Research

We decode the microstructures of prediction markets to identify and capture structural alpha. By combining Bayesian inference with graph-based relationship detection, we maintain probabilistic coherence across fragmented liquidity pools, building execution strategies that remain durable across market regimes.

Research Philosophy

Targeting Structural Inefficiency

We target persistent structural inefficiencies in how prediction markets price risk, not short-lived patterns. Focus areas: fragmented liquidity, delayed arbitrage, resolution-rule ambiguity, and violations of basic probability constraints that persist across regimes.

Adversarial Stress-Testing

Signals are treated as adversarial hypotheses and stress-tested across regimes, execution costs, and failure modes. Proprietary event-driven backtesting reconstructs historical market sequences through replay, exercising the same signal pipeline, risk constraints, and execution logic used in live trading. Synthetic simulation generates scenarios across calm, volatile, and panic regimes.

Real-Time Feedback Loops

Live monitoring feeds back into the research loop. Edge decay is detected early, and capital is dynamically reallocated. The system maintains observability from raw data ingestion through portfolio impact.

Execution Infrastructure

Low-Latency Infrastructure

We treat the operating system as a legacy bottleneck. Our proprietary environment utilizes kernel-bypass architecture and bare-metal isolation to achieve deterministic execution. By stripping away system jitter and shielding our models within the processor’s cache hierarchy, we have collapsed the window between signal and impact—ensuring our edge is captured at the absolute limit of the hardware.

Core Research Pillars

Market Microstructure & Dynamics

Regime Forecasting: Modeling phase-aware transitions and liquidity shocks.
Decay Modeling: Calculating information relevance with regime-aware decay rates.
Order-Flow Dynamics: Analyzing cross-venue dislocations and execution routing.

Structural Coherence

Graph-Based Detection: Multi-factor scoring across logical, temporal, and structural dimensions.
Resolution Parsing: Machine-readable rule extraction to identify contract ambiguities.
Probabilistic Scoring: Enforcing internal consistency across related market outcomes.

Simulation & Validation

Event-Driven Backtesting: End-to-end market reconstruction via high-fidelity state replay.
Synthetic Simulation: Monte Carlo stress testing against structural failures.
Adaptive Sizing: Reinforcement learning for execution under variable constraints.

Interested in research collaboration?

Let's Talk