07.11.2024
As we've seen with the U.S. elections, prediction markets aren't just forecasting outcomes; decentralized, crypto-powered platforms are capturing the pulse of public perception and reshaping our understanding of political forecasting.
In politics, it’s often said that form is substance, which becomes especially relevant when so much of the political sphere revolves around mastering the art of appearances.
Within this art of appearances, political marketing, and communication have reached previously unimaginable heights, especially with the amplification provided by social media. But there’s a less glamorous, perhaps cruder and rougher element, which plays a crucial role in the political game of appearances: polls.
Forecast models, projections, or simply surveys are tools to statistically determine that a candidate in a political race is ahead and has a higher chance of winning. The problem with this is that sometimes the polls have been “wrong.” In other words, after an electoral process, the winning candidate (often from the opposition) wasn’t the favorite in the polls. Why does this happen? Are statistical methods and forecast models broken?
Well, the answer is simpler than a deeper analysis might suggest: polls belong to those who pay for them. The mass media, with their own agendas, publish projections and offer the “hard data” from their polls to contribute to a narrative of victory for a candidate and serve their own interests. This is why it has become so common for polls to “miss the mark,” only for the opposite candidate to win. Polls are a centralized instrument of power.
The rise of prediction markets.
But what if a decentralized platform, accessible to most people with just a cell phone and internet connection, could allow people to place bets on a wide range of political scenarios and outcomes? It might sound like what it is: the ability to wager on whether a candidate will win or whether a certain scenario will unfold. This is the core of prediction markets like Polymarket, which faced its "trial by fire" during this year's U.S. election “Super Tuesday.” Perhaps to the surprise of many (or maybe no one), these markets turned out to be more accurate than both institutional and traditional media polls.
One of the aspects that give prediction markets a better ability to reflect reality is decentralization. Since they don’t rely on a centralized institution or serve a specific interest, platforms like Polymarket—which is powered by the Ethereum and Polygon blockchains and funded with USDC—are accessible to any user. Unlike Kalshi (another popular platform), Polymarket doesn’t operate officially in the U.S., but this can be easily bypassed with a VPN.
Another factor that provides consistency and accuracy to prediction markets is their liquidity and volume. The ability to buy and sell bets adds liquidity to the positions, while trading activity contributes to volume. Of course, this opens up the possibility of manipulation in low-volume, illiquid markets.
But what do we mean when we say that prediction markets have proven to be more accurate than other projections? For example, let’s look at what happened with polling in the U.S. election.
Before this, one might argue that this market could have been manipulated by Trump supporters—a group of tech-savvy Republicans or even crypto enthusiasts who still see him as an ally. Or we might speculate that, since foreigners can also place bets, it could be a manipulation scheme by a foreign power, as happened with the so-called "whale bet" that placed over $26 million in favor of Trump. However, it’s essential to note that this particular poll had a trading volume of over $3.6 billion, making it difficult to manipulate.
For example, the ABC News/Ipsos poll projected Kamala Harris as the winner by 3 points, while the NBC News poll and Emerson poll showed a technical tie between Harris and Trump. Polymarket, however, showed a variable lead for Trump at different points during the campaign, with a decisive break in his favor beginning in the second week of October.
Source: Polymarket
It’s also worth highlighting that this surge in Trump’s odds after this period coincides with Harris’s disastrous media tour. In this sense, and now with the results in hand, we can confirm that Polymarket was giving Trump a 60/40 chance of winning, moving beyond the 50/50 split that most mainstream polls were indicating. We could say that prediction markets weren’t afraid to pick a winner, unlike many other polls that remained neutral or showed a close tie. This is because prediction markets reflect the genuine interests of those placing their bets.
Another example can be seen with the odds for each candidate to win the so-called swing states—states that can be won by either the Democratic or Republican candidate. In this regard, The New York Times and Siena College poll projected Harris with a slight advantage to win most of the swing states, which would give her the necessary points to secure the electoral college.
Source: Polymarket
For example, in North Carolina, which according to The New York Times was projected to favor Harris, Polymarket showed a very different picture. Ultimately, Polymarket’s projection turned out to be closer to the outcome.
These are just a few examples, and delving deeper would likely provide a more complete vision of what has been an exercise in liberty and democracy. Much of the reason why prediction markets have passed this "trial by fire" is because they allow data and sentiment to be transmitted in real-time, without interference from communication or vested interests.
In the future, it would be exciting to see more crypto-powered decentralized applications that bring the power of perception, stakes at reality, and popular will into the open—without biases, and manipulation. In an era where there’s so much talk of a "post-truth" world, being able to get a real pulse on a truth more grounded in reality is profoundly impactful for everyone.