Predicting Market Liquidity Provision: Looks Easy but Actually Full of Minefields

An explanatory article about Polymarket’s shared order book mechanism has become a hot topic on X, leading to widespread attention on the formula Yes + No = 1. Many DeFi projects like Buzzing have also expressed the view that this formula is an innovative mechanism second only to x * y = k, bringing infinite possibilities to information dissemination markets.

Indeed, this design lowers the barrier for “anyone to become a market maker and earn fees.” However, there is a significant pitfall hidden within this optimistic outlook.

Market Making in Prediction Markets Is Far More Difficult Than Imagined

To understand the gap between ideal and reality, it is necessary to recognize the fundamental differences between the AMM model and the order book model.

In AMM markets like Uniswap V2, liquidity providers simply deposit ETH and USDC into pools at appropriate ratios. Impermanent loss occurs due to price fluctuations, but continuous trading fees compensate for it. Once positioned, the market mechanism functions automatically.

However, prediction markets are entirely different. Suppose you are market making on Polymarket and place a buy order at $0.56 and a sell order at $0.60 when the YES price is $0.58. Then, four scenarios could occur:

Scenario 1: Neither order executes → No spread profit, but liquidity subsidies are earned
Scenario 2: Both orders execute → Spread profit earned, no subsidies
Scenario 3: One order executes, and the market price remains within the order range → Directional position arises (inventory risk)
Scenario 4: One order executes, and the market price suddenly moves in the opposite direction → Loss state with no subsidies

The key point is that low-frequency operations may yield different results, but in actual operation, continuous losses are likely.

Why Do AMM Market Makers Succeed While Prediction Markets Fail?

The reason lies in the fundamentally different operation mechanisms:

Operation Mechanism: AMMs automatically distribute liquidity based on formulas, whereas order book models require manual orders at specific points
Continuity: AMMs remain effective within price ranges; order book models require active order management and price adjustments
Risk Structure: AMMs mainly face impermanent loss; order book models have near-infinite inventory risk

In Polymarket, profits are limited to “spreads” and “liquidity subsidies,” but the risks of directional positions far exceed these. In other words, for market makers to sustain profits, they need advanced strategies to minimize risk positions and maximize revenue opportunities.

Tracking external events, understanding market structure, grasping settlement rules, and real-time order adjustments—these are skills that even experienced market makers on CEXs or Perp DEXs cannot rely on in prediction markets.

Prediction Markets Are More “Wild” and More “Dangerous”

In traditional crypto markets, prices fluctuate cyclically over the long term. But prediction markets are inherently different:

Event-Driven Nature: Each contract has a clear settlement time, and ultimately, the outcome is either YES or NO, both converging to 1 dollar. In other words, the market will inevitably become one-sided in the end.

Discontinuous Price Movements: Unlike normal markets where emotions and capital move prices continuously, prediction markets see real-world events cause leap-like price jumps. If the previous price was 0.5, the next could suddenly jump to 0.1 or 0.9, leaving market makers with extremely limited reaction time.

Serious Information Asymmetry: Prediction markets have “insider players” who receive real-time information. They trade with definitive knowledge, providing liquidity that market makers supply. Thus, market makers are always at an informational disadvantage.

While gathering information through DeFi projects like Buzzing is important, in front of players with genuine inside information, ordinary market makers are close to “prey.”

Intensified Competition and Subsidy Wars Between Platforms

Faced with these challenges, Polymarket and Kalshi are competing with substantial liquidity subsidies:

• Polymarket: Invested about $10 million in liquidity subsidies, paying over $50,000 daily at peak, currently offering $0.025 subsidy per $1 traded
• Kalshi: Invested over $9 million in subsidies, and in 2024, signed a market-making agreement with Susquehanna International Group (SIG)

These leading platforms have capital advantages. Polymarket is valued at $8 billion, with a $2 billion investment, aiming for over $10 billion in the next round. Kalshi is valued at $5 billion, having raised $300 million.

Is the “Blooming” of Prediction Markets an Illusion?

Currently, prediction markets are a hot spot for startups, with new projects emerging one after another. However, Haseeb Qureshi, an analyst at Dragonfly, warns:

“Prediction markets are rapidly developing, but 90% of prediction market products are completely ignored and will gradually fade away by the end of the year.”

This judgment is realistic. As leading companies continue to provide subsidies and establish partnerships with regulators, it is extremely difficult for new projects with less capital to compete.

Some new projects may survive with backing from parent companies, but not all. If you truly want to bet on prediction markets, focusing on leading players is more practical.

In summary, market making in prediction markets is not a “easy side hustle” to make quick money, but a domain where only specialized players with deep understanding of market structure and excellent risk management can achieve sustainable long-term profits.