The pursuit of wisdom in any walk of life quickly reveals that what you think you know is not nearly enough to get you to where you want to go. As I'm starting out in my football scouting journey I challenged myself to reach out to those already working in various roles in football to answer a short list of questions. My goal wasn't to get answers but relevant perspectives on the game within the game.
Here is Ryan Beal:
Here is Ryan Beal:
How did your first opportunity in sports come about?
I’m lucky to be able to focus my PhD on the use of AI in team sports which came about through pursuing a sports project for my dissertation while studying for my undergraduate degree. I was then offered the chance to stay on a research in the area full-time. For my dissertation project I used machine learning and optimisation to form optimal teams for daily fantasy sports in the NFL. In terms of sports in general, I played football growing up and then played at a non-league level for a few years.
What attracted you to analytics? What’s more intriguing now names or numbers?
While studying computer science and coding I always found the data science models/projects the most interesting, which led me to go deeper into more machine learning and AI work. Probably numbers to me, I enjoy getting into big datasets and finding new insights. Although, one of the best things about sports is the human aspect and the numbers alone don’t always tell the whole story.
What is the first model/concept you "found"? Python or R? Why?
One of the first sports papers I read was “Competing with humans at fantasy football: team formation in large partially-observable domains” which is an AI for fantasy football paper written at my university and consistently gets into the top 5% of FPL teams (Matthews, Ramchurn and Chalkiadakis, 2012). I also read papers such as Dixon and Coles (1997) and more recently really enjoyed the paper “Actions speak louder than goals: Valuing player actions in soccer” (Decroos et al., 2019).
Python for me, I use a lot of machine/deep learning tools from sklearn and keras which I have implemented in Python and I also use some optimisation packages that are only available on Python. Occasionally, I will also use some Java for some optimisation problems.
What is the biggest misconception in sports? Why?
In terms of analytics, I think students and people starting up think it’s impossible to get good quality datasets for free. However, there are now some great online resources with really good data such as on fbref.com for some basic stats like xG and StatsBomb for event-based data. My first paper on teamwork was started by using the 2018 World Cup dataset that StatsBomb provide for free. The exception though is with tracking data - hopefully this will change as more companies start to use computer vision on broadcast footage to collect their tracking data.
If you could start over what skill would you build on first?
I think starting from scratch a grounding in maths and statistics is really helpful, it’s what got me into computer science and coding to start with. From there definitely learn how to code, any language will do to start with and then build from there. I learnt in a language called Delphi which I haven’t used since but it allowed me to learn many more languages by understanding the fundamentals behind it.
Do you see player development as more of an art or a science? Why?
A bit of both, there is plenty you can use science for to optimise and manage the development of a player but there’s more to it than just on the pitch especially with younger players. I know teams with very successful academies have a philosophy focused on building soft skills and education as well as just developing football skills.
What is your favourite sports moment? Why?
For me it was Southampton’s promotion back to the Premier League in 2012. All my family are big fans and growing up I watched them get relegated down to League 1 (with a 10-point deduction) and nearly go out of business. Then watching them rebuild and achieve back to back promotions was brilliant, especially watching players like Lallana, Lambert and Fonte improve at each step up the leagues.
What coach/player/team inspires you? Why?
I was, as I’m sure many others are, inspired by Billy Beane and the book Moneyball. I think in general a lot of American sports are more advanced in adopting newer techniques such a data analytics and AI. I’ve seen some amazing work in basketball and ice-hockey, it will take a few more years for more football clubs to buy into similar techniques. My favourite player has always been Matt Le Tissier who is a legend in Southampton and could do things that no other player could!
What advice would you give to someone wanting to get into analytics?
Think differently and do new things that people haven’t thought of before and then find a way to get your ideas out there. I try to publish a lot of the work that I do on social media so that people can read about it and so that the techniques can be used for other applications in the wider AI community.
What is your favourite app/tool to use (for work or fun)?
For work I use scikit-learn a lot, it’s really useful for modelling and testing new machine learning techniques in Python. For fun I like to play Football Manager, I’m sure I will be playing lots of this in the coming weeks/months without any sport!
What other sport/hobby/discipline do you feel improves your work as an analyst/scout? Why?
I think working at a university helps with this as I work in an AI research group (AIC Group at the University of Southampton) with lots of people aiming to solve different problems in AI and multi-agent systems. As a group we focus on looking at the way in which autonomous entities (be they biological organisms, humans, software agents or robots) interact with one another in order to achieve individual and collective goals. This means that there is lots of interesting ideas and collaboration that can help apply new ideas to football and sports.
Here is a link to my google scholar page where my papers are available if people are interested: https://scholar.
No comments:
Post a Comment