FYOS 1101: Risk, Analytics, and Football
UGA's First-Year Odyssey seminars are designed to introduce first-year students to the academic life of the University. These seminars will allow these students to engage with faculty and other first-year students in a small class environment to learn about the unique academic culture the University offers.
Course description
A single football game involves thousands of decisions by coaches and players. Millions more decisions are made over entire seasons and over a player's career. All of these decisions involve risk – there are many possible outcomes, which may be good or bad. The analytics revolution seeks to help inform these decisions using statistical analysis. In this course, we will use statistics to understand how players and coaches measure, evaluate, and manage risk. Exercises may involve deciding the best course of action on a single play, quantifying risk attitudes, predicting game outcomes, and evaluating players. This will be a math-heavy course with extensive use of Microsoft Excel. Students unfamiliar with Excel are expected to learn the software independently (YouTube is a great resource).
Course Materials
The "Weekly Agenda" document provides a description of what we are doing each week.
The "Class Materials" folder includes subfolders for each week with the requisite class materials. If a week's subfolder is missing, it is likely we had a guest speaker with proprietary materials.
Prompts for each of the semester's four writing assignments are in the "Writing Assignments" document.
Links (articles, videos, etc.)
My repository of articles on football analytics, organized by tag
My YouTube playlist for football analytics
My Twitter list of football analytics folks
Datasets
NFL
nflfastR - a set of R functions to scrape NFL play-by-play data since 1999 (here's a good video tutorial). Part of the larger nflverse library (if you install nflverse, it will include nflfastR). OpenSourceFootball does some great work using this data and posts a lot of their code.
Stathead Football - many similar stats to nflfastR but more user-friendly. Web-based and searchable if you're looking for a specific stat or play. Also includes trades, draft picks, and combine results. More advanced searches require a subscription, and they have a 50% student discount.
NFL Big Data Bowl 2023 - very detailed data provided to competitors in the NFL Big Data Bowl. Includes play-by-play, PFF's scouting data, and player location tracking data. Not sure how available it is outside the competition. Kaggle also has this data from previous years.
ffanalytics - R package that collects projected stats and points scraped from sites with publicly-available projections.
Pro-football-reference - I think this is a partner of Stathead as they have much of the same data (e.g., combine results) and link to each other.
For NFL draft data and contract values, take a look at spotrac and OverTheCap.
College
CollegeFootballData - a huge repository of college football data (play-by-play, recruiting, advanced stats, betting lines, etc.). See the API docs for all the available data. There's a good guide to accessing the data using Python here.
Sports-Reference CFB - many similar stats to CollegeFootballData. Kind of like the equivalent of Stathead for CFB.
PFF Advanced Stats - team and player-level data collected by Pro Football Focus. Most of it requires a subscription, but they have a 50% student discount for all paid tiers.
GameOnPaper - very neat tool to look at how two teams match up in a game.
Winsipedia - historical matchups between teams.
Data visualization
Football
Grinding the mocks charts expected draft position
BCFToys has some great visuals
General
Data visualization guide from Johns Hopkins
DataVizCatalogue - catalogue of different types of graphs/charts
R Graph Gallery - different types of graphs/charts and how to make them in R
Intro to Excel and graphics (a YouTube playlist I created, there are many others out there)