Quantitative R&D for systematic trading
Turn a market thesis
into something you can trust.
Volatility Street turns trading ideas into falsifiable quantitative research, validated strategy systems, and production implementations. We define the objective, reconstruct what was actually knowable at decision time, test simple explanations before complex models, and carry the specifications that survive into production.
Deep specialization in volatility, derivatives, and market structure.
Why the process exists
Most false alpha is created before the model is trained.
Wrong target.
The model predicts something adjacent to the trade rather than the actual economic outcome.
Future information.
A threshold or historical field quietly uses information that wasn't available at the time.
Redundant evidence.
Five correlated variables look like five confirmations.
Complexity without a benchmark.
A sophisticated model "works" but fails to beat a one-line rule.
Research/live drift.
The strategy that gets deployed isn't the strategy that was tested.
Our process is designed to catch those failures before capital depends on them.
Research in practice
A volatility thesis changed materially before the first fitted model was allowed to run.
The value wasn't a more complicated model. It was discovering what the model actually had to explain.
- 01
The original prediction target did not represent the trade.
- 02
An inherited threshold was found to contain future information.
- 03
Two apparently separate signals were shown to encode the same condition.
- 04
An early historical feature did not represent a stable tenor and was removed.
- 05
The original futures source was rebuilt from official exchange archives.
- 06
A simple structural rule became the benchmark any ML system had to improve on.
Bring the market judgment. We build the quantitative research function around it.
Research protocol
Specify the question, reconstruct the information set, establish the simplest explanation, and hold out the verdict. The model comes later.
How the research worksFor investment managers
Strategy R&D, production systems, and technical diligence support - a quant research function without building the whole desk.
Working with managersInfrastructure
Point-in-time data, experiment tracking, validation harness, and live serving - built before the engagement starts.
What's already builtIn the past year, four candidate strategies entered our validation protocol under pre-registered criteria. Most failed. Every verdict is provable.
Most strategies don't survive the process.
The ones that do are worth trading.
Tell us what you're working on. We'll let you know if there's something testable.
Discuss a research problem