The dispersion of trading capital across multiple independent order books when a single question is split into many binary contracts. In prediction markets, this creates ghost markets where tail outcomes receive little or no volume, reducing the information captured by the market structure.
Cluster: Liquidity & Trading
The dispersion of trading capital across multiple independent order books when a single question is split into many binary contracts. In prediction markets, this creates ghost markets where tail outcomes receive little or no volume, reducing the information captured by the market structure.
Referenced in 6 articles
Deep-dive analysis of Polymarket International's likely blockchain migration as it pursues CFTC compliance. Covers three technical migration paths — mirror-and-cutoff, dual chain instances, and sunset switchover — along with changes to KYC requirements and dispute resolution mechanisms. Argues dual chain instances are the most probable outcome, balancing compliance with continuity.
Binary prediction markets struggle with continuous outcomes like oil prices because capital fragments across individual strike prices. This piece uses Polymarket's crude oil dataset to illustrate the staircase hedging problem. It then shows how continuous probability markets solve this with a single density-based pricing mechanism.
Follow-up to functionSPACE's V1 discretisation analysis, splitting Polymarket's 18,863 multi-market events into continuous (price brackets, weather ranges, margin percentages) versus categorical (teams, candidates) and re-running the pathology tests. Both types concentrate 90% of volume in the top 5-6 markets, but ghost markets turn out to be largely a categorical phenomenon: continuous events distribute volume more evenly across buckets and survive the liquidity cliff longer at high N. With continuous events overtaking categorical by event count in 2026Q1, the case for a continuous-distribution primitive applies to a growing share of the platform.
Survey of recent research on why conventional market making fails in prediction markets. Covers cross-venue fragmentation (the same contract at 58-67 cents on different platforms), the January 2026 Polymarket XRP exploit that paid $231K on thin weekend liquidity, Kalshi's structural longshot bias, and evidence from 150M Polymarket trades that the top 5% skilled traders earned $228M while spread capture barely moves P&L. Concludes passive LPs on these venues behave more like underwriters of terminal risk than classical market makers.
Announces Lotus, an execution layer for prediction markets that aggregates fragmented liquidity across venues without owning any itself. The same events trade in isolated silos with different formats, resolution rules, and depth, producing poor price discovery and high slippage — the same problem DeFi DEXs faced before aggregators. Lotus combines orderbooks and odds into a unified market view, canonicalizes truly equivalent markets via a verification process called Bloom, and smart-routes orders across venues and market formats to minimize slippage, accounting for differences in oracles, resolution methods, and counterparty risk. Users submit one order, pay one fee, and receive a single aggregated position — positioned as the 1inch prediction markets have needed, with the post citing a $35M infrastructure fund backing the thesis.
Analyzes 36,777 Polymarket events to understand what happens when continuous questions are split into dozens of independent binary contracts. Volume follows an extreme Pareto distribution: the top 3 markets capture over 75% of trading activity regardless of event size, leaving a large fraction as untradeable ghost markets. The $0.01 tick size compounds the problem, creating a rounding tax that makes low-probability contracts structurally imprecise.