Methodology
For each tournament game, both teams are mapped to their university's home state (or city). A state- or city-level business metric is compared between the two locations. The team from the location with the higher (or lower) value is predicted to win. Each metric is tested in both directions; the better direction is reported. Same-location matchups are excluded.
Data: Hundreds of business metrics computed from Enigma's operating locations dataset. Men's tournament: 737 games across 10 tournaments (2014-2025, excluding 2020). Women's tournament: 683 games across 10 tournaments. Upset analysis uses tournament seed data.
Important caveat: All business metrics reflect today's data, not historical snapshots. When we test a metric against a 2015 game, we're comparing current business data for those cities or states. The underlying assumption is that the relative economic character of places is stable enough over a decade for this to be meaningful. (This is entertainment, so we're not losing sleep over it.)
Limitations: This is entertainment. A 55% accuracy rate is marginally above a coin flip. State-level metrics correlate with population and economic activity. City-level metrics are noisier due to smaller sample sizes. Multiple-comparison effects inflate the apparent significance of top performers.