Project · 2026
The Tale of Two Collapses
A win-probability study of two of the biggest blown leads in sports — the Spurs' 29-point 2026 Finals collapse vs. the Falcons' 28-3 — plus an NBA contract-value analysis, all from live ESPN and Basketball-Reference data.
Two teams crossed a 99% win probability — and lost. This pulls ESPN’s win-probability and play-by-play data to overlay the Spurs’ Game 4 Finals collapse (a blown 29-point lead) against the Falcons’ 28-3 Super Bowl, normalized to share of game so two different sports compare fairly.
How safe is a “sure thing”? Scraping every game of the 2026 NBA playoffs (90 games, ~44k win-probability readings), a team reached a 95%+ win-probability spot and still lost in 11 of them, and 99%+ three times. The model is well-calibrated overall — shown with 95% Wilson intervals — but slightly overconfident at the extremes.
Bonus — contract value. A small scraper refreshes current-season advanced stats and salaries from Basketball-Reference to surface the league’s best-value contracts: each player’s production priced at the market rate of ~$3.3M per win share, minus actual salary.
Built on a reusable ESPN / Basketball-Reference scraping core and a two-tier rendering system — a bold shareable style and a clean analytics style — sharing a single data pipeline.