There is a version of quantitative consulting that produces a model so complex, so dependent on the consultant's institutional knowledge, that the client cannot use it without calling the consultant back. This is a business model. It is not a good one for the client. Every model we build at Quantum Lattice is designed from the start to be maintained by your team after we leave.

Documentation as a First-Class Output ¶

We treat documentation as a deliverable, not an afterthought. Every model we build is accompanied by a written document that explains the structure of the model, the source of every input, the logic of every formula, and the assumptions that would need to change if the model were to be updated. This document is written for a technically literate reader who was not present when the model was built.

Annotated Code ¶

For models built in Python or R, every function is annotated with a plain-English explanation of what it does and why. We follow a consistent style guide that we have developed over nine years of practice. The goal is that a competent analyst who joins your team two years from now can open the code and understand it without needing to contact us.

The Handover Session ¶

Every engagement ends with a structured handover session, typically half a day, in which we walk your team through the model in detail. We ask them to run scenarios we have not run, to change assumptions, to break it deliberately. The session is not a presentation. It is a working session, and it is the point at which we find out whether the documentation is as clear as we thought it was.

Choosing the Right Tool for the Team ¶

We build models in Python, R, or Excel depending on what your team is most likely to actually use. A Python model is more powerful and more reproducible. An Excel model is more accessible to a team without a data science background. We have a preference for Python, but we have learned that a well-maintained Excel model is more valuable than an abandoned Python one. The right tool is the one your team will open.

If you have inherited a model you cannot interrogate, or if a previous engagement produced outputs your team cannot reproduce, that is something we can help with. It is one of the more common situations we are brought in to address.