<100 subscribers

The Past, Present, and The Future of Prediction Markets

Prediction markets can be created around almost anything with a clear outcome and timeline. In crypto, they might ask whether BTC will cross $70K by a certain date or if a token’s market cap will hit $1B. In tech, markets can form around product launches, IPOs, or major releases. Culture and internet trends show up as bets on whether a creator will hit a milestone in number of followers or if a movie will cross a revenue mark. Sports markets focus on match outcomes, player performance, or tournament winners.

Skeptics dismiss prediction markets as gambling or speculation. The way I see it is that some forms of speculation are more useful than others. Prediction markets create incentives for people to put money behind their views, which leads to more honest signals about how the future might play out. In a world where most information on social media is noise, marketing, or disguised opinion, prediction markets offer a clearer, more grounded sense of what people actually believe will happen.

One of the earliest recorded instances of prediction markets on political events was around papal elections in 1503. Even then, it was described as an "old practice".

The 1503 Papal election and its associated prediction markets bear a striking resemblance to the dynamics of the 2024 U.S. Presidential election in the sense that how the markets were able uncover an outcome that didn't seem likely in first glance. The market was facilitated by Roman Banking Houses who acted as bookmakers offering odds. Bankers often formed partnerships with cardinals' attendants who were inside the conclave. These attendants would leak private information regarding voting tallies to the merchants and bankers

The main contenders in the election were Cardinal Francesco Piccolomini, Cardinal Giuliano della Rovere, and Cardinal Georges d’Amboise. The market had Cardinal Piccolomini as the favorite with 2X better odds.

As the voting started, Cardinal Piccolomini was trailing with 4 votes to 13 for d’Amboise and 15 for della Rovere. However, in a surprising turn of events, as the election ran into a standstill, Cardinal d’Amboise threw his support behind Cardinal Piccolomini who became Pope Pius III proving the bookmakers right.

By the 19th century, prediction markets had spread across Europe and the United States. People were actively betting on parliamentary outcomes, royal succession, and even wars.

But a clear pattern started to emerge. Whenever these markets grew, they faced pushback. The Catholic Church tried to curb betting on papal elections, and governments across countries began cracking down on them as well.

Prediction markets saw a revival in the late 20th century, mostly in controlled environments within academia, corporate, and defence research. Researchers at University of Iowa founded the Iowa Electronic Markets back in 1988. IEM allowed people to buy contracts on events ranging from political events such as US Presidential Elections to Fed Funds Rates. In the 2008 US election IEM predicted the final vote count to within half a percentage point. IEM never attempted to scale beyond its research purpose. It operates under a special understanding with CFTC where the maximum amount a user could participate with in a market was capped at $500.

In July 2003, DARPA, the US Defence Think Tank, proposed the Policy Analysis Market, an initiative designed to trade in geopolitical risk to determine if such markets could help predict future events like civil stability, economic health, and military disposition. The project intended to offer contracts based on specific indicators, such as the growth of a country’s non-oil output or potential military withdrawals. However, a major political controversy ensued, with critics labeling the program "terrorism futures". Due to the intense public and political backlash, the proposal was dropped. Ironically, the predictive power of these markets was demonstrated when an offshore betting exchange accurately forecasted the resignation of DARPA’s head.

Technology Companies such as HP, Google, and Siemens experimented with prediction markets internally to forecast sales, project completion, and OKR accomplishment. Employees could trade on questions like product launch timelines, sales targets, or project completion dates. Because participants often had direct knowledge of what was happening on the ground, these markets were able to surface more accurate and real-time signals than traditional reporting or top-down forecasts. In many cases, they outperformed internal estimates and expert opinions. However, these markets were limited to employees within the company, and while useful for decision-making, they never extended beyond internal use.

Intrade arguably was the first modern attempt at Prediction Markets that gained broad adoption. Founded by by Ron Bernstein and Sean McNamara in 1999, it allowed users to trade on a myriad of events ranging from elections to weather. It was incorporated in Ireland and was able to bypass regulatory hurdles. Similar to how Polymarket and Kalshi gained traction during the 2024 US elections, Intrade saw its breakout moment during the 2008 US elections. In 2008, John Delaney, wrote a letter to CFTC seeking clarification on their regulatory stance

While Intrade serves a global community and has registered members from 162 countries, our 82,000 plus membership are predominantly resident in the United States ... it is perversely unclear as to whether Intrade, and indeed myself, are considered persona gratis by the United States.

In 2012 US Elections, Intrade reached a peak of million of monthly visitors. By 2013, combined with legal pressure and internal issues, the platform shut down, cutting short what was the first real attempt at scaling prediction markets on the internet.

The Ethereum Whitepaper was published in 2013. One the applications that Vitalik proposed in it was Prediction Markets.

Prediction markets. Provided an oracle or SchellingCoin, prediction markets are also easy to implement, and prediction markets together with SchellingCoin may prove to be the first mainstream application of futarchy as a governance protocol for decentralized organizations.

Augur was conceived in 2014 and ICOed in 2015 raising $5 Million and distributing its native Reputation (REP) tokens. After a beta in 2016, the protocol went live on Ethereum Mainnet in 2018. Augur allowed anyone to create markets on any question, with outcomes determined by crowdsourced reporting from REP holders. As a pioneer in the space, Augur faced almost every challenge that a decentralized prediction market could.

The first version of Augur was a full fledged Ethereum Node Client that users had to install on their computers. The app itself would take hours to launch. Trading on Augur v1 involved an order-book model and required multiple transactions (e.g. buying complete sets of outcome shares), incurring high Ethereum gas fees. The experience was clunky.

That wasn't the only challenge. With in weeks of launch, multiple "assassination markets" had cropped and the decentralized nature of Augur meant that the community was divided on how to address them.

A flaw in how markets were invalidated allowed market creators to exploit market participants. Scammers create markets with unclear questions. They placed small bets on unlikely outcomes to make the market look real. Then they sold shares in the more likely outcome to other users.Once enough money came in, people would notice and vote the market as invalid. The money in the market however was split evenly.

These hurdles meant that by 2019, activity on Augur had stagnated, and many markets had little to no liquidity.

Augur V2 went live just in time for the US 2020 Elections. For a brief period, it felt that it could all work out. However, after the end of the elections, usage cratered. Augur attempted a new release called Turbo, this time on Polygon, to take advantage of lower gas fees. However, the tides had turned by then.

Polymarket and Kalshi are the two giants of prediction markets today. What’s interesting is how different their paths have been.

Kalshi was founded in 2018 and chose the regulated route from day one. It spent years working with the CFTC to get approval before even launching. This meant slow progress, but strong legitimacy. At one point, Kalshi sued the CFTC to list markets on US presidential elections, arguing that these are valid financial events people should be able to trade on.

Polymarket, founded in 2020, took the opposite approach. It went crypto-native and launched on Polygon, allowing users globally to trade on real-world events using stablecoins. This gave it speed and flexibility. Markets could be launched quickly and participation wasn’t restricted to the US. But this path came with its own challenges. In 2022, Polymarket was fined by the CFTC and had to block US users.

Both approaches worked, in different ways. Today, both are doing billions in monthly volume. Kalshi has seen strong traction in sports markets, which make up a large part of its volume, while Polymarket is more balanced across politics, crypto, and global events.

Like in previous cycles, the 2024 US Presidential Elections marked a turning point. But it wasn’t just about the surge in trading volumes. September, 2023 - Polymarket launched a market speculating President Joe Biden dropping out of the race. The initial months saw the odds of that happening hover around 22.5%.

With time, the odds started climbing. In July, 2021 despited Biden explicitly saying that he would be in the race, the odds stayed at 66%.

2 weeks later, Biden announced dropping out of the race. The theme repeated again as the conventional polls suggested that elections were going to be close with Kamala Harris in lead where as Prediction Markets such as Polymarket had Trump in lead. When elections finally concluded and Trump was elected as the next president, the world took notice. The crowds saw it coming before the experts did.

The success of Polymarket and Kalshi has triggered a surge of new prediction market startups and tools. From the outside, it may look like the category is already figured out and heading toward maturity.

Prediction markets are already on a run rate of over $250B in annual volume in 2026. But most of this activity is concentrated in a small set of markets. On Polymarket, for example, there may be ~12K open markets at any time, but only about 1K of them drive more than half the volume.

The user base is still small, with around 1–2 million transacting users.

This is not a mature market. It’s just getting started.

There are three broad themes emerging in prediction markets today.

First is scaling supply. Prediction markets are an infinite canvas. But even with around 12K open markets today, it barely scratches the surface of all the questions humanity has about the future. New products are focused on increasing the number of markets. Some are going vertical, like Noise exploring social trend markets and Limitless focusing on short-duration price markets. Others are pushing toward permissionless creation, like Context Markets where anyone can create a market. At the infrastructure level, teams like Melee Markets are working on new AMMs to better support long-tail markets. The idea is simple: more markets means better coverage of reality.

Second is scaling demand. It’s not just about creating markets, but getting more people to participate. Products are rethinking the experience. Pred.app is making prediction markets feel like trading apps, especially for sports. Kash.Bot lets users participate directly from social feeds. Instead of asking users to come to a trading app, these products are bringing markets to where users already are. The focus is on making participation easier and more natural. Gondor.fi is building DeFi primitives like lending and borrowing on Prediction Market positions to improve capital efficiency.

Third is aggregating and improving trading. As markets spread across platforms, tools are emerging to bring everything together. These tools help users discover markets across platforms, compare prices, and trade more efficiently. This is similar to what happened in crypto, where aggregation became important once liquidity fragmented. The goal here is to make trading simpler, faster, and more efficient across the ecosystem. Fireplace and Tradefox are two players working towards this.

One idea I keep coming back to is this: prediction markets as an oracle for the internet.

Today, most of what we see on social media is noise and propaganda. Opinions, narratives, and incentives to say what sounds good and gets you the clicks, not what is true. Prediction markets flip that. They force people to put skin in the game and express what they actually believe will happen.

Instead of scrolling through endless takes, you get a probability and instead of narratives, you get a price.