Paris · quant research · automation
Quant researcher
and developer.
I design and test systematic strategies for asset management: signal research, portfolio construction, statistical validation. I also build, and I take on freelance work in automation and LLM agent integration.
01About
An engineer by training, I do quantitative research for a Paris-based asset manager. My job is to turn a market intuition into a testable hypothesis, then into a strategy that holds up under scrutiny.
I also build the tooling that carries that research, because an idea you can neither test quickly nor put into production is worth very little. That same taste for reliable systems drives my freelance work, where I help companies replace manual data entry and processing with automated, supervised pipelines.
02Quant research
Designing strategies, and proving they hold
Signal research on market and alternative data, factor selection, portfolio construction, performance measurement and attribution.
Most of the work goes into what separates a real result from an artefact: out-of-sample test protocols, controlling for overfitting and survivorship bias, sensitivity to transaction costs and market regime. A backtest only proves something once you have seriously tried to break it.
03Freelance
Automation and LLM agent integration
I integrate LLM agents and automated workflows into existing business processes, from scoping to production.
-
Document extraction
Processing pipelines that read your documents, extract structured fields and push them into your tools, with human review reserved for doubtful cases. Your staff validates instead of typing.
-
Automated workflows
End-to-end automation of repetitive processes, with logging, error handling and monitoring. To me, a deliverable is a system running in production.
-
White label
For agencies with more inbound demand than delivery capacity: I take a defined scope, I deliver, you keep the client relationship.
How we work together
- Scoping
A conversation to bound the perimeter, then a written note: what can be automated, what cannot, and what it changes in time and reliability. You leave with a decision, even if that decision is to do nothing.
- Pilot
One chosen use case, delivered to production on a reduced perimeter. That is where it gets checked against your real data, not a prepared demo.
- Run
Once the pilot holds, extension to the other cases and upkeep under real conditions: monitoring, error recovery, changes.
04Projects
Recent work
-
Signal research
Building and validating signals on market data, from research prototype through to production monitoring.
-
Investment tooling
A suite unifying positions, performance, benchmarks and news behind a single access layer, usable from a web interface and by LLM agents alike.
-
Automated investment commentary
Weekly and monthly writing, with a strict split between attribution computed by code and prose written by a model.
-
Market data as a wallpaper
An implied volatility surface and a correlation map of the equity universe, rendered in 4K from public data. Free to download.
-
Document extraction
Pipelines that read documents, extract structured fields and only escalate doubtful cases to a human.
-
Online slot games
Casino game math engines, statistical verification of return to player by simulation, and web rendering integration.
05Contact
A process to automate, a project to scope?
Describe what you need in a few lines, I will get back to you quickly.
contact@nazzareno.fr