Insights topic

AI & Data

How we build research data that never sees the future, where machine learning helps our investment process, and the controls that keep it honest.

Latest on AI & Data

What we are writing on AI and data.

Newest first

Lit office windows forming a grid at night

What we cover

Better data before better models.

Most research errors start in the data. Our AI and data coverage explains how we store information as it was known on each date, and where models, including language models, earn a place in the process.

  • What did we know, and when?

    Point-in-time storage, data revisions and how look-ahead bias enters a backtest.

  • Where do models help?

    Reading documents, reviewing code and finding patterns, with a person accountable for every decision.

  • How are models controlled?

    Validation, logging and model risk reviews before anything reaches a portfolio.

AI and data research is written by engineers and researchers on our Technology and Research team in San Francisco, who run the shared platform used by every desk.