Career
Sports analytics as a career, in four stages: what the roles are, what gets you hired, what to build first, and how to read a posting. Each stage names the session or course that teaches it.
The titles vary by sport and by front office, and the work underneath them sorts into four kinds. The analyst answers questions with data: a projection for a player, a read on an opponent, a number for a negotiation. The craft is the model and the sentence that explains it, and it is taught in Know Exactly When an MLB Player Starts to Decline, What Fills the Seats and Becoming a BayeZian.
The engineer makes the data arrive and the models run without anyone pressing a button: feeds, pipelines, schedules, alerts. Game-Day Automation and Dockerizing a Match-Day Shiny App So It Runs Anywhere are that job in miniature. The tools builder turns a model into something a coach or a scout opens on a phone, which is the whole of ProductioniZing Shiny Applications, The Farm System and The No-Look Pass. The lead decides which questions are worth a quarter of the group's time and owns the system the other three work inside.
Most people enter through the first or third door and move toward the fourth. The roles share one spine: code that reruns, a model you can defend, and a result someone else can use without you in the room.
A body of work that runs. A notebook that answers a question nobody handed you, with the data pulled, cleaned and plotted in one script that gives the same answer tomorrow: FoundationZ of Data Science in Python, BreeZing Through the Tidyverse in R. A model where you can say how sure you are and why, which is the subject of MCMC in Practice and Hierarchical Modeling using PyMC in Baseball. A deployed app or pipeline with a URL, because a thing someone else can open is a different category of evidence from a thing you describe.
Then the conversation about it. The work is judged by people who do the work, and they ask the same question every time: what did you decide when the data ran out, and why. The sessions here are built to give you that answer in your own words, because every one of them is taught by someone who has had to give it.
In order, and each one small enough to finish. First, a reproducible report on one season of one sport, from raw data to a chart with a sentence under it. Second, a model with uncertainty: a hierarchical model that pools a player toward the league and says how far, the material of Hierarchical Modeling using PyMC in Baseball and Becoming a BayeZian. Third, an app that puts that model in front of a person who does not code, scaffolded the way Kickoff to Production in Minutes shows and made production-grade the way ProductioniZing Shiny Applications teaches. Fourth, a pipeline that refreshes it on game day without you, which is Game-Day Automation.
Add the AI layer as you go, not at the end: a natural-language front door over your own data is Ask the Box Score Anything, and an agent that turns unstructured text into rows is Build an AI Agent That Turns Cricket Commentary Into Data. Three worked problems in this shape, with code, are on the problems page.
Read the verbs before the nouns. The tools list tells you the stack; the verbs tell you the job. "Build and maintain" is an engineering role whatever the title says. "Communicate findings to coaching staff" means the deliverable is a sentence, and the model is the means. "Own" means a lead.
Separate required from preferred, and read the required list as the floor you must clear, not the ceiling. Then find the one sentence about a decision the group makes, because that sentence is the real job description, and build your first piece of work to answer it. The postings we are reading now, each mapped to the sessions and courses that answer them, are on the jobs board.
Every session and course named on this page is on the home page, with its price and its buy link. Each is a recording you buy once.
1-on-1 mentorship and certification are coming. Neither is open yet, and nothing about them is for sale on this site today. If you want to talk about either before they open, book a short call.
First name and email. One message when a session recording, a skill, a cheat sheet or a posting worth your time arrives here, and no other mail. Replies reach a person.