a16z Cripto, latest post: Why do we need to predict markets

Author:Scott Kominers

Other Organiser

The forecast market allows people to trade in the results of events. Last year, it entered the United States on a massive scale and is now being used to track events ranging from geopolitical to recreational awards. But what is it exactly

As an economist who has long studied markets and incentives, my answer is simple: predicting markets is essentially markets. Markets are the basic instrument for allocating resources to ensure that goods and services flow to those who value them most。

In the process, the market also aggregates information:Market opening (i.e. achieving a balance between supply and demand)In essence, it is a mechanism to aggregate the perceptions of all participants and to refine them into price signals。

Predicting market platforms and products is the direct use of this information aggregation capability to predict specific future events: They design an event-related asset that generates benefits once a particular result occurs and is then traded on the basis of their own judgement as to whether the outcome will occur。

It's already been used。

EnterpriseThe forecast market was used early to obtain hidden information from employees, such as predicting whether a significant product could be released on timeScientistsUse it to assess which experiments are expected to be successfully re-emerged; there are now many othersMediaWorking with the predictive market, using "mass wisdom" to supplement home-based information sources and journalists' coverage。

It is predicted that the market will gather information directly from the participants, that is, everyone's judgement about the future, and then bring it together in a market, in order to answer how much it is possible that something might happen。

One can bet on these events, as it is in the stock market to bet on the future value of a company, or on the future price of large commodities such as oil. In contrast, the demand for assets such as oil is influenced by a number of factors at the same time, while predicting market-designed assets generate gains only when a given event occurs。

If oil prices rise, we know that demand has risen relative to supply, but we do not necessarily know why: it may be that the Middle East conflict is expected to escalate or that someone has found new uses for oil。

And by predicting the market, you can single out every possibility。

For example, a forecast market for "Will the Strait of Hormuz remain open at a certain point in time" could be built around a contract that, once an incident has occurred, pays a dollar per unit contract。

As people buy and sell such assets repeatedly, market prices become a "probability mark", reflecting the overall perception of the likelihood of the event。

How does it work? Assuming the unit market price for a given result is 0.50, which is equivalent to 50% probability. If you think it's 50% more likely that the strait will remain open, like 67%, you'll buy it; if you're right, you'll get $0.50 at a cost of $0.67。

And this buy-in pushes up market prices and corresponding probabilistic estimates, which is tantamount to saying, "Some people think the market underestimated it." The same is true of the reverse: when people are perceived to be overpriced, they are sold (or empty) at lower prices, thus lowering the overall probability of the market。

When the market is forecast to work well, it has a few obvious features compared to other methods of forecastingAdvantagesI don't know。

First of all, it gives a probabilistic estimate, and that alone is a "superpower."。

The poll and the questionnaire give a "opinion ratio" that you have to do statistically to figure out the relationship between this ratio and the total. Moreover, polls are often only a snapshot of a certain point in time, and predict markets that can be updated in real time with new participants and new information。

More crucially, the forecast market incentives are self-incentives: both buyers and sellers take in real money and silver and pay for the wrongs. This prompts participants to take careful account of the information in their hands and to invest the money in their own best interests。

In turn, the ability to benefit from information and professional judgement in predicting markets also inspires people to take the initiative to study and clarify issues。

A well-known oneExamplesYes: Before the presidential election in 2024, a market participant predicted that he had even made a poll himself, using an unconventional method to dig for information that could not be obtained by a standard polling agency

Finally, the projected market also has significant advantages in terms of coverage. A person who knows which events may affect oil demand can, in principle, do more or empty oil; but many of the results that we want to predict do not have the corresponding large commodity or stock markets to bet on. At that point, forecasting the market became the ideal option。

IN RECENT TIMES, FOR EXAMPLE, THERE HAVE BEEN A SERIES OF FORECAST MARKETS THAT HAVE BEEN USED TO BRING TOGETHER JUDGMENTS SUCH AS WHICH AI MODEL IS BEST PERFORMING ON MISSIONS, WHICH ARE TOO DETAILED TO BE REFLECTED IN TRADITIONAL COMMODITY MARKETS. ANYONE CAN CREATE AND FUND A FORECAST MARKET FOR SUCH BREAKDOWNS。

These ideas are not new. As early as the 16th century in Europe, a similar practice existed, when people used it to predict the next Pope。

Modern forecasting markets are rooted in economics, statistics, market design and computer science. Charles Plott and Shyam Sunder introduced the first formal academic framework in the 1980s, and shortly thereafter the first modern forecast market, the Iowa Electronic Markets, was born。

Through the Internet, this model is now able to bring together scattered information across the globe. There are also many prerequisites for the market to truly realize its potential。

The first is..Infrastructure issues• How to verify and reach consensus on whether an "incident has occurred", how to ensure transparent, auditable market operations and how to deal on a large scale with contractual settlements that may be controversial or even manipulated。

Another category is market design challenges. First, those who really have the relevant information must be willing to come. If the participants are not informed, the price signals do not mean anything; in turn, the estimates of the market are not distorted only if people with all types of information are involved。

I was in 2016NoteIn the past, the forecast market may have underestimated the probability of first-time election to the British Deutsche and Trump, as participants at the time did not know enough about the rise of populism。

Another problem is that if someone has "perfect" information, such as knowing the real results in advance, it's equally troublesome, especially when he can influence the course of events。

Imagine: what if an insider of the pope's secret election conference went to the prognosis market of the next pope, bought a deal before the news of Leo's election was officially announced, and even tried to influence the election so that he could win

As a result, once potential participants anticipate an insider's presence, the rational choice is simply to stay away and the market collapses。

Finally, there are those who may deliberately distort the prices of predicting markets to influence public perceptions of the probability of an outcome, moving it from a tool of "convergence" to a tool of "manipulation"。

For example, a public relations team of a candidate who wants to convince the outside world that he or she is a steady winner may spend a portion of his or her campaign funds on the market。

On this point, however, there is some self-correcting ability to predict markets: once the probability of a contract is pushed to an unreasonable level, there will always be a willingness to stand on the other side of the deal。

All this points to the need for greater transparency and clarity at the level of participation in management, contract design and operations. But as long as designers solve these problems, predicting markets is expected to be one of our core tools for predicting the future。