AthlyticZ · Sport, data science and AI, on sports data

Real code on sports data, in your browser.

Pick a dataset, edit the code, press Run. R runs on webR and Python on Pyodide, loaded on your first run; nothing you type leaves the page.

# Pitch movement: one cloud per pitch type, every pitcher, and the per-pitch summaries the chart labels.
import json
import pandas as pd

pitches = pd.read_csv("data/pitch_movement.csv")

summary = (
    pitches
    .groupby(["pitcher", "pitch_type"], as_index=False)
    .agg(n=("hb_in", "size"), velo=("release_speed", "mean"), hb=("hb_in", "mean"),
         ivb=("ivb_in", "mean"), swings=("swing", "sum"), whiffs=("whiff", "sum"))
    .assign(usage=lambda d: d["n"] / d.groupby("pitcher")["n"].transform("sum"),
            whiff_rate=lambda d: d["whiffs"] / d["swings"].clip(lower=1))
    .round(3)
)

# a sample of each pitcher's pitches per type keeps the page light; the summaries use every pitch
points = (
    pitches
    .groupby(["pitcher", "pitch_type"], group_keys=False)
    .apply(lambda g: g.sample(min(len(g), 18), random_state=7))
    [["pitcher", "pitch_type", "hb_in", "ivb_in", "release_speed"]]
    .rename(columns={"hb_in": "hb", "ivb_in": "ivb", "release_speed": "velo"})
)

json.dumps({"points": points.to_dict("list"), "summary": summary.to_dict("list")})

This chart is the code's result. Run recomputes it in your browser.

Horizontal break against induced vertical break, one cloud per pitch type, contours of where each pitch lives. Pick a pitcher; hover for the averages.

Edit me: Change the sample size from 18 to 120 per pitch and run again, or add a column to the summary.

Data: Pitch movement and whiffs (pitch_movement.csv). Simulated for practice by AthlyticZ (agents/sim_data.py, seed 20261109). License CC BY 4.0.

The code runs in your browser on webR or Pyodide; nothing you type leaves the page.

Senior or Lead Analyst (Forecasting)

Colorado Rockies · Denver, CO

Build, validate and deploy predictive models and forecasts, communicate the uncertainty plainly, and turn one-off analyses into repeatable work; R or Python, SQL, reproducible workflows, four or more years in the role.

Teaches toward itSessionKnow Exactly When an MLB Player Starts to Decline SessionMCMC in Practice CourseBecoming a BayeZian

The posting, read 2026-11-08.

Vice President, Data Systems

Los Angeles Lakers · El Segundo, CA

Own the cloud data platform end to end, from ingestion to the web and mobile apps that put model output in front of coaches and the front office; orchestrated pipelines, CI/CD, infrastructure as code, Python, SQL and TypeScript, ten or more years with five leading.

Teaches toward itSessionGame-Day Automation SessionDockerizing a Match-Day Shiny App So It Runs Anywhere CourseMobile StructureZ

The posting, read 2026-11-08.

Data Scientist, Baseball Research

Toronto Blue Jays · Toronto, ON, Canada

Design and maintain predictive models and metrics for scouting, development and projections, with a demonstrated ability to use LLMs in the research process; SQL, R and Python, Bayesian modeling welcome, four to five years building models and explaining them to non-technical staff.

Teaches toward itSessionHierarchical Modeling using PyMC in Baseball SessionAsk the Box Score Anything CoursePython MachineZ: ML in Sports

The posting, read 2026-11-08.

Sport, data science and AI, moving together. Learn the methods where they are being built.

The 2026 line-up, each session a recording you buy once.

Real code on sports data, in your browser.

Every live room, and the whole catalog, for a year.

The postings we are reading now.

Senior or Lead Analyst (Forecasting)

Vice President, Data Systems

Data Scientist, Baseball Research

Four ways in, beyond a single session.