Quantitative research on prediction markets — microstructure, forecasting, on-chain forensics, and AI agents. Research department of Devnull FZCO (Dubai, UAE).
Prediction markets are now a third venue for price discovery alongside equities and crypto, but they lack the research depth that classical markets developed over fifty years. ForesightFlow exists to close that gap.
We publish open research, develop benchmarks and detection systems, and release datasets that make the field reproducible. Our long-term thesis is that quantitative trading, market design, and AI forecasting on prediction markets will become a serious sub-field of quantitative finance — and the research infrastructure should be public.
- Microstructure — informed-flow detection; PIN, VPIN, and Kyle's λ adapted to discrete-outcome markets; order imbalance and variance-ratio diagnostics.
- Forecasting & AI agents — proper-scoring evaluation, calibration, LLM ensembles, and on-chain agent reputation.
- On-chain forensics — wallet clustering, funding-flow analysis, novelty scoring, and cross-market behavior.
- Quantitative strategies — alpha generation, market making, arbitrage, execution, and slippage modeling on hybrid CLOB venues.
- Mechanism design — resolution typology, oracle integrity, and market designs that resist manipulation.
- Event-linked derivatives — resolution-aware perpetual futures on binary prediction markets; margin calibration under discrete-outcome risk; behavioral tier classification and fill-side microstructure of non-retail participants; manipulation and regulatory analysis of leveraged event-driven products.
Event-Linked Perpetuals programme (four-paper series, 2026):
- Fill-Side Non-Retail Trading on Polymarket: Behavioral Tiers and Microstructure Signatures — Paper 4, arxiv-ready. 13.4M on-chain fills · 43,116 markets · 77,203 addresses · behavioral cluster pipeline + six-tier classification. Dataset: pmxt-behavioral-clusters-v1.
- Manipulation, Insider Information, and Regulation in Leveraged Event-Linked Markets — Paper 3, working paper.
- Resolution-Aware Perpetual Futures on Binary Prediction Markets — Paper 1, working paper. Dataset: pmxt-stylized-facts-v1.
- Counterfactual Replay of Resolution-Zone Protocols on Polymarket — Paper 2, working paper. Dataset: pmxt-counterfactual-replay-v1.
Other recent preprints (2026):
- ForesightFlow: Real-Time Detection of Informed Trading in Decentralized Prediction Markets
- Foresight Arena: A Decentralized On-Chain Benchmark for Evaluating AI Forecasting Agents
- Focal: Signal Credibility Index for Prediction Markets
Full list, PDFs, and BibTeX at foresightflow.org/publications.
- 📄 Publications — foresightflow.org/publications
- 📦 Datasets — github.com/ForesightFlow/datasets (CC-BY-4.0)
- 💻 Code — across this organization (MIT, where applicable)
- Web — foresightflow.org
- Email —
info@foresightflow.org - Affiliation — Research Department, Devnull FZCO, Dubai, UAE