Sign in

Derek Grifka

@derekgrifka.bsky.social
55 followers 84 following 8 posts

Data Scientist. Marquette basketball enjoyer. dgrifka.github.io Creator of bsky.app/profile/mlb-simulator.bsky…

PostsRepliesMedia
Reposted by Derek Grifka
MLB Deserve-To-Win-O-Meter @mlb-simulator.bsky.social · 18/01/2026
Hi! Launched a Streamlit app for the Deserve-to-Win-O-Meter. You can now easily search any game from the 2025 season (soon 2026). Note I added spray angle to the model, which accounts for where balls are hit on the field, not just how hard. dtw-str.streamlit.app
dtw-str.streamlit.app
111
Reposted by Derek Grifka
MLB Deserve-To-Win-O-Meter @mlb-simulator.bsky.social · 04/06/2025
Introducing a daily playoff probability model and a "rooting interest guide" (inspired by @PFF_Moo from Twitter).   Playoff %s from 10K sims w/remaining SOS & home/away advantages. Rooting Interest analyzes who each team's fans should root for today, based on playoff implications
112
Derek Grifka @derekgrifka.bsky.social · 24/03/2025
I'm excited to announce I've accepted a role as Data Scientist for the Chicago Fire. Thanks to everyone who has given me feedback and advice. I'm ecstatic to be able to learn from/work with some very bright individuals and contribute to the success of a storied MLS franchise!
010
Derek Grifka @derekgrifka.bsky.social · 04/02/2025
I'm a Houston Texans fan and have never seen my team win the Super Bowl, which made me think: If you were born today, then how likely is it that you would be alive to see your favorite team win the Super Bowl? You have a 50% chance of seeing your team win 1 by age 22. #NFL #SBLIX
110
Derek Grifka @derekgrifka.bsky.social · 22/01/2025
Skeptical of Notre Dame's "middle 8" advantage, I thought good teams dominate throughout. But data shows ND scores 3-18 more points (89% confidence) in these minutes vs other periods. Whether or not this matters, they do in fact score more in the middle 8. www.wsj.com/sports/footb...
010
Derek Grifka @derekgrifka.bsky.social · 21/01/2025
When McLellan took over, the Red Wings were .433. Since then? 8-4. Using a Beta-Binomial model with a .500 prior, there's an 81% chance they're now truly an above-.500 team. Best estimate of their true win rate: .591 (90% CI: .418-.755)
010
Derek Grifka @derekgrifka.bsky.social · 31/12/2024
Hockey is cool
000
Derek Grifka @derekgrifka.bsky.social · 18/11/2024
I developed a neural net for this year's NFL's Big Data Bowl. Have a lot of ppl to thank, but wanted to share bc I enjoyed working on it and hope you enjoy reading it www.kaggle.com/code/derekgr... github.com/dgrifka/nfl_...
kaggle.com
Overlap: How can pre-snap motion exploit it?
Explore and run machine learning code with Kaggle Notebooks | Using data from No attached data sources
011