Research/Stealth
How we think about markets.
Research is the core of what we do. We study how markets are structured, where edges come from, and how to turn evidence into systems that hold up out of sample. Most of our work is private while we are in stealth — this is the shape of it, and the principles behind it.
00—How we work
A small set of principles keeps our research honest and our results reproducible.
Evidence over narrative
We start from data, not from a story we want to be true. Hypotheses are written down before they are tested, and a result only counts once it survives costs, slippage and out-of-sample checks.
Reproducible by default
Pre-registrations, frozen configs and validation scripts are part of the work, not an afterthought. If we cannot rerun it and get the same answer, we do not trust it.
Explainable by design
A signal we cannot explain is a signal we cannot defend. We build decisioning that exposes its drivers, so risk, compliance and clients can audit why a model did what it did.
01—Research areas
The questions we keep coming back to. These areas span our DeFi, systematic-trading and AI-infrastructure work, and inform the engagements we take on.
Market microstructure
How orders, liquidity and latency shape price formation. We study order-flow, spread dynamics and execution to understand where structural edges live and what they cost to capture.
Factor & alpha research
Designing, testing and combining signals — momentum, low-volatility, reversal and beyond — with transaction-cost modelling and honest out-of-sample evaluation, so an edge survives contact with the market.
Regime detection
Markets do not behave the same way for long. We work on identifying volatility and market regimes and adapting allocation and risk accordingly, rather than assuming one model fits every state.
DeFi & MEV
On-chain market structure: arbitrage cycles, pool imbalances and MEV-aware execution across EVM chains. From graph-model pathfinding research to detection and optimization for live routing.
Explainable & agentic systems
Making automated decisions auditable, and making multi-agent automation safe to run. Causal decomposition and confidence scoring for signals and risk, plus governance and durability for agentic workflows.
02—Publications
We publish when the work is ready to stand on its own. Today that means one reproducible-research companion, with more in progress while we are in stealth.
03—Research notes
We share selected research as notes with partners and clients. A public stream is on the way.
More write-ups go to partners and clients directly while we are in stealth. Ask to be notified when the public stream expands.
Research you can interrogate, not just trust.
If you want a team that shows its work — and can put it behind your risk and compliance — let's talk.