ANC Labs
ANC Labs is a one-person, independent research lab working on robust machine learning for low signal-to-noise financial data. The echo in the name is no accident: the core of the work is suppressing noise so the signal can be heard. Research is the tool; the end goal is systems that earn sustainable returns in live markets, after costs.
What we work on
Research
Noise-robust model architectures, synthetic labs with a known signal, and the limits of learning at low signal-to-noise.
Measurement
Pre-registered tests, after-cost evaluation, and test infrastructure that catches the errors that make results look better than they are.
Engineering
Data pipelines, point-in-time datasets, a decision engine that runs in real time, and the tools that monitor it.
One pair of hands, a process that checks itself
The upside of a one-person lab is that every layer comes from the same hands: from data collection to model development, from test infrastructure to the live engine. The path from an idea in the synthetic lab to the production system is short and unbroken.
We know the downside too: there's no team to catch your mistakes for you. So the process is built to check itself. Every test is written down before it runs, every number that drives a decision is reproduced independently, critical safeguards are tested by breaking the code on purpose, and every idea we try is recorded, whatever the outcome.
Founder · quantitative developer and researcher
Designs and runs the research, data infrastructure and live system at ANC Labs.
Get in touch
For questions about the research, possible collaborations or the system, send an email.