Alpsilon Research

Alpsilon Research explores alternative data and AI methods for questions in quantitative trading, model risk and portfolio research. We turn overlooked observations into testable hypotheses, then examine whether they add information beyond established baselines.

Alps, epsilon and AI.

Rooted in the Swiss Alps, Alpsilon uses AI to uncover alpha within epsilon. The name brings together the Alps and epsilon, with alpha as the research objective and AI as a set of methods for exploration.

The Alps inspire perspective and discipline. Epsilon represents the residual: what a model does not explain. An unexplained observation can open a useful question; evidence determines whether it becomes a meaningful pattern.

Curiosity, with a standard of evidence.

Our approach starts with data provenance, timestamps and a clear question. Evaluation uses time-aware tests and independent baselines, with attention to leakage, multiple testing, trading costs and model uncertainty.

Quant and AI risk are part of the research question. A pattern must be examined across conditions, including where it fails. Reproducibility, uncertainty and limitations belong alongside results.

Explore our methodology

Research and project collaboration.

We welcome discussions with potential clients, data providers and research partners about alternative datasets, quantitative research questions and AI project collaboration. A useful starting point is the objective, available data, evaluation criteria and intended deliverables.

Research collaborations are scoped individually. Signal data subscriptions and the research platform are in development; availability and project scope should be discussed directly.

Contact Us