Original title: The Super Bowl of prediction markets
Original author: Scott Duke Kominers, a16z crypto
Original translation by: Saoirse, Foresight News
On February 8th, US time (7:30 AM on February 9th, Beijing time), hundreds of millions of NFL fans watched the Super Bowl on their screens. Many of them were also watching another screen at the same time—keeping an eye on the trading dynamics of the prediction market, where the betting categories are all-encompassing, from the champion and final score to the passing yardage of each team's quarterbacks.
Over the past year, the trading volume in the US prediction market has reached at least $27.9 billion, covering a wide range of topics, from sports results and economic policy formulation to new product launches. However, the nature of these markets remains highly controversial: are they trading activities or gambling? Are they news tools that gather public wisdom, or methods of scientific verification? And is the current development model the optimal solution?
As an economist who has long studied markets and incentive mechanisms, my answer begins with a simple premise: predicting the market is, in essence, the market. And the market is the core tool for allocating resources and integrating information. The operating logic of the predictive market is to introduce assets linked to specific events—when the event occurs, traders holding the asset can obtain returns, and people trade based on their own judgment of the event's outcome, thus realizing the core value of the market.
From a market design perspective, referencing prediction market information is far more reliable than listening to the opinions of a single sports commentator, or even looking at Las Vegas betting odds. The core objective of traditional sports betting organizations is not to predict match outcomes, but rather to "balance betting funds" by adjusting odds, attracting funds to the side with lower betting volume at any given time. Las Vegas betting aims to get players willing to bet on unexpected results, while prediction markets allow people to make trades based on their own genuine judgment.
Prediction markets also make it easier to extract useful signals from massive amounts of information. For example, if you want to predict the likelihood of new tariffs, deriving this from soybean futures prices would be very indirect—because futures prices are influenced by multiple factors. But if you directly ask this question in a prediction market, you can get a more intuitive answer.
The earliest form of this model can be traced back to 16th-century Europe, when people even placed bets on the "next pope." The development of modern prediction markets is rooted in the theoretical frameworks of contemporary economics, statistics, mechanism design, and computer science. In the 1980s, Charles Pratt of Caltech and Shyam Sand of Yale University established its formal academic framework, and soon after, the first modern prediction market—the Iowa Electronic Markets—was officially launched.
The working mechanism of prediction markets is actually quite simple. Take, for example, a bet on whether Seattle Seahawks quarterback Sam Darnold will pass the ball within one yard of the opponent's end zone. The market would issue corresponding trading contracts, and if the event occurs, each contract would pay the holder $1. As traders continuously buy and sell these contracts, the market price of the contract can be interpreted as the probability of the event occurring, representing the traders' overall assessment of the outcome. For example, a contract priced at $0.50 means the market believes the probability of the event occurring is 50%.
If you judge the probability of an event occurring to be higher than 50% (e.g., 67%), you can buy the contract. If the event ultimately occurs, your contract, purchased for $0.50, will yield a profit of $1, resulting in a gross profit of $0.67. Your purchase will drive up the market price of the contract, and the corresponding probability valuation will also increase. This sends a signal to the market that someone believes the market is currently underestimating the likelihood of the event. Conversely, if someone believes the market is overestimating the probability, selling will lower the price and probability valuation.
When forecasting markets function well, they demonstrate significant advantages over other forecasting methods. Opinion polls and questionnaires can only yield percentages of opinions; to convert these into probability estimates, statistical methods are needed to analyze the correlation between the survey sample and the overall population. Furthermore, these survey results are often static data at a specific moment, while forecasting market information is continuously updated with the addition of new participants and the emergence of new information.
More importantly, prediction markets have a clear incentive mechanism, and traders are all "hands-on." They must carefully analyze the information they possess and only invest funds and bear risks in areas they understand best. In prediction markets, people can convert their information and expertise into profits, which incentivizes everyone to actively seek deeper understanding of relevant information.
Finally, prediction markets have a far wider reach than other tools. For example, someone might possess information about factors influencing oil demand and profit by long or short on crude oil futures. However, in reality, many outcomes we want to predict cannot be achieved through commodities or stock markets. For instance, specialized prediction markets have recently emerged, attempting to integrate various assessments to predict the timeline for solving specific mathematical problems—information crucial for scientific advancement and a key benchmark for measuring the level of artificial intelligence development.
Despite their significant advantages, prediction markets still face numerous challenges in realizing their full value. Firstly, at the market infrastructure level, there are persistent questions that need clarification: How can we verify whether an event has actually occurred and achieve market consensus? How can we ensure the transparency and auditability of market operations?
Secondly, there are challenges in market design. For example, there must be participants with relevant information to trade—if all participants are ignorant, market prices cannot convey any valid signals. Conversely, various participants with different relevant information need to be willing to participate in trading; otherwise, the valuation of the prediction market will be biased. The prediction market before the Brexit referendum is a typical example of the opposite.
However, the entry of participants with absolute insider information can also create new problems. For example, the Seahawks' offensive coordinator clearly knows whether Sam Darnold will pass the ball within one yard, and can even directly influence the outcome. If such individuals participate in trading, market fairness will be severely compromised. If potential participants believe there are insider traders in the market, they may rationally choose to leave, ultimately leading to a market collapse.
Furthermore, prediction markets are also susceptible to manipulation: some may turn this tool, originally intended to consolidate public opinion, into a means of manipulating public discourse. For example, a candidate's campaign team might use campaign funds to influence the valuation of prediction markets in order to create an atmosphere of "victory in sight." Fortunately, prediction markets possess a certain degree of self-correction in this regard—if the probability valuation of a contract deviates from a reasonable range, traders will always choose to operate in the opposite direction, allowing the market to return to rationality.
Given these risks, prediction market platforms must focus on improving operational transparency and clearly disclosing the rules for each stage, including participant management, contract design, and market operation. If these issues can be successfully resolved, we can foresee that prediction markets will play an increasingly important role in the future of prediction.
