Market Microstructure
Understanding how information, liquidity, and participant behavior interact at the finest observable timescales.
Research
Find the structure.
Test the evidence.
We study markets as adaptive systems, combining statistical rigor, economic reasoning, and implementation-aware experimentation.
Research disciplines
Many lenses. One standard of evidence.
Useful insight can come from many methods. What matters is whether it survives costs, instability, and honest attempts to disprove it.
Predictive Modeling
Combining statistical learning with domain structure to build models that generalize beyond their training window.
Portfolio Intelligence
Translating many uncertain signals into coherent exposures under explicit capacity and risk constraints.
Adaptive Systems
Detecting regime change, monitoring model health, and adjusting decisions as market structure evolves.
Alternative Data
Evaluating novel datasets through a disciplined process centered on provenance, stability, and economic meaning.
Execution Science
Measuring implementation cost and designing policies that balance urgency, information leakage, and liquidity.
The research loop
Designed to learn, not confirm.
Each stage has explicit artifacts, owners, and tests so an idea can be reproduced, challenged, and improved.
01Form
Hypothesis
Begin with a falsifiable idea grounded in market behavior, not an unexplained backtest.
02Test
Evidence
Stress assumptions across time, assets, costs, and changing conditions with reproducible experiments.
03Build
System
Engineer the full decision path, including data quality, risk, execution, and failure behavior.
04Learn
Feedback
Monitor live evidence, diagnose drift, and continuously improve the complete research-to-production loop.
Research culture
Strong opinions, loosely held. Precise records, permanently kept.
We value intellectual independence and direct debate. Ideas are judged by the quality of their reasoning and evidence, not by title or ownership.
Reproducibility is part of that culture. Data versions, assumptions, experiments, and decisions are recorded so results can be understood long after the original run.
Join the research team