Statbotics has gone through some major changes this season. Let's take a look at what's new!
My name is Abhijit Gupta, and I'm a student at Yale and former member of FRC Team 5511 Cortechs Robotics. Cortechs competed at IRI in 2019, where Zebra Motionworks robot tracking was first introduced. I became interested in FRC data analysis and eventually built Statbotics to analyze historical FRC data. In 2022, I modified Statbotics to support ongoing events. Thousands of users engaged with the site, but the slow data refresh reduced the total impact. This year, I have rebuilt everything from the ground up, with an increased focus on performance and scalability, and a broader vision of creating a platform for both realtime and historical FRC data analysis. I'm excited to share the new Statbotics platform with the community!
Statbotics V2 is a complete rewrite of the original Statbotics platform. This includes changes to the rating model, backend infrastructure, website, and Python API.
The new Expected Points Added (EPA) model is a significant improvement over the previous Elo and OPR based rating models, providing increased prediction accuracy, interpretability, and accessibility. Read more about the EPA model here and read how it compares to existing approaches here.
The new backend infrastructure better interfaces with the TBA API and is designed to support realtime data analysis. During the competition season, match results will be updated every minute, compared to every couple hours last year. The website has been rewritten to reduce loading time and provide a stronger foundation upon which to continue building. Already, some new features have been added, such as bubble plots, match breakdowns, and a revamped team page.
The Python API has been rewritten to support the new EPA model and use the new backend infrastructure.
Given the complete overhaul, most of my effort was spent recreating the existing functionality. I expect to add new features and improve the platform over the coming months. I am very open to community feedback and feature requests. I plan to maintain this blog and use the EPA model to answer interesting questions. Thanks for reading, and I hope you enjoy the new Statbotics!