Duncan Anderson
Back to Projects

Sports Analytics Platform

The Lineup

Full-stack sports analytics platform with ML projections, real-time odds comparison, and automated betting analytics.

01

The Problem

Sports bettors lacked a single platform to get reliable player projections, compare odds across sportsbooks, and identify mispriced lines. Existing tools were either prohibitively expensive, too shallow, or fragmented across dozens of sites — forcing users to rely on intuition over data.

02

What I Built

I designed, built, and operate The Lineup end-to-end as a solo founder. The platform covers NBA, NFL, NHL, and college basketball.

ML PipelineLightGBM models trained on 198 engineered features including exponentially-weighted moving averages, defense-vs-position matchup adjustments, pace factors, and usage deltas. Models use per-stat Huber loss tuning to handle outlier performances. The pipeline processes nightly projections across full league slates with a bettable-population MAE of ~4.9 points, ~2.0 rebounds, and ~1.5 assists.

Real-Time Odds EngineIngests and normalizes odds from multiple sportsbooks, surfacing line discrepancies, expected value calculations, and arbitrage opportunities. Includes automated settlement that grades every pick against actual results.

Full-Stack PlatformFastAPI backend (Python 3.13) with Redis caching, Next.js 15 frontend with TypeScript, PostgreSQL on Supabase, and deployed on Railway. Stripe-integrated subscriptions, email campaigns via Postmark, and a native iOS app via Capacitor.

03

The Outcome

Launched to 200+ paying subscribers generating recurring revenue. The accuracy tracking system — which publicly grades every projection against results — became a key trust signal and growth driver. I built, shipped, and continue to operate every layer: ML training, data pipelines, backend, frontend, billing, and growth.

Stack

Tech Used

PythonReactMachine LearningPostgreSQLAWS

Interested in something like this for your business?

Let's Talk