Johan

Coaches need to embrace AI driven analytics to acquire valuable insights on how to improve performance

[Johan @soccerlogic]

@ReadingFC just hired football’s first ‘Head of AI,’ but most clubs are about to fall into a multi-million dollar infrastructure trap trying to copy them. Here is the lean, on-pitch blueprint clubs actually need before next season: https://t.co/XsCrOOvZx9

[Johan @soccerlogic]

Football clubs focus on data/AI infrastructure because it is concrete and familiar. But the real challenge isn't assembling hardware, pipelines, and feeds. It is working out with coaches on how to use AI creatively to provide insights to win games.
https://t.co/YQTL6Dc8lG

[Johan @soccerlogic]

Why England’s Data Revolution Failed Its Biggest Test https://t.co/KXmWJ1CgUW

[Johan @soccerlogic]

Wonder if Frank had the same analytics support @Spurs_THA he enjoyed @BrentfordFC . Probably not!

[Andrej Karpathy @karpathy]

Love this project: nanoGPT -> recursive self-improvement benchmark. Good old nanoGPT keeps on giving and surprising :)

- First I wrote it as a small little repo to teach people the basics of training GPTs.
- Then it became a target and baseline for my port to direct C/CUDA re-implementation in llm.c.
- Then that was modded (by @kellerjordan0 et al.) into a (small-scale) LLM research harness. People iteratively optimized the training so that e.g. reproducing GPT-2 (124M) performance takes not 45 min (original) but now only 3 min!
- Now the idea is to use this process of optimizing the code as a benchmark for LLM coding agents. If humans can speed up LLM training from 45 to 3 minutes, how well do LLM Agents do, under different kinds of settings (e.g. with or without hints etc.)? (spoiler: in this paper, as a baseline and right now not that well, even with strong hints).

The idea of recursive self-improvement has of course been around for a long time. My usual rant on it is that it's not going to be this thing that didn't exist and then suddenly exists. Recursive self-improvement has already begun a long time ago and is under-way today in a smooth, incremental way. First, even basic software tools (e.g. coding IDEs) fall into the category because they speed up programmers in building the N+1 version. Any of our existing software infrastructure that speeds up development (google search, git, ...) qualifies. And then if you insist on AI as a special and distinct, most programmers now already routinely use LLM code completion or code diffs in their own programming workflows, collaborating in increasingly larger chunks of functionality and experimentation. This amount of collaboration will continue to grow.

It's worth also pointing out that nanoGPT is a super simple, tiny educational codebase (~750 lines of code) and for only the pretraining stage of building LLMs. Production-grade code bases are *significantly* (100-1000X?) bigger and more complex. But for the current level of AI capability, it is imo an excellent, interesting, tractable benchmark that I look forward to following.

[Johan @soccerlogic]

Chelsea's issues aren't just the players, but also "one-trick-pony" coaches who can't adapt and try different tactics. This requires thinking outside the box: a rare skill among coaches who've spent their lives primarily on the pitch. #ChelseaFC

[Johan @soccerlogic]

Coach or AI? Who would you trust to make the best decisions? 🤖⚽️
#AI, #Football, #Soccer, #Soccerlogic

[Johan @soccerlogic]

Join me at @BigData_LDN 2024, register free here! https://t.co/Ix3I4V5mV5 #BigDataLDN

[Johan @soccerlogic]

It's rare to see a coach publicly credit his analytics team as Klopp as done after the Fulham match. But while their observation in this case may be considered common sense, it highlights the potential for analytics to inform tactical decisions on the pitch

[Johan @soccerlogic]

@JSportsSci @WeAreTandF you need to do more than encouraging if you want to be considered a scientific publication

[Johan @soccerlogic]

@JSportsSci what is you policy about sharing the data of your articles?

[Johan @soccerlogic]

New @SpursOfficial and @ChelseaFC managers face high expectations. But will they use analytics to succeed? Old-school managers often rely on experience, but @soccerlogic AI-driven analysis can help them make better decisions. #analytics #football

[Johan @soccerlogic]

Hey @ChelseaFC, looking for help to win trophies again? Look no further than SoccerLogic! We have over 20 years experience in finding valuable insights in match data, and we're confident that we can help you take your team to the next level.

[Johan @soccerlogic]

@soccerlogic, @SRMcIntosh , @Robertson_SJ Science journals should not publish papers by researchers who refuse to share data. Data sharing plays a pivotal role in scientific research by fostering transparency, reproducibility, collaboration, and knowledge advancement

[Johan @soccerlogic]

Managers should move away from blaming 'lack of investment' for failures and embrace football analytics. By leveraging AI data-driven insights, particularly in tactics, they can enhance the performance of their existing squads. @SpursOfficial @ManUtd
https://t.co/Nq47762rWs

[Johan @soccerlogic]

https://t.co/87FOQElrx9

[Johan @soccerlogic]

Managers should not be sacked solely on results, analytics (AI driven analysis) should play a major part in judging their performance. But most clubs are run by people who barely understand stats, never mind analytics. @LUFC , @SpursOfficial

[Johan @soccerlogic]

@SRMcIntosh @MathieuFeigean @Robertson_SJ Hi Sam, it would be nice to have the data too. Can you pls share it. Thanks!

[Johan @soccerlogic]

AI for football is here, but coaches do not want it...
https://t.co/fdrNKBJ9U8

[Lord Jason Stockwood @jstockwood]

New @guardian article. Thoughts on #chatgpt and #ai in sport. #gtfc

[Johan @soccerlogic]

Football coaches searching for new players to improve performance and win games are like generals who can only think of winning battles and wars by getting more and better arms. Instead, they should deploy analytics to achieve their aims #football #analytics #performance