Work / Independent / 2025
Performance Analytics Platform
Personal project · Overwatch competitive match data · 2024–2025
An end-to-end pipeline from a 15-metric PostgreSQL schema to a 2-page Power BI dashboard on Overwatch match data. Hero selection swung solo win rate by 43 points. Map choice swung it by 100.
The question
I coached Overwatch at a top-100 North America level. Players obsess over mechanics, but which decisions actually move the win rate, and by how much?
Approach
Designed a 15-metric relational schema in PostgreSQL. Wrote five production queries using window functions, CTEs and aggregations, then modeled the results in Power BI with DAX measures across a 2-page dashboard.
What I found
Hero selection produced a 43-point swing in solo win rate between the best and worst picks. Map choice produced a full 100-point swing. Some maps were near-certain wins and others near-certain losses. Choices made before the match even starts mattered more than anything during it.
Why it matters
It’s the same workflow as any KPI project. Define the metrics, model the data, find the lever, and show it so someone can act on it Monday morning. Here the lever was “stop queueing on your worst maps.”