. So when traders see Trump significantly beat poll expectations in less competitive states like New York City (a classic Democratic state) or Florida (a classic Republican state), it means that the national poll error may be large.

Polymarket caught on quickly, realizing that the swing states were no longer competitive. At 11:30 p.m., Polymarket had already put Trump’s odds in Pennsylvania at 90%, even as only a fraction of Pennsylvania’s votes had been counted.

Prediction markets don’t wait for red tape or commentators’ analysis. They don’t care if the traditional ritual of waiting for the votes to be counted is broken. Remember the outrage when Fox News announced the Arizona results early in 2020 (which turned out to be correct)? Trump even threatened to boycott the channel. It reinforced the lesson that the major media must count the votes honestly and not be too smart.

Markets, however, don’t care about the drama; they only care about the results. It’s obviously very difficult to explain to CNN viewers that the election is over, that the polls in the non-contested states are too far off, that Kamala’s prospects are bleak, and that you should go to bed instead of waiting for the results in the swing states. This goes against the narrative that the media has been reinforcing for months. The public wants a simple, easy-to-understand story where everyone knows where the story is going — you wait until the results in the swing states come in, until a certain color bar crosses the 270-vote mark.

At 12:51 a.m., the New York Times is still displaying dramatic charts and headlines. By this time, Polymarket has already given Trump a 98% chance of winning.

So election watchers are staying up all night just for the media to complete its meaningless ritual of coloring in bars.

Polymarket traders are not bound by narrative, nor motivated to hype drama for ratings — they just give their verdict.

@shayne_coplan, founder of Polymarket, says the Trump campaign is watching Polymarket to try to better understand how to read the odds. The media even dared to complain about Trump declaring victory at 267 votes — by which time Polymarket’s odds were already so low they were showing 100%.

The beauty of markets is that they react quickly to new information. Traders who can integrate information the fastest are rewarded — profits. This is something that traditional media cannot do because they need to process events through multiple layers of interpretation, narrative construction, and internal politics (like Murdoch’s meddling in the 2020 Arizona results).

Polymarket’s decentralized nature bypasses all of this red tape, allowing information to flow freely and without interference.

Last night’s events are worth pondering.This election was a stark warning to the Democratic Party, a repudiation of the expert class, and an immune response to the arrogant media.

However, for Polymarket, this night was a perfect demonstration of its value. The lesson learned for me is: when something important happens in the world, skip the op-eds and go straight to the Polymarket odds. Disclaimer: I am an investor in Polymarket. I have long been passionate about prediction markets, and I am very relieved that their value is finally being recognized. Also, I may be wrong in some details because I am very tired, but you should get my point.

Original link