Polymarket and On-Chain Prediction Markets: What Information Aggregation Actually Looks Like

# Polymarket and On-Chain Prediction Markets: What Information Aggregation Actually Looks Like

Prediction markets have been an academic obsession since at least the 1990s, when economists demonstrated that market prices aggregate dispersed information more accurately than expert panels or polls. Robin Hanson's work on decision markets influenced a generation of researchers. The internet created Intrade, which demonstrated real-world viability and was then killed by regulatory action. For a long time, the idea sat in academic limbo.

Polymarket, built on Polygon, has become the first prediction market to achieve meaningful scale without being shut down.

## What Polymarket actually is

Polymarket's approach: decentralized settlement, resolution by community consensus and named data oracles (UMA protocol), accessible through a simple web interface. It's attracted hundreds of millions in monthly volume at peak, with the 2024 US election cycle driving its most significant growth to date.

The 2024 US election was the clearest test case for the information aggregation thesis. Polymarket's odds on Trump winning the presidency were consistently higher than traditional polling models in the weeks before the election. Whether this represented genuine information advantage — sophisticated traders with better models — or a skewed user base (crypto-adjacent, internationally heavy, not representative of American voters) is still debated. The market was correct on the outcome. It was also more volatile than traditional polling models in ways that suggest speculative noise alongside signal.

## What prediction markets demonstrably do well

Highly specific, verifiable events with clear resolution criteria. "Will Ukraine reclaim Crimea by December 31, 2025?" is a better prediction market question than "Will Ukraine win the war?" because the first has a binary outcome and an unambiguous resolution date. Polymarket's most liquid markets are structured this way. When the question is clear and the resolution is unambiguous, market prices track calibrated probability estimates reasonably well.

## What prediction markets do less well than advocates claim

Long-horizon complex questions, questions where resolution can be contested, and questions where a small number of large traders can move the market without having superior information. Thin markets are particularly susceptible: a well-capitalized actor taking a large position will shift prices even if their information content is zero.

The on-chain dimension matters for a specific practical reason: Polymarket can't be shut down by a single regulatory action against a central operator. Traditional prediction markets died when regulators pursued Intrade's leadership. Decentralization isn't just a philosophical position here — it's the product's durability feature. (Polymarket did block US users following a CFTC settlement in 2022, but continues operating globally.)

## The broader implication

The more interesting question for 2025–2026 is whether prediction market infrastructure can be embedded in governance, forecasting platforms, and corporate decision-making. Several projects are exploring this. The mechanism design challenges are real — sufficient liquidity, clear resolution criteria, user sophistication — but not insurmountable.

Polymarket's growth proved there's demand. The information aggregation thesis is real under the right conditions. Whether on-chain prediction markets consistently meet those conditions depends heavily on the specific market, the liquidity behind it, and how resolution disputes are handled when the question turns out to be less clean than it appeared.

This raises an important question: prediction markets are only as good as their resolution process. In most cases, that process involves humans making judgment calls. The mechanism design problem hasn't been fully solved.

> **Key Takeaway:** Polymarket demonstrated that a decentralized prediction market can survive and scale. Whether its prices represent the best available estimate depends on the question, the liquidity, and the resolution mechanism — not on decentralization alone.

Polymarket and On-Chain Prediction Markets: What Information Aggregation Actually Looks Like

// COMMENTS

ON THIS PAGE