Work
Two shelves: what I've done in games, and what I chase for its own sake.
The professional work is generalized — no former employer's absolute revenue or player numbers appear here. The personal projects are entirely my own data and methods, and I'm happy to show the workings.
In games
Measurement tied to a real product or revenue decision.
The forecast that caught an eight-figure error
A long-term revenue model, built in R and Python for an M&A assessment, that found and corrected a material multi-million-dollar overestimation before the deal closed. Diligence tends to reward the person who reads the footnotes.
Feature lift without a clean experiment
Matched-cohort causal inference to separate real monetization lift from self-selection when an A/B test wasn't feasible — the evidence base that kept a companion web app funded, later cross-validated by a randomized messaging test.
An alliance recommender that moved retention
A Python recommendation engine and churn diagnostics for a live-service strategy game, tied to a measurable, mid-single-digit lift in annual revenue.
Fourteen hours to under one
Modernized a studio's analytics stack — fragile R ETLs into incremental dbt/Athena models and Prefect workflows — cutting the daily refresh from up to fourteen hours to under one, at lower cloud cost.
A fuller, permission-cleared write-up of the companion-app work lives on the Savepoint Analytics site.
Off the clock
Projects I built because the question wouldn't leave me alone.
The Election-Year Install Tax
US install prices rise in the fourth quarter of an election year because presidential campaigns are bidding for the same impressions. It is real, it is specific to one market, and it is the one piece of seasonality you cannot fit.
Cohort LTV Forecasting
Spend to CPI to installs to actives to spenders to revenue per user to LTV. Seven links, each with its own estimator, and a chain whose weakest point is not where anyone expects it.
8 parts
A/B Testing in Games
A decade of running experiments on live games, reduced to seventeen rules anyone can apply, a worked case study for each way an experiment lies to you, and the module I built so the pipeline refuses to make the same mistakes twice.
10 parts · 6 interactive reports
Why Simulate Demand
Bias requires truth. A simulator lets you set the truth, hand an estimator only the observable view, and measure how far off it lands — the one measurement real market data can never give you.
4 parts · 1 interactive report
Savepoint Analytics
I believe in the 80/20 rule — right up until the last 20% is the answer. What Savepoint is, why it exists, and the one idea it is organized around.
The sabermetrics and U.S. elections work is still in the notebook stage — it lands on the writing page as it's finished.