Greg Everett

@yureedelahi Thanks very much!!

@greg_everett Thanks Greg! 1) Min 5 actions in a 30s window before the shot. Set pieces excluded. Tested 10s-60s, 30s seemed most stable. 2) Features like width, switches, direction changes, tempo, central %, final third %, spatial acceleration, recycles, start depth, progressive pass ratio.

Trying to see what happens in the 30 seconds before a shot? I clustered every PL build-up into 6 patterns. An example of how the model reads a single goal traced action by action. A work in progress (as always😅)

Euro 2024 kicks off this evening, so we set out simulation engine the task of predicting it. Using information about squads, tactics, and style to simulate the tournament 1000s of times, it thinks 🏴󠁧󠁢󠁥󠁮󠁧󠁿 England, 🇫🇷 France and 🇩🇪 Germany have the best chance of success. #euro2024

Excited to announce that we will be organizing the 11th Edition of Workshop on Machine Learning and Data Mining for Sports Analytics @ECMLPKDD in September!  w/@JanVanHaaren @azimmerm_dm and Ulf dtai.cs.kuleuven.be/events/MLSA24/

The 11th Workshop on Machine Learning and Data Mining for Sports Analytics (MLSA) will take place in September in Vilnius, Lithuania. Our workshop welcomes sports analytics research papers involving machine learning and data mining. @ECMLPKDD (1/2)

It'd be great to hear your thoughts and feedback. Me and the team are looking forward to presenting this work next month!

Excited to announce that my research paper, "The Strain of Success: A Predictive Model for Injury Risk Mitigation and Team Success in Soccer", in collaboration with Sentient Sports, has been accepted at @SloanSportsConf! The full paper is available here: sloansportsconference.com/research-paper…

Have you seen the 2-minute video abstract for our upcoming #KDD2023 paper? 👉 youtube.com/watch?v=hV3whg… We're in the running for the audience appreciation award, and you can make a difference! Show your support by giving it a 👍 and let's aim for the highest view count! 🏆😁

I just wanted to send this farewell message to all Manchester United supporters. I would like to express my unwavering gratitude and appreciation for the love from the last 12 years. We’ve achieved a lot since my dear Sir Alex Ferguson brought me to this club. I took incredible pride everytime I pulled on this shirt, to lead the team, to represent this institution, the biggest club in the world was an honour only bestows upon a few lucky footballers. It’s been an unforgettable and successful period since I came here. I didn’t think from leaving Madrid as a young boy we would achieve what we did together. Now, it’s the right time to undertake a new challenge, to push myself again in new surroundings. Manchester will always be in my heart, Manchester has shaped me and will never leave me. We’ve seen it all. 🤘🏼❤️

Thrilled to announce we've won a grant from the UK's innovation agency @InnovateUK to fund research into trustworthy generative AI systems for sport!

@Anuraag027 Thank you very much! 😁

@greg_everett @devinpleuler This is super interesting! Will definitely give this a read!

In soccer, has there been any work that takes coordinates from event data and predicts true player-level average positions as defined by tracking data? Seems like this would be pretty useful, and not super hard.

@devinpleuler Thank you very much! Would be interested to hear your thoughts!

@greg_everett Very cool. I'm looking forward to digging into this.

@hugoriosneto @MaaikeVanRoy @wagnermeirajr @jessejdavis1 Thank you Hugo!

@greg_everett @MaaikeVanRoy @wagnermeirajr @jessejdavis1 Thanks, Greg! Congrats to you as well on your paper at AAMAS, very interesting! 👏

Very happy to have been selected for presentation in the Pro Forum's Algorithm Track. The project was done in collaboration w/ @MaaikeVanRoy, @wagnermeirajr, and @jessejdavis1. Looking forward to going to London for the event next week!

This work was led by myself, and co-authored by Ryan Beal and Tim Matthews (@sentient_sports), Joseph Early (@JosephAEarly), Tim Norman (@proftjn) and Gopal Ramchurn (@gramchurn).

This increases accessibility of estimated off-ball downstream analytics tasks (e.g., physical player metrics, pitch control calculations, player heatmaps) to researchers and clubs who are only able to access event data.

Excited to announce our paper "Inferring Player Location in Sports Matches: Multi-Agent Spatial Imputation from Limited Observations" has been accepted for publication in the Main Track at #AAMAS2023. The pre-print paper is available here ⬇️ arxiv.org/abs/2302.06569

Fascinating! Informed decision making in sport using both human knowledge and the power of AI models seems an exciting way forward and this feels like a new AI staff member that can collate and present insightful reports quickly and have discussions regarding these results.