A Short History of Prediction Markets

Prediction markets feel modern. They are not. Organised betting on political outcomes was a large, public and thoroughly reported business in the United States for the better part of a century, and understanding that history explains a lot about why the current venues are shaped the way they are.

Wall Street's election markets

From roughly the 1880s until the Second World War, formal election betting markets operated in the New York financial district, with prices carried in the newspapers alongside stock quotes. Volumes were substantial in real terms, participation was broad, and the prices were treated as news.

Researchers who later reconstructed those markets found something uncomfortable for modern polling: the prices were a good forecast. They were not perfect, and they had biases, but as a running estimate of election outcomes they held up well against what came later.

The markets faded for mundane reasons — the rise of scientific polling gave newspapers a cheaper story, legal pressure increased, and the money moved elsewhere.

The academic revival

The idea returned through universities. From the late 1980s, small-scale real-money markets run for research purposes demonstrated the same thing repeatedly: prices from a market of modestly informed participants tracked outcomes about as well as, and often better than, contemporaneous polling, particularly in the final weeks.

This body of work matters because it established two things. First, the aggregation effect is real and reproducible, not a story told after the fact. Second, it works at small scale, which was surprising: you do not need enormous liquidity to get a usable signal, only enough to make being wrong cost something.

The same period produced the theoretical framing — markets as information aggregation mechanisms, prices as a consensus estimate weighted by both confidence and capital — that the field still uses.

The internet venues

Through the 2000s, offshore internet exchanges ran real-money markets on politics, economics and current events for an international audience. They proved the model worked online and at scale, produced a great deal of the price history researchers now use, and ran into the problem that has shaped everything since: these instruments look like derivatives, and derivatives are regulated.

Meanwhile, corporate internal markets had a brief vogue — companies running play-money or small-stakes markets on their own project deadlines and sales figures. The results were interesting, the politics were difficult, and most programmes quietly ended. The lesson was not that the mechanism failed but that organisations do not always want an unfalsifiable public estimate of whether a project will ship on time.

The regulated era

The current generation of US venues took the opposite approach: go through the regulatory process rather than around it. That means registering as an exchange, listing contracts under a supervised process, holding customer funds under rules, and accepting constraints on what can be listed.

It is slower and narrower than the offshore model, and it has a large compensating advantage: the venue is a market operator rather than your counterparty, customer money is segregated, and the whole thing is durable. What Kalshi is, and what being regulated changes covers what that means day to day.

The on-chain era

In parallel, event markets moved on-chain. Contracts became tokens, collateral became a stablecoin, settlement became a smart contract, and resolution became an oracle process with a dispute mechanism.

The tradeoffs are almost the mirror image of the regulated route. Access is broad, listing is fast, the long tail of questions is enormous, and the whole book is publicly inspectable in a way a traditional venue's never is. Against that: resolution depends on a social process rather than a contractual source, and the rules around access differ by jurisdiction and change. What Polymarket is covers the model.

Where that leaves things

For the first time there are two liquid venues with genuinely different architectures pricing overlapping sets of events. That is what makes cross-venue comparison interesting rather than academic: when a regulated exchange and an on-chain venue disagree about the same question, the disagreement has content. It might be a difference in participants, in resolution terms, in fees, in who can access what, or simply in how fast each one repriced.

Measuring those disagreements continuously is what the gap board on this site does, and why two venues price the same event differently is the guide to reading it.

What the history suggests

Three things, none of them new.

Markets aggregate well when participants have something at stake and the question is clear. Both conditions are load-bearing.

Thin markets are opinions with a spread. The historical Wall Street markets were large. Modern long-tail markets often are not, and the research on large markets does not automatically transfer.

Regulation shapes the product more than technology does. The differences between today's venues are mostly downstream of which regulatory path each one took.

Next

Your first week watching prediction markets turns all of this into a routine.