Writing

Notes from the team.

Technical posts on synthetic market data, backtest validation, and the joint dynamics that make a strategy survive contact with reality.

  • June 12, 2026

    In practice: selecting strategies on synthetic histories

    Three ways to pick the best trading strategy out of 34: a backtest, a bootstrap, and a generative model. Two of them picked a strategy that memorized its own training data. Only one didn't.

    Read →

  • June 4, 2026

    Introducing Sablier-Flow

    A Python SDK that turns your backtest from a single P&L number into a distribution, by simulating thousands of alternative versions of history that share your data's statistical fingerprint.

    Read →