The horse

We start with the mundane: data pipelines, operational teams, data quality processes and systems, high-quality data sources, ideally some of those unique, strict architectural reviews, and strict “infrastructure-as-code” platforms. We always build on the top cloud platforms and we never skimp on tools. We create — and then assiduously pay off — technical debt. We want the behind-the-scenes stuff to gleam even more than the part that is visible. This is true for data, data operations systems, infrastructure, and code that generates predictions from the model — the inference code. Data pipelines should gleam.

The astronaut

Only once we have great basics do we turn to the math and the science. We build a platform to do data science, and then build predictive models. We create and validate and re-validate those models. We turn to the top academic sources, and often leverage our personal contacts with professors. We test our hypotheses. We leverage first principles. But we also spend enormous time comparing our real-world results to our hopeful hypotheses.

The princess

The third piece is the princess, which represents imagination, emotion, and creative problem-solving. It also represents deep interest in the first principles and mechanics relevant to the domain. In a way, the princess represents an attention to detail outside of both math and engineering. When there is an end user, the princess includes art and design, and allowing for an emotionally satisfying connection for the user. In investing, the princess includes making research and data frameworks that are infinitely flexible — in recognition that the quant researcher is a creative, and an artist, who relies on a kind of rigorously tested artistic intuition to guess at a solution. We recognize that our kind of rigor and engineering have to be flexible to support and cultivate creative thinking.

Paradoxically, the princess converts mathematical precision and engineering to real-world power.