Lessons Learned: Attending a Football Analytics Conference as a Newbie

TL;DR:

I attended the online 2020 Sounders FC Soccer Analytics Conference and it was amazing. Since this was the first conference I attended in the football analytics space, I summarised my learnings in this post.

Short summary;

  • The beautiful game will probably always stay true to the basics; judgement based on the professional’s own eyes. Data plays a role in optimising the process of generating objective insights, complimenting the subjective intuition of football professionals.
  • There is a gap in understanding about data between the data and coach “brains” at football clubs. Blending intuition and data capabilities and marrying both worlds is critical for the success of football analytics.
  • There are certain limitations to the technology and availability of data, which limits the expansion of the industry — but there is huge progress every single day and many innovations are expected in the near future

Grab a cup of coffee (or tea if you like) and read more if interested in more details.

A bit of background for the curious reader

In the past few years, football analytics became my #1 passion - one that grew into further than just a simple hobby. During these unprecedented times of staying at home, I had a chance to spend more time on personal interests and dive even deeper into the fantastic work being shared by the community and surprisingly overwhelmed by the support given by the leading members of it. Being a product manager with virtually no real value add to the industry at this stage, I was thinking of ways to give back to the community. Having seen that Ravi Ramineni (@analyseFooty) from Seattle Sounders FC was making one of the top events in football analytics space more accessible to everyone by making it online, reducing the cost from $500 to $50 and more importantly committing to donating all proceeds to a COVID-19 Relief Fund, I thought why not share the highlights of the amazing “2020 Sounders FC Soccer Analytics Conference” with the rest of the community. After all, documenting stuff is what I do all the time, right? :)

A bit of context for people who are alien to MLS football like me, Seattle Sounders FC are the latest champions of MLS football in the US and are known by their evidence driven approach to soccer (let me use this one last time). They won the most recent final against Toronto FC, which are another team known for their successful usage of analytics with Devin Pleuler (@devinpleuler) leading their analytics department. A message to the ignorant reader: this final is widely classified as the cornerstone of the impact analytics is having on the success of football clubs (setting Liverpool’s great story aside).

Sounders FC Soccer Analytics Conference is organised yearly and focused on discussing “how to integrate analytics into a decision-making workflow at a soccer club”. This year the conference took place online, which enabled me to attend my first ever sports analytics conference, featuring a wide variety of guest speakers and panelists.

My experience attending the conference

First things first: Having been to different product/tech conferences over the years, I can easily say that this conference was by far the best value for money. Had it been $500 + return tickets to US + all other expenses, it’d still be the case. :) It could be because I’m a novice in the field, but I found the content shared in the conference mostly eye-opening and I learned from every single speaker.

Now, let’s do a bit of retro with the spirit of a product manager.

What went well?

  • Variety of speakers: There was a very good balance of professionals from all the fields related to football analytics, ranging from football club staff to data visualisation experts. This gave me a chance to learn more about the dynamics of the industry, the current status and the future of analytics in football. This was gold, especially for someone like me, who is an outsider to the football world.
  • Chance to engage with speakers: Ravi made sure to spare some time for questions for every single session, and the speakers were super kind to go through all the questions even if they were not able to answer them during the talk. This made it possible for me to ask any questions and get answers from the experts in the industry. Even only this is well worth far more than 50 bucks.
  • Online setup: Well, realistically I could’ve never made it to this conference in Seattle had it been not organised online. There are a bunch of negativities surrounding COVID-19 crisis, but the possibility of attending overseas conferences online is maybe the one thing that I would hate to see return back to “normal”.

What could’ve been better?

  • Short / no breaks: During all 3 days, I barely had a chance to get up and get some water, as the schedule was pretty intense with not enough breaks in between talks. As Ravi mentioned last day, the break between the announcements for the competition and Bruno’s talk was the only proper break we had. Intense stuff. It’d have been better to have some time between talks for refreshments.
  • Buggy Q&A functionality of Zoom: Due to the tight schedule most Q&A happened through the chat. Zoom, with all its popularity, was used as the video conferencing solution and the conference showed the flaws of Q&A functionality of Zoom and how buggy / primitive it is. A lot of great questions were asked and it was really hard to keep up with the amazing answers given. It’d be great to have a compilation of Q&A inputs, if available, for future reference. (@Ravi this one is for you, if you read this).
  • Details of the schedule & themes: Although the timing of talks and the speakers were announced far in advance, the content of their talks were not. It’d have been nice to see these in advance as well, to be able to plan the day, or night in my case as I watched the conference from Amsterdam. The content each day could also have been grouped in themes, to make following it easier based on specific interests of attendees. (I made an attempt to summarise the general vibe of the talks though, stay tuned.)

Highlights from the conference

The content of speakers were super rich and I got some interesting insights from every single talk. When I check my very unstructured notes though, I can see some highlights from my learnings and I tried to condense these for each day of the conference, here you go.

Day 1: Various applications of data & analytics in football

First day of the conference felt like an intro to the different applications of data & analytics in football. Having club staff, analytics departments and media on the same day, talking about a wide range of topics created that feeling of kickstarting the conference. Let’s take a look at what has been shared the first day.

Gregg Berhalter, Head Coach at US Men’s National Team — Using data to support the decision making process
In his talk, Gregg shared the journey they’re having at the national team with regards to the data collected about the USMNT players before, during and after an international game. Having a sophisticated and thorough process in place to collect and analyse their player’s data, USA seemed to set themselves up for success for many years to come. One eye opener from Gregg’s talk was the emphasis he put on the communication and integration between coaching staff and analytics departments. He touched upon the importance of having these two sitting closer to avoid passing insights through a filter. This dichotomy of traditional vs. analytical “brains” was mentioned by quite a lot of speakers during the conference, definitely an important takeaway.

Ravi Ramineni (Director of Soccer Analytics) & Gonzalo Pineda (Assistant Coach), Seattle Sounders FC— Analytics in Workflow
Hosts of the conference followed up with putting further emphasis on how they work together every single day. They demonstrated a well functioning workflow between their coaching, video and data analytics departments having a week of cadence to generate insights before the games and gave specific examples of how this relationship works on a day to day basis. Seeing the both sides of the coin was very insightful, making the analytics-in-action visible to us, attendees.

Tom Worville, Football Analytics Writer at The Athletic — Analytics, a bite at a time
I’m a huge fan of Tom (@Worville) and an avid reader of his analytical, visual heavy* pieces on The Athletic ever since he moved there (what a match!). During his talk, he gave insights into his interpretation of data available to him and how he translates this to compelling stories using data visualisation, with the aim of making his vizardry accessible to the non-data-native reader. If you’ve never read one of his stories, you’re missing out a lot, check my favourite piece here (Disclaimer: I don’t earn anything from this link. You’d need a subscription on The Athletic, which is 90 days free right now)

* If you’re curious what Tom uses for his vizs, here is his answer; “All my stuff is built out in R. Use ggplot2 for graphics, Shiny for apps, GT for tables”

Marcel Daum, Assistant Coach - Analysis at Bayer Leverkusen
Having been pleasantly surprised with his surname and curious to check the connection, I learned that Marcel has worked in Turkey before at Christoph Daum’s Fenerbahce. His talk gave insights into how they use data at Bayer04 (null vier it is) to measure and explain the highly intensive & innovative tactical style of Peter Bosz, and gave examples of how they act on these insights even during the course of the game. Marcel’s talk was really eye opening for me as a product person, as he shared the suite of tools they use at the club — pretty sophisticated, right?

I was already pretty tired by the time the roundtable discussion started and needed to skip it for day 1. I still feel bitter for missing it, but I hope the videos will be published some time and I can catch up.

Day 2: Data & analytics in talent identification

It was hard for me to come up with a theme for the day especially because of the extraterrestrial work of Karun Singh (details below), but I think the dominance of talks around the usage of data & analytics in talent identification and player performance won the race. Having a specific interest in youth talent identification, I binge-watched the sessions almost open-mouthed. I’ll count myself successful if I’m able to touch upon 5% of what was shared, sorry all.

Joe Mulberry, Director of Recruitment at Nordsjaelland & Right to Dream

I prefer giving direct reference to Joe and his team‘s work with Right to Dream as I don’t want their meaningful work lost in translation. One part of Joe’s talk that stood out for me was how he decided to become data native and started learning coding at a relatively later age after reading Daniel Kahneman’s “Thinking, Fast and Slow” — I can totally empathise. During his talk, he mentioned that the professionals in the analytics field should respect intuition, learn to question and be vulnerable to be able to achieve a successful transition to using analytics at their clubs. I had the privilege to get an answer to my question about how they collect and interpret data for youth football and he was kind enough to answer transparently; saying that it’s a real challenge to collect meaningful data, if any, and they use proxies like played matches and debut dates for different age groups, as well as creating systems for young players to codify themselves and their days. Thanks again, Joe!

Sudarshan Gopaladesikan, Head of Sports Data Science at Benfica — Data Analytics for Youth Football

As if he heard my question to Joe, Sudarshan, aka Suds from #FoT* (@suds_g), thoroughly explained Benfica’s methodology about using data analytics for youth football. As most football fans would know, Benfica have been producing ridiculous amounts of young talent for at least a decade; Bernardo Silva, Ederson, Joao Felix to name a few. Stuff that Suds talked about were of course of highest quality and every slide taught me a new thing, but I can’t imagine how valuable these must be for football clubs and professionals. It’s virtually impossible to summarise his talk due to the vast amount of learnings spread across his slides, but for me the most striking point was his prioritisation to support the pressing needs of the coaching staff at the club. The need for football analytics people inside the clubs to produce results for their main stakeholders, the coaching staff, was later discussed during the panel together with Ted Knutson’s (@mixedknuts) appearance, and their situation is contrasted to that of the people working at analytics tools. Enlightening, really.

*If you don’t know what this is and have interest in football analytics, I urge you to check Friends of Tracking Youtube channel and #FoT hashtag on Twitter to see how far the support of community has gone— In addition to Suds, I want to take this chance to thank all the leaders of this initiative here, as they are the ones behind my increased understanding of the space and ultimately the reason for this Medium article. Thank you David Sumpter (@Soccermatics), Javier Fernández (@JaviOnData) and Laurie Shaw (@EightyFivePoint), and all the week in-out contributors.

Karun Singh — Analysts and Algorithms: Human — Machine Collaborations to Aid Match Analysis

One of the leading inventors of the football analytics space, Karun (@karun1710) gave one of the most interesting, and certainly the most tech-heavy, talks of the conference. The man behind Expected Threat (xT) started off with pointing out to the dichotomy mentioned earlier between analytics departments (i.e. Algorithms) and the traditional staff (i.e. Analysts). From this perspective, he laid out his thought framework of deciding on solving macro vs. micro level problems based on the push vs. pull mechanism happening between builders of “Algorithms” and “Analysts” who make use of these. Following this logic, he focused his efforts on finding anything in the existing and up-and-coming data set that he can make the jobs of video analysts better and created two new products based on tracking data; “Situational Search” and “Auto-tagging”.

Let’s try and imagine a user scenario to make things a bit more concrete: You are a video analyst at Manchester United and you are analysing Manchester City’s attack patterns ahead of your derby day, you notice Kevin de Bruyne picks up the ball in the right half space and makes a dangerous cross inside the box, instead of watching the rest of the game to search for similar patterns of play, the “Situational Search” simply shows you other moments this had happened throughout the game. Not only this though, you can also label this move, and a few other examples from different games if you like, as a “right half space cross” and the “Auto-tagging” functionality would tag all the other game situations at scale and show you how similar a pattern of play to a “right half space cross”. Can you imagine the efficiency of gaining real insights for opposition analysis? Pretty impressive, I’m looking forward to seeing this working and commercialised.

Javi Garcia, GK Coach ex-Arsenal, PSG, Sevilla — Using Analytics for Goalkeepers

Well, I was a goalie, a pretty decent one at a young age if you ask my father. I remember the first English books I read were goalkeeper training books my mom would bring from the UK and I think I might have learned English while translating the exercises and practicing them. Hence, I was very excited to see such a top level GK coach in the conference speakers line-up. Javi Garcia gave great examples of how he uses data before, during and after a game to train and measure the progress of the goalkeepers in his teams. He gave pretty specific examples about how the data provided by the analytics teams help him instruct his goalkeepers about their decisions and prepare them for the upcoming game (e.g. facing penalties, ball distribution from the back). Practical as it goes, I had reminded of my beautiful childhood memories as a GK.

Roundtable Discussion: Event vs. tracking data and what’s next?

As I mentioned earlier, the roundtable discussion of the 2nd day was joined by the CEO of Statsbomb, Ted Knutson, which is the company that has been providing event data freely to the community for quite some time (You can even find data about World Cup 2018 there). Not so surprisingly after Karun’s innovative work with tracking data and the analytics-heavy panelists with Suds, Ted and Joe, there was some really good discussion about the differences between event vs. tracking data and the pros & cons of each to explain football. One clear expectation from the panelists was that there would be more and better data available to explain football moving forward and they agreed that the context would be the main driver defining the innovation in the field. The absence of context about the event data seems to be the biggest disadvantage right now, even if it’s available across leagues and although tracking data is trying to address this issue, its collection is pretty limited due to needed infrastructural investment, and its applications are nascent, Karun’s “Situational Search” and “Auto-Tagging” being one more promising & sophisticated example. As a result, I posed my catch-all question; “If you had a magic wand, what would be the data that you’d like to add to your existing suite of data points?”. Leaving the answers below to give you some indication of where we are heading to;

  • Tracking proxies to fundamental football intelligence e.g. body moves, scanning over the shoulder
  • Contextual event data, explaining the situation of how a particular event has happened
  • For youth, measuring their growth, motivation and perception is crucial. I think Gary Lewis, Academy Director at Seattle Sounders FC, mentioned it’d have been great to see what the players were thinking just before taking an action and how their decision making process was — that’s what I call a request if you had a magic wand! :)
  • Cognitive understanding about how players see the situation in relation to team members, having an understanding on the group dynamics, team setting
  • Training data of players
  • VR situations together with event and tracking data — a VR lens is maybe not that of a SciFi anymore?

Day 3: The Red Carpet & Curtain Call

Closing day of the conference felt like a celebrity panel with the likes of death charts guy John Burn-Murdoch (@jburnmurdoch), “Mr.Moneyball” Billy Beane and the co-author of “The Numbers Game” Chris Anderson (@soccerquant) on the speaker line-up. This was for sure a once-in-a-lifetime experience, not only due to the opportunity to hear from the experts & creators of the field, but also witnessing the potential next frontier to scale the collection of football analytics data; the non-corrected data extraction from training videos by Metrica Sports that is announced with an open call from the CTO of the company, Bruno Dagnino (@brunodagnino).

John Burn-Murdoch, Senior Data Visualisation Journalist at Financial Times — Data Visualisation as Communication

Most of you, if not all, would know this chart below. John is the creator of it.

For me the main takeaway from John’s talk was the importance of knowing your audience when using data visualisation and telling a story that would be understood by them, in his words; “People are taking a message from the graph, not numbers”. He repeatedly mentioned about thinking about “non-chart” people and the beauty of being able to tell a sophisticated analysis with simplicity. Having shared examples of how this perspective is translated on graphs using elements like the design of the headers, usage of text and annotations, John made me realise that most of the graphs that I can recall are indeed following the same pattern, which is a great learning for even outside of the football context for my work.

I’m pretty sure some readers would be interested in what tools John uses, and of course he’s been asked this question. Although he referred to Tableau and R (ggplot) as tools he uses, his answer was aligned with the rest of his talk; “Start with using what enables you to translate a heavy analysis into a simple message”

Q&A with Billy Beane, Executive VP of Baseball Operations at Oakland A’s

If you haven’t read Moneyball or watched the movie with the same title starring Brad Pitt, you are probably lacking proper context to understand the triumph of sports analytics. Although Billy Beane himself said that he probably was just a person who used data with success and not the first one by any means, his work at Oakland A’s is a historical turning point for usage of data and statistical methods in sports. During this Q&A session, Beane gave insights into his perspective on how the data revolution in player recruitment all started and what he expects in the coming years. One particular highlight was his reference to recruitment being a combination of art and science using the example of Liverpool’s Coutinho sale. Statistics might tell you a story, he elaborated, but the decision of replacing a player of his calibre vs. not directly replacing him and thinking about the holistic pieces of a team is where the art part kicks in. With his vast experience, he touched upon a lot of other topics as well, but this particular one gave a hint about the real life decisions made by recruitment teams.

After the Q&A of Billy Beane, results of the Data Analytics Competition were announced. I was amazed by the vast amount of sophisticated analysis done by people from outside the football industry, which is a sign of very rapid development in the short to mid-term. I’ll just leave a reference to Ravi’s tweet below for the interested reader.

Bruno Dagnino, CTO at Metrica Sports — Bridging the gap, but which one?

Closing talk of the conference was made by Bruno (@brunodagnino) who leads the tech at Metrica Sports, a fellow Amsterdam-based company building a video analysis platform for football professionals. Metrica gained additional popularity in the community by making 2 full games of tracking and event data freely available, enabling analysis with tracking data curriculum of Laurie Shaw for #FoT— something that can be a turning point for the usage of tracking data to create value for the industry.

Bruno started by differentiating between the “horizontal” and “vertical” gaps we have today in the football analytics industry. The “horizontal gap” is between the data and coaching brains, which was previously mentioned multiple times by other speakers as well. However, Bruno made the first clear reference to the “vertical gap”, which is the fact that only a handful of clubs have access to data and make use of analytics today.

As Metrica Sports, they commit to bridging this vertical gap in the near future and Bruno announced a very exciting new technology the company has built: getting non-corrected tracking and event data using the training camera of any team — a 10x more affordable way of collecting data compared to the existing methods. He also made an open call to clubs, federations and likes, who hold the right to publish data openly, to participate in “Project Open Data” and create a proof of concept of their technology in the coming months. Exciting times ahead!

Roundtable Discussion: The Football Analytics Ecosystem

Roundtable session of the final day was moderated by Rory Smith (@RorySmith), who posed some very interesting questions about the ecosystem surrounding football analytics: likes of fans, agents and media. Fans acceptance of data is widely regarded as a must-have for further growth of football analytics, and the example of the usage of xG at MoTD in the UK last year was referred to as an important milestone — albeit a confusing one. Although media coverage about analytics is increasing, especially with the success stories of Liverpool, there is still a long way before the mass adoption of the topic. Agents are seen as another up and coming user of data analytics, especially since more and more clubs are asking questions to them about their players’ data. Chris Anderson made a nice point that if agents are interested in a topic, it’s a good sign that it’ll become more prominent in the coming years.

Who knows, maybe the acceptance of fans, maybe the media giants or maybe gambling will simply fuel the growth of football analytics. One thing for certain, the future of football definitely looks more data & analytics driven.

Closing Thoughts

The conference gave me amazing insights about how football analytics is used at clubs and helped me understand where the industry is and where it is heading to. Once again, a huge thanks to Ravi Ramineni and Seattle Sounders FC for making the conference accessible online.

I’m looking forward to seeing more innovations in the coming years, and excited to keep being in this wonderful community.

If you liked the content of this article, please clap and share so that others can benefit from it as well!

Follow me on twitter @ulkerca

Product, technology, start-ups, music, data, football and life.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store