ForesightFlow
← Publications

ForesightFlow: Real-Time Detection of Informed Trading in Decentralized Prediction Markets

Maksym Nechepurenko · 2026 · Working Paper

Abstract

We introduce the Information Leakage Score (ILS), a market-microstructure framework for detecting informed trading in prediction markets in real time. We adapt classical PIN, VPIN, and Kyle's λ to discrete-outcome on-chain CLOB venues, and combine them with on-chain wallet features (novelty, funding concentration, cross-market behavior). The framework is validated on the ForesightFlow Insider Cases inventory of eight documented episodes spanning 24 markets.

This paper introduces the Information Leakage Score (ILS) framework for real-time detection of informed trading in decentralized prediction markets, with a focus on Polymarket.

Motivation

Prediction markets have emerged as a third venue for price discovery, yet they lack the regulatory infrastructure and academic research depth of equity and derivatives markets. This creates both risk — informed traders can profit systematically from non-public information — and opportunity for detection using tools developed for classical microstructure.

Framework

We adapt three classical microstructure estimators to discrete-outcome on-chain CLOB venues:

  • PIN (Probability of Informed Trading) adapted for binary markets
  • VPIN (Volume-Synchronized PIN) for real-time monitoring
  • Kyle's λ (price impact coefficient) as a liquidity-based signal

These are combined with on-chain wallet features: novelty score, funding concentration, and cross-market behavior clustering.

Validation

The ILS framework is validated on the ForesightFlow Insider Cases (FFIC) inventory — eight documented or strongly-suspected insider-trading episodes spanning 24 markets between April 2024 and April 2026.

Cite this work

@misc{nechepurenko2026foresightflow,
  title  = {ForesightFlow: Real-Time Detection of Informed Trading in Decentralized Prediction Markets},
  author = {Nechepurenko, Maksym},
  year   = {2026},
  note   = {Working paper, v0.95}
}