Sports Analytics Weekly by kubeia.io - 25/2023
- Week 52
- Week 51
- Week 49
- Week 48
- Week 47
- Week 46
- Week 45
- Week 44
- Week 43
- Week 42
- Week 41
- Week 40
- Week 39
- Week 38
- Week 37
- Week 36
- Week 35
- Week 34
- Week 33
- Week 32
- Week 31
- Week 30
- Week 29
- Week 28
- Week 27
- Week 26
- Week 25
- Week 24
- Week 23
- Week 22
- Week 21
- Week 20
- Week 19
- Week 18
- Week 16
- Week 14
- Week 13
- Week 12
- Week 10
- Week 9
- Week 6
- Week 5
- Week 4
- Week 3
- Week 1
📝 Sports Analytics
Does 'Packing it in' really work in short yardage situations?
statsbomb.com - Mike Bursik
Short yardage situations are among the most exciting plays in football. Defined for this analysis as any 3rd or 4th down play with less than 2 yards to go, these plays offer intrigue for a number of reasons.
Death of a Target Man: Exploring The Efficiency of Route One Football
statsbomb.com - Scott Johnson
This article will investigate whether there is merit to the casting aside of the target man, setting realistic expectations for those that do play Route One football, and an analysis of a team at the top level who do Route One well: Brentford.
👁️ Computer Vision
Fast Segment Anything
arxiv.org - Xu Zhao , Wenchao Ding , Yongqi An , Yinglong Du , Tao Yu , Min Li , Ming Tang , Jinqiao Wang
The recently proposed segment anything model (SAM) has made a significant influence in many computer vision tasks. It is becoming a foundation step for many high-level tasks, like image segmentation, image caption, and image editing. However, its huge computation costs prevent it from wider applications in industry scenarios. The computation mainly comes from the Transformer architecture at high-resolution inputs. In this paper, we propose a speed-up alternative method for this fundamental task with comparable performance. By reformulating the task as segments-generation and prompting, we find that a regular CNN detector with an instance segmentation branch can also accomplish this task well.
💰 Quantitative Finance
Machine Learning Trading Essentials (Part 1): Financial Data Structures
hudsonthames.org - Michael Meyer
Trading in financial markets can be a challenging and complex endeavour, with ever-changing conditions and numerous factors to consider. With markets becoming increasingly competitive all the time, it is a never ending struggle to stay ahead of the curve. Machine learning (ML) has made several advances in recent years, particularly by becoming more accessible. One might think then why not use ML models in markets to challenge more traditional ways of trading? Well the answer is, unfortunately, that it is not so simple.
🤖 Machine Learning
How RLHF actually works
interconnects.ai - Nathan Lambert
The question I still get the most is "Why does reinforcement learning from human feedback (RLHF) work?" Until last week, my answer was still "no one knows." We are starting to get some answers.
Time Series Forecasting with ARIMA: Predicting Future Trends and Patterns in Sequential Data using Python
medium.com - The AI Quant
This tutorial aims to provide a comprehensive guide to time series forecasting using the ARIMA model in Python. We will explore how to analyze sequential data, identify trends and patterns, and make accurate predictions for future values. By the end of this tutorial, you will have a solid understanding of ARIMA and be able to apply it to your own time series data.
What Is a Transformer Model?
nvidia.com - Rick Merritt
A transformer model is a neural network that learns context and thus meaning by tracking relationships in sequential data like the words in this sentence.
Transformer models apply an evolving set of mathematical techniques, called attention or self-attention, to detect subtle ways even distant data elements in a series influence and depend on each other.
First described in a 2017 paper from Google, transformers are among the newest and one of the most powerful classes of models invented to date. They’re driving a wave of advances in machine learning some have dubbed transformer AI.
The Little Book of Deep Learning
fleuret.org - François Fleuret
I am also finalizing "The Little Book of Deep Learning", a short introduction to deep learning for readers with a STEM background. It is distributed under the Creative Commons BY-NC-SA 4.0 International License, and in the month following its public release, it was downloaded more than 200'000 times.
🕰️ Blast From the Past
Models, Balancing and Unconvinced at 18-0
blogspot.com
l0l0 started the ball rolling with some thoughts on the bookmaker's process:
1)calculating "true" probabilities
2)knowing the habits of your clients, distort the offer in order to have an equilibrate state of the book at the end to match the "true" probs
3)make noise(publish 'betshares') when the book isn't close to equilibrium
How bookmakers track your every move & how to get around it (from an industry insider)
daily25.com - Matthew Trenhaile
This article will give a small glimpse into the lengths that bookmakers go to to profile every single customer and then weed out any that may one day make a profit. This is a major issue in our little sports-betting world, and an important story to get out there to the general public.
Do Naive xG Models underestimate Expected Goals for Top Teams?
thesignificantgame.com
My intuition for this phenomenon is that high shooting output is a proxy for the quality of a team and that naive xG models underestimate the xG values for top teams. Given that xG models are usually by construction unbiased over all teams this also means that they should overestimate xG for poor teams (one channel how quality could impact xG over/underperformance is the finishing quality of top players that tend to be employed by top teams).
In this post I am trying to find evidence for my intuition.
🎙️ Podcast
Commoditizing the Petaflop — with George Hotz of the tiny corp
latent.space
How tiny is taking on Nvidia, Google, and PyTorch, building in public with AMD, hot takes on ggml, Mojo, Elon, e/acc, and GPT-4, and why AI Girlfriend is next. Now on YouTube!