Inspiration
Our spark was a local pressure point: the SEPTA service cuts. When FanDuel stepped in with an $80,000 one-time sponsorship to restore Broad Street Line express service for the Philadelphia Eagles’ 2025 home opener—including free post-game rides—it underscored how fragile transit funding can be and how sports can mobilize solutions. That civic moment pushed us to build something useful around sports data.
What it does
Gemline ingests MLB game data from Sportradar (an official distributor) and pairs it with Google’s Gemini API. We model player and team performance at the game level, compare sportsbook lines, and generate matchup analyses with predicted edges and narrative context. The system surfaces what matters—who’s trending, where odds diverge, and why.
How we built it
We started with wireframes and user stories, then divided roles. Two teammates focused on implementation and UI/UX; the rest drove research, story mapping, and design polish. The stack integrates two Sportradar APIs for structured data and routes curated features into Gemini for reasoning and generation.
Challenges we ran into
We initially aimed for industry-grade depth with MLB Statcast features. Given time and access constraints, we narrowed scope to reliable game-level stats from Sportradar—enough to ship a solid MVP without overreaching.
Accomplishments that we're proud of
We delivered something that looks sharp and holds up technically. We chose a hard path on purpose, wrestled with unknowns, and still shipped on time. That balance—clarity in the UI, discipline in the data, and measured claims—was the win.
What we learned
Gemini is a flexible layer for structured sports data: great at turning numbers into explanations fans and analysts can use. Despite AI in sports being relatively niche, we believe that in the long run the demand for timely, explainable analysis is real.
Built With
- gemini
- nextjs
- sportradar
- tailwind
Log in or sign up for Devpost to join the conversation.