Author: Scott Duke Kominers, a16z crypto Research Partner
Compiler: Chopper, Foresight News
Prediction markets allow users to trade on the outcomes of various events. These platforms began scaling up in the United States last year, and their coverage now spans from geopolitics to entertainment award winners. But what exactly is a prediction market?
As an economic researcher who has long studied market mechanisms and incentive systems, my answer is simple: Prediction markets are, at their core, just ordinary markets. Markets are fundamental tools for allocating resources, enabling goods and services to flow to those who need them most. In this process, markets also aggregate information: the supply-demand clearing process integrates information held by all participants and converts it into signals like prices.
Prediction market platforms and related products directly leverage this information aggregation capability to forecast the direction of specific future events. Platforms issue assets tied to specific events; holders earn a payoff if the preset outcome occurs. Users trade these assets based on their judgment of the event's probability. For a long time, many companies have used prediction markets to tap into employees' implicit knowledge to assess whether key products would launch on schedule. Researchers also use these tools to evaluate which experimental conclusions are replicable. Today, several media organizations partner with prediction markets, supplementing frontline reporting and traditional coverage with collective intelligence to enrich their content.
Prediction markets gather all participants' individual judgments about the future and integrate these views into a trading market to estimate the probability of various events. Users betting on outcomes on such markets operate on the same logic as those predicting a company's stock price in the stock market or trading oil prices in the commodity market. The difference is that the price of an asset like oil is influenced by multiple complex factors, while a prediction market asset yields a return only if a specified event occurs.
When oil prices rise, we can infer that current demand exceeds supply, but we may not know the underlying reason: is the market worried about escalating Middle East tensions, or has oil found a new application? Prediction markets can create separate trading instruments for single possibilities, enabling precise and granular forecasting. For example, a market for "whether the Strait of Hormuz will be open for normal traffic at a specified time" could have contract rules set as: if true, each contract pays out $1. As users continuously buy and sell, the market price becomes a probabilistic indicator, reflecting the collective judgment of all traders on the likelihood of the event.
The operating logic is as follows: If the current price per contract is $0.50, it means the market judges the probability at 50%. If you believe the probability of passage is higher, say 67%, you would buy the contract. If correct, your $0.50 purchase would yield a $0.67 payoff. This buying action would push the market price and estimated probability higher, indicating that a trader believes the market previously underestimated the event's likelihood. Conversely, if someone thinks the current price is too high, they would sell or short the contract, pulling down the market's probability estimate.
Compared to other forecasting methods, well-functioning prediction markets offer significant advantages. First, they directly output quantitative probability results, which is a core highlight. Polls and surveys can only count the proportion of opinions; translating that into event probabilities requires additional statistical methods to analyze the relationship between sample data and the overall population. Furthermore, poll results are mostly static data from a single point in time, while prediction markets update judgments in real-time as new participants enter or new information emerges.
More crucially, prediction markets have built-in incentive constraints. Both buyers and sellers invest real money and will suffer losses if their judgment is wrong. This forces participants to carefully assess the information they possess and prioritize trading in areas where they are more familiar and have greater informational advantages. Conversely, the desire to profit from information and expertise incentivizes people to actively conduct research and dig deeper into event-related clues. A well-known case is that, prior to the 2024 U.S. election, some prediction market participants conducted unconventional polling specifically to gather information traditional pollsters couldn't access.
Finally, the coverage scope of prediction markets is extremely broad. In theory, a trader with oil industry information can express their view by going long or short on oil contracts. However, in reality, there are numerous event outcomes that cannot be forecast through mainstream commodity or stock markets. These scenarios are precisely where prediction markets can play a role. For example, many prediction markets have recently launched contracts to comprehensively evaluate the performance of various AI models across different tasks. Trends in such niche areas are difficult to reflect in traditional commodity markets. Anyone can create and fund a prediction market to address such specialized questions.
Prediction markets are not a new phenomenon; their earliest forms can be traced back to 16th-century Europe, where they were used to predict the next Pope. Modern prediction markets combine knowledge from economics, statistics, market design, and computer science. In the 1980s, Charles Plott and Shyam Sunder first established a formal academic framework for this mechanism. Soon after, the world's first modern prediction market—the Iowa Electronic Markets—launched. Leveraging internet technology, this model has integrated fragmented information from around the globe and continued to develop.
However, several challenges remain to be solved to fully realize the potential of prediction markets. First, at the infrastructure level: how to determine the final outcome of an event and reach consensus, how to ensure transparent market operation and traceable transactions; and how to implement large-scale dispute resolution mechanisms when contract payouts are contested or even subject to manipulation.
Second, there are challenges in market design. First, people with key information must participate. If all participants are uninformed, the market price signal is worthless. Conversely, if various informed parties are unwilling to enter, the prediction will be biased. I argued as early as 2016 that prediction markets underestimated the probability of events like Brexit and Trump's first presidential election win because participants at the time failed to perceive the trend of rising populism.
Furthermore, the participation of individuals with insider information poses risks, especially if they have the ability to influence the event's outcome. Imagine if insiders in a papal conclave placed bets in a "next Pope" prediction market using privileged information for front-running; or worse, secretly intervened in the election to favor their own holdings. If participants widely believe the market is rife with insider trading, they will leave, ultimately causing the market to collapse.
Another risk is that someone might deliberately manipulate prediction market prices to influence public perception of event probabilities. In this case, the prediction market becomes a tool for manipulating public opinion rather than aggregating views. For example, a campaign could use its funds to artificially inflate its own market probability of winning, creating a false sense of lead. However, prediction markets have some self-correcting ability: as long as prices deviate significantly from a reasonable range, traders will place opposing bets to hedge against unreasonable pricing.
These issues indicate that prediction markets need further refinement of rules, clarifying standards for participant access, contract design, and overall operation. But if industry practitioners can solve these problems one by one, prediction markets will ultimately become an important tool for humanity to forecast the future and navigate uncertainty.
