Prediction markets have a very interesting history. Not sure if this is the first that they were ever written about but Robin Hanson is generally seen as the father of prediction markets. He first wrote about them in Idea Futures in which he describes how policy makers should turn to betting markets rather than experts for decision making (later this became Futarchy).

Here is a list of many more Hanson publications.

A few years after Hanson’s initial Idea Futures post multiple big corporations picked up the idea. Companies like Microsoft, Eli Lilly, and Hewlett-Packard have run experiments in which employees trade around new drug candidates, approval of drugs, etc.

Google ran internal prediction markets to “forecast product launch dates, new office openings, and many other things of strategic importance to Google”. After 2 years of continually running their internal markets Using Prediction Markets to Track Information Flows was released, detailing many biases and outcomes. They have their application public on GitHub.

Even crazier in 2005 the Pentagon set up a prediction market website in which terrorist attacks and assassinations.

The Pentagon is setting up a stock-market style system in which investors would bet on terror attacks, assassinations and other events in the Middle East. Defense officials hope to gain intelligence and useful predictions while investors who guessed right would win profits.

Prediction markets have gone through many iterations been tried in a number of applications from private corporations, open on blockchains, to the government.

Although, outside of U.S. election cycles and certain internal company markets they have never really caught much mainstream interest. Liquidity, Regulation, Duration of the predictions all play into this.

Sports betting on the other hand has always been popular and with recent regulation changes seen tremendous growth. Sports betting benefits from short duration bets and “always on” liquidity provided by the House.

Sports betting and prediction markets are one in the same, they are both Event Markets.

Forecasting Events

The idea of forecasting events has been very popular with the EA (effective altruism) community for a while and now more so with the e/acc (effective accelerationism) community. Basically the slate star codex / astral codex ten, scott alexander, LessWrong type. Ironically perfectly described in scott alexander’s Bay Area House Party Substack Series (Every Bay Area Party, Another Bay Area House Party, Even More Bay Area House Party)

While different than EA, e/acc seemingly only became popular and mainstream after SBF ruined the EA name in 2022. Notably SBF offered conditional markets on FTX and FTX’s Future Fund made donations to the popular prediction market manifold.markets.

The forecasting community is niche but very popular. In fact prediction market site manifold.markets recently held their first conference Manifest. Here is some coverage on it in the New York Times.

Superforecasting by Philip E. Tetlock and Dan Gardner details the government funded Good Judgment Project. The yearly tournament poses between 100 and 150 questions on geo-politics*.* You can join challenges at gjopen.com/challenges, some of them have close to 100k forecasters participating.

Far easier to follow is the Annual Forecasting Contest held by Astral Codex Ten. In this contest a variety of 50 questions were asked. Rather than yes/no the questions asked for a probability that an event happened. Here are the results from the 2023 competition. I look forward to these competitions each year.

Forecasting for the most part is pure reputation based and forecasters take pride in getting probabilities exactly right year after year.