01
Data and integrity
The Python project includes data validation and processing code, historical-data tests, temporal-feature tests, and a benchmark-integrity script. These checks help make assumptions and data problems visible.
Case study 03 / Independent research
NFL Statistical Research System
A Python-based football analytics project focused on data processing, model evaluation, integrity checks, and repeatable experiments.
The problem
Context & approach
Statistical experiments are only useful when the inputs, evaluation process, and limitations can be examined.
I built an independent research workflow around processing football data, evaluating models, and checking the integrity of experiments before interpreting results.
The implementation
Engineering decisions
02
Evaluation workflow
Modeling and evaluation code is paired with tests and repeatable experiment workflows. The emphasis is on inspecting evidence and maintaining reproducibility.
03
Engineering practice
The project uses pytest, type checking, and linting alongside its analytics dependencies. The presence of this tooling is not a claim that a model has demonstrated reliable predictive performance or profitability.
Let’s discuss the work.
I’m happy to walk through the implementation, verification process, and project limitations in a conversation.
Email Jeff ↗