RESEARCH
The questions
behind the software.
Notes on data, model evaluation and research automation. Each one examines a practical problem that shapes how we build.
01 — THE READING ROOM
Research notes
Browse by topic or search for a question.
3 notes
01Research the data you could have known.
Observation dates, publication dates and vintages answer different questions. A historical research dataset needs all three.
What a strategy test needs to tell you.
A useful test explains the rules, trading costs, evaluation period and conditions under which the result changes.
Where AI belongs in a research workflow.
Use language models to help read and organise source material while keeping numerical work and trading decisions accountable.
ABOUT THESE NOTES
These notes describe our research principles and development direction. They are explanatory articles, not performance reports or trading recommendations.