.png)
About Basenerd
Basenerd is a stats-first MLB analytics platform built for fans who want more than box scores. We combine real-time MLB data with proprietary machine learning models to deliver insights you can’t find anywhere else.
What We Do
Live Scores & GameCast — Every game, every pitch. Our gamecast tracks the action in real time with pitch-by-pitch data, live score updates, and play-by-play breakdowns.
Matchup Predictions — Our XGBoost machine learning model evaluates every batter-pitcher matchup and predicts the probability of every possible outcome: strikeout, home run, single, walk, and more. Predictions update live during games with Bayesian adjustments based on pitcher velocity and fatigue. Read how the model works.
Pregame Predictions — Before first pitch, see projected outcomes for every batter in both lineups against the opposing starter. HR probability, K rate, hit rate, and hot/cold streaks — designed for fans who want to understand matchups and bettors looking for edges on player props.
Player Profiles — Deep-dive pages for every MLB player with year-by-year stats, career totals, awards, and accolades.
Pitcher Reports — Detailed pitching dashboards with arsenal breakdowns, pitch movement charts, release point analysis, location heatmaps, and our proprietary BNStuff+ and BNCtrl+ grades for every pitch type.
Standings & Schedules — Live standings for every division, team schedules, and transaction logs.
Stat Leaderboards — Sortable leaderboards across dozens of batting and pitching metrics.
Team Pages — 40-man rosters, schedules, transactions, and team-level analytics.
The Analytics
Basenerd isn’t just a stats aggregator. We build our own models:
- BNStuff+ — Our proprietary pitch quality model that grades every pitch type on a 100-point scale based on velocity, movement, spin, and release characteristics
- BNCtrl+ — Command and control grade measuring a pitcher’s ability to locate pitches in the zone
- Matchup Model — 86-feature XGBoost classifier trained on 730,000+ plate appearances that predicts PA outcomes using batter profiles, pitcher arsenals, pitch-type matchups, park factors, game context, and rolling 14-day form
- Pitch Selection Model — Predicts what pitch a pitcher is likely to throw given the count, situation, and arsenal
All models are trained on Statcast pitch-level data from 2021-2025.
The Tech
- Backend: Python / Flask
- Data: MLB StatsAPI + Statcast (PostgreSQL)
- ML: XGBoost, scikit-learn
- Frontend: Vanilla JS, server-rendered templates
Built by Nick Labella.
