A16z Crypto latest article: Why do we need to forecast the market?

Null

Author: Scott Kominers

Translation: Jiahui, ChainCatcher

Prediction markets allow people to trade on the outcomes of events. Last year, they entered the U.S. on a large scale, and now they are used to track various events from geopolitical issues to entertainment award results. But what exactly are they?

As an economist who has long studied markets and incentive mechanisms, my answer is simple: prediction markets are essentially markets. Markets are fundamental tools for resource allocation, ensuring goods and services flow to those who value them most.

In this process, markets also aggregate information: market clearing (i.e., reaching supply and demand equilibrium) is essentially a mechanism that consolidates all participants’ knowledge and distills it into price signals.

Prediction market platforms and products directly leverage this information aggregation ability to forecast specific future events: they design assets linked to events, which generate profits once a particular outcome occurs, and people trade these assets based on their judgments about whether the outcome will happen.

Such usage has existed for a long time.

Businesses have long used prediction markets to gather implicit information from employees, such as predicting whether a key product will be released on time; scientists use them to assess which experiments are likely to be successfully replicated; today, several media outlets collaborate with prediction markets to supplement their information sources and reporting with "crowd wisdom."

Prediction markets collect information directly from participants—each person’s judgment about the future—and aggregate this information into a market, thereby answering how likely a certain event is to happen.

People can "bet" on the future value of a company in the stock market, or on the future price of commodities like oil, and so on. The difference is that demand for assets like oil is influenced by many factors, whereas assets designed in prediction markets only pay out when a specific event occurs.

If oil prices rise, we know demand has increased relative to supply, but we may not know why: perhaps people expect an escalation in Middle East conflicts, or maybe someone has found a new use for oil.

With prediction markets, you can isolate and forecast each possibility separately.

For example, a prediction market on whether the Strait of Hormuz will remain open at a certain time could involve a contract: once the event occurs, each contract pays one dollar.

As people repeatedly buy and sell this asset, the market price becomes a "probability indicator," reflecting traders’ overall assessment of the likelihood of the event.

How does it work specifically? Suppose the market price for a certain outcome is $0.50, which corresponds to a 50% probability. If you believe the chance of the strait remaining open is higher than 50%, say 67%, you would buy; if your prediction is correct, you earn a total of $0.67 on a $0.50 investment.

This buying activity pushes the market price and the estimated probability higher, effectively saying "someone thinks the market is undervaluing this." Conversely, if someone believes the price is too high, they will sell at a lower price (or short), pulling the overall probability estimate downward.

When prediction markets operate well, they have several clear advantages over other forecasting methods.

First, they can directly provide a probability estimate—this alone is a "superpower."

Public opinion polls and surveys only give "opinions," and converting these into probabilities requires statistical inference to determine the relationship between the observed proportion and the overall likelihood. Moreover, polls are often just snapshots at a specific time, while prediction markets can update in real-time as new participants and information enter.

More importantly, prediction markets have built-in incentives: buyers and sellers put real money on the line, and if they are wrong, they lose. This encourages participants to carefully consider their information and invest in the issues they are most confident about.

Conversely, the ability to profit from information and expert judgment in prediction markets motivates people to conduct research and clarify issues.

(A well-known example is that before the 2024 U.S. presidential election, a prediction market participant even conducted a poll using unconventional methods to extract information unavailable to standard polling agencies.)

Finally, prediction markets also have a significant coverage advantage. Someone knowledgeable about events that could influence oil demand can go long or short on oil; but many of the outcomes we want to predict do not have corresponding commodity or stock markets for betting. In such cases, prediction markets become an ideal choice.

For example, recently, prediction markets have emerged specifically to assess "which AI model performs best on various tasks." These questions are too granular for traditional commodity markets to reflect, and anyone can create and fund a prediction market for such specialized issues.

These ideas are not new. As early as the 16th century in Europe, similar practices existed for predicting the next papal election.

The foundation of modern prediction markets lies in economics, statistics, market design, and computer science. Charles Plott and Shyam Sunder proposed the earliest formal academic frameworks in the 1980s, shortly before the first modern prediction market, the Iowa Electronic Markets, was established.

With the help of the internet, this model now can aggregate dispersed information from around the world. However, for prediction markets to truly realize their potential, several prerequisites remain.

One is infrastructure: how to verify and reach consensus on whether "a certain event has already occurred," how to ensure market transparency and auditability, and how to handle large-scale settlement of contracts that may be controversial or even subject to manipulation.

Another is market design challenges. First, those who truly hold relevant information must be willing to participate. If participants are uninformed, price signals are meaningless; conversely, only when those with various information participate can the prediction be accurate.

I pointed out in 2016 that prediction markets might underestimate the probabilities of Brexit and Donald Trump’s first election victory because participants at the time did not fully understand the rise of populism.

Another issue is if someone possesses "perfect" information, such as knowing the true outcome in advance—especially problematic if they can influence the event’s course.

Imagine: if an insider at a secret papal conclave bets on the next pope, trading before the official announcement of the election results, or even tries to sway the election to ensure their preferred candidate wins, what would happen?

Because of this, once potential participants anticipate the presence of insiders trading, rational choices would be to stay away, and the market would collapse.

Finally, some might deliberately manipulate prediction market prices to influence public perceptions of the likelihood of certain outcomes, turning the market from a "belief aggregator" into a "belief manipulator."

For example, a candidate’s PR team might spend part of their campaign funds to sway relevant markets to create the impression of confidence.

However, prediction markets have a certain self-correcting ability: if a contract’s probability is pushed to an unreasonable level, someone will be willing to take the opposite side of the trade.

All of this indicates that prediction markets need higher transparency and clarity in participation management, contract design, and operation. But if designers can solve these challenges, prediction markets could become one of our core tools for forecasting the future.