The Projects
The Dawgometer
A custom multi-factor model that rates NHL skaters on a single scale, published under the @dawgometer name. Designing it meant deciding what a skater's value actually consists of, turning several different kinds of on-ice contribution into one comparable number, and keeping the scale honest as the season's data accumulates.
Weekly power rankings — three model variants
Every week I publish NHL power rankings in three variants of the model: Balanced, Results-First, and Process-First. Same league, same week, three defensible answers — publishing the variants side by side is a deliberate choice to show that model output depends on modeling judgment, not pretend there is one objective ranking.
Monte Carlo playoff simulator
A Monte Carlo simulator that runs 5,000 simulations of the playoff picture, turning a single predicted outcome into a distribution of outcomes. The point of simulating instead of forecasting once: decision-makers deserve probabilities, not false precision.
Earlier: Tableau for camp decisions
Before Organon and Mass General, at Pro Ambitions Hockey I used website data to create Tableau visuals that supported decisions on new camp locations — an early version of the same pattern: operational data in, a visual a decision-maker can act on out.
What This Shows About My Data Modeling
- Modeling is choosing. A multi-factor rating lives or dies on which factors you include and how you defend them — the same judgment as designing the fields and definitions in an enterprise data model.
- One dataset, multiple honest views. The three rankings variants make model assumptions visible instead of hiding them — the analytics version of documenting your data definitions.
- Distributions over point estimates. The 5,000-simulation playoff model reports likelihoods, which is how uncertain real-world systems should be presented.
- A pipeline, not a one-off. Weekly rankings mean recurring data collection, a repeatable process, and output that has to be right on a schedule — the discipline production data work runs on.
The projects are public: search @dawgometer on X for the Dawgometer, the weekly rankings, and the playoff simulator output.