Alter Everything

MaddieJ · ‎04-19-2023

For football clubs, data collection and analytics opportunities are growing faster than you can say “GOOOOAAALL!” But what are the metrics that matter? How are football clubs advancing their analytics maturity? With support from Alteryx and Continuum, Richard Battle of Left Field Football Consulting will share insights from his recent report “Analytics in Football: The State of Play” and will share tips for how to break into the sports analytics game.

Panelists

Richard Battle - @BATTLE, LinkedIn
Megan Dibble - @MeganDibble, LinkedIn, Twitter

Topics

Youtube Thumbnail.png

Transcript

Episode Transcription

MEGAN: 00:04	[music] Imagine you're playing your favorite sport, and location data is captured at 25 frames per second. Now imagine that you can use that data along with technology like machine learning to improve your performance. When you think about it, that's actually a lot of data capture, and the analytics opportunities are vast. So what are the metrics that really matter? And how can analytics help sports teams succeed? I'm Megan Dibble, data journalist for the Alteryx Community, and today we're talking with Richard Battle from Left Field Football Consulting based out of Liverpool in Northwest England. Richard is a consultant in football working across strategy and analytics.
RICHARD: 00:44	When I say football, perhaps I should clarify for our audience, I'm talking about soccer as some might refer to it, but I've worked in football for 10, 15 years now with clubs, leagues, federations, and investors. So very happy to come and talk to you guys about some of that today.
MEGAN: 01:00	With support from Alteryx and Alteryx partner Continuum, Richard helped drive Analytics in Football: The State of Play, which is a deep dive report into the building blocks of analytic strategy at football clubs. Fun fact, I also helped a bit with this report. In this episode, you'll hear how football clubs think about data literacy, analytics maturity, AI, and spatial analysis. And as a bonus, if you love when the analytics world collides with sports, be sure to check out Alteryx Fanalytics for riveting, in-depth looks at the world's most popular sports. Links to Richard's report and Alteryx Fanalytics will be in our show notes at community.alteryx.com/podcast. Let's get started. So I'm curious what your role as a football consultant looks like. What kinds of things do you advise on?
RICHARD: 01:54	So there's quite a wide variety. So I suppose my background going back a long time is in finance and accountancy. Much more recently, I've worked in football on topics such as player development, squad development, and high performance. So my work can really be across any of those strands. And my focus is very much on what I'll call the football side of football. So over recent years, the last couple of decades, football has become an industry, as well as a game. That industry creates lots of work in terms of marketing, in terms of fun engagement, all those things that go around operating a professional sports club or legal franchise. But my focus within football is very much on the game itself and on the disciplines which support player development and team performance.
MEGAN: 02:49	Okay. Awesome. Yeah. And I mentioned the analytics report. So you work with football data, and I'm curious to know what kinds of data do football clubs have and use.
RICHARD: 03:03	Good question. The report focuses on different kinds of data. So there are lots of data types which football clubs get given based on the league in which they play. And that can include event data, which tells you things like who kicked the ball, how they kicked it, where they were when they kicked it, and so on. You can also get what's called tracking data, which tracks the position of every player and the ball at a rate of kind of 25 or 50 times per second and gives you X, Y, and Z coordinates of the location around the pitch, which is kind of a more recent development in the football data industry and has the potential to give a much greater and deeper and richer insight into what it is that goes on the pitch than event data historically has done. So those are probably two of the main types of data that football clubs use, but equally, there are tons of other kinds. So it could be related to player contracts, player salaries, player injuries, player transfers. All different elements really, which help clubs and teams, and other organizations to better understand A, what goes on the pitch, and B what the market for talent in the industry looks like around the pitch.
MEGAN: 04:18	Gotcha. Yeah, the data you mentioned about player location on the field, what was the frequency?
RICHARD: 04:26	So most commonly is 25 frames per second.
MEGAN: 04:30	That's a lot of data. Do clubs ever have trouble analyzing that much data or visualizing it?
RICHARD: 04:37	So it is a lot of data. It's where you go from thousands of rows per game in event data to millions of rows per game. And yeah, as a result, there are fewer clubs, let's say, which are fully competent in processing and manipulating, and extracting value from that data. The industry is moving along its analytics journey, let's say, to a point where more clubs are becoming tracking literate, but event data is still the staple for the vast majority. I think it's a question not just to volumes of data, but also then of the skill sets needed to process and understand that data and the types of insights it gives you. I think event data is much more intuitive. You can create shot counts, you can create pass counts, you can create derivatives of those metrics. Whereas, tracking data is essentially a series of points which move in different ways through an area over time. And I guess in the world of Alteryx, people are very competent in talking about the world of points and lines and polygons, and football really is based on all those same principles, even if we use slightly different language to describe it.
MEGAN: 05:50	Yeah. You mentioned all this tracking data that football clubs have. So how do they end up analyzing that? What tools do they use? What tools have you used?
RICHARD: 06:02	So for me personally, I'm a big fan of using Alteryx for that data. I guess my understanding of Alteryx is that, many years ago, it was initially created for use cases around spatial data. Megan, you'll correct me if I'm wrong. But I think that that was where the product started. And those spatial use cases clearly weren't football, but actually, you can repurpose almost all of the spatial tools to help you analyze tracking data if the spatial tools are built around points and lines and polygons, then so is the game of football. And much as we might use different words to describe them - for points, we'll say players; for lines, we'll say units; for polygons, we'll say teams or areas of space, areas on the pitch - the fundamental principles are the same. So for me, where I have worked with tracking data, I've always turned to Alteryx to do so. And it's been ideal, to be honest.
MEGAN: 07:00	Yeah. Yeah. So geospatial data. That's cool. I know Maddie had an episode about that a couple back. I think location data's really interesting. And that makes sense what you're saying about, it's very clear who's making the shots, and lots of the metrics are very obvious. Whereas could be kind of just this mine of data, all of this location data, how do you get insights out of that? That's definitely further along on the analytics journey where you would get to that as an organization.
RICHARD: 07:29	Exactly. It takes a maturity and a skill set both in terms of kind of process and infrastructure and also appetite from the organization to invest in--
MEGAN: 07:40	True.
RICHARD: 07:40	--deriving insights from that data.
MEGAN: 07:43	Yeah. Totally. So I'm also wondering about what kinds of technologies clubs are using to bridge that gap between where they're at now and what their future goals for analytics are.
RICHARD: 07:57	So it's a really interesting question. And there's some quite different approaches, get taken across the industry. I mean, each club will be at a different place in its analytics journey. They will have started at different times. They will be going at different speeds. And not everyone's journey, even then, will look the same. They will have different priorities, different circumstances, and take different directions. But what we really see is, I guess, a fundamental choice for these organizations, which is presumably the same as the choice that companies in other industries make. Which is, do they want to use kind of the established industry BI tools, or do they want to create their own customized frontends in which to deliver the insights that they're building? And we see a fairly even split across the industry. The BI tools that get used are the ones you might expect. As is the case for people who are creating their own front ends, they do that using the same kind of languages and programming that you would expect them to that are the industry standards. Beyond what's used to visualize the insights, we haven't yet seen massive tech adoption within the industry, in my view. Solutions like ETL tools, like AutoML, like kind of enterprise level, no code or low code app development I think all have a good distance yet to run. And I think we'll see much-increased adoption of those over the next three to five years within football clubs.
MEGAN: 09:37	Gotcha. So what do you think right now holds back that adoption?
RICHARD: 09:42	I think it's a couple of things. I think firstly, cost or let's say perceived cost. There's a lot of good work getting done within the industry by analysts or people who have perhaps been the first through the door so to speak in a club's data science or analytics department. Budgets for those departments tend to go more on data than they do on technology. I think technology is a harder sell at a business case level. And part of the reason for that is this is a journey that not many people in football have been on before. There's no template for what good looks like. There's no template for the right way to do things. So clubs are really discovering as they go - as of course, are many companies in other industries - as they go through their own processes of digital transformation.
MEGAN: 10:34	Yeah. Totally. So yeah, for the report that you wrote, you ended up talking to 27 football clubs about their analytics departments. And I'm curious to know what trends you saw in the current state of football analytics and there were any surprising conversations that you had preparing for that.
RICHARD: 10:55	So I think probably one trend which was ubiquitous, if you like, was wherever there are analysts who are being able to do good work combining data and technology to create insights that they are not running out of valuable work to do. So there's definitely scope for more insight to come when the resources are put in place to facilitate that. And I'm very much of the view that, as that insight comes, so will the return on investment from a business point of view. So that insight will be important in improving decision-making across all facets of a football club. I think beyond that, we will see an expansion in the departments themselves in terms of analytics departments or data science departments within football clubs. I think the number of stuff in those will grow. Equally, I think we'll see an evolution in the competence and the skill sets, and the data literacy of people in other departments who are increasingly able to self-serve in terms of analytics, who are more competent in coding or in BI tools or in using data to inform day-to-day practice than their predecessors have been just because that's the way they've grown up. They've grown up with greater availability of data and greater availability of the tools to analyze it. So I really think we'll see not just increased value come out of centralized departments, but increased value come out of all the disciplines around a training ground. Whether that be performance analysis, sports science, recruitment, coaching. Think all of those disciplines will become increasingly self-sufficient.
MEGAN: 12:48	Awesome. Did you say one of the disciplines is sports science?
RICHARD: 12:52	I did, yeah.
MEGAN: 12:53	What does that mean? I'm just curious.
RICHARD: 12:55	So that includes many different things, like performance nutrition and conditioning, performance psychology, those kind of elements which can contribute to player development and elite performance.
MEGAN: 13:09	Oh, that's really cool. So then my next question I wanted to ask was, well, it was really because AI, artificial intelligence, is a hot topic right now in the industry and data, and even in all industries I've been seeing a lot about it on LinkedIn. So I'm curious to know what kinds of applications of AI or machine learning do you see in sports.
RICHARD: 13:35	It's a good question. I share your view that you can't move these days for seeing something new about AI.
MEGAN: 13:43	Yeah. For real. [laughter]
RICHARD: 13:44	It Is used to cover kind of all manner of vagaries, but also some very valuable stuff. So where is it used? Where can it be used in football? I guess you can break it down into two different areas. There are AI solutions where football clubs might benefit from acquiring a service from a third party. So, for example, the tracking data we talked about earlier was once only available where you had access to a stadium with cameras in it, and those cameras would capture the data. Computer vision now allows tracking data to be derived from broadcast feeds, which makes it much more widely available, and which will increase the scope of what it can be used for, but also kind of will act as a catalyst for clubs upskilling themselves to be able to use it.
MEGAN: 14:37	Yeah. That's a good point.
RICHARD: 14:38	Equally, I guess, a very important area of AI is machine learning. And I think the majority of clubs I spoke to as part of the project had at least one machine learning model in production. I think soon that will be all clubs. Much as clubs might often buy the outputs of suppliers' models, they will also build their own kind of proprietary models, which will be particularly important in the area of player recruitment and predicting future player performance. Whether that be for fully grown players playing in the first teams or whether it be for young players coming through clubs' academy to be able to talk with greater certainty and greater precision about what those players might go on to achieve in the future in footballing terms and also what they might be able to generate for the club in transfer fee terms is going to be hugely valuable. And I guess to do that, as is the case in any industry, to build and productionize good quality models, you've got to have a certain quality of data processes and data infrastructure and data governance underpinning it all. And for clubs who have laid those foundations already as part of what they've done to date, they will be in a very good position to go on a build those models and take advantage of AI in that regard. For clubs whose infrastructures isn't quite so robust, there might be a bit of retrofitting, let's say, to be able to strengthen it to the point where they can generate those kinds of insights. Because I see a time very soon where it will be normal, completely normalized, that the output of machine learning models is integrated into conversations around player recruitment across all the top leagues in Europe. And clubs who aren't able to do that and aren't able to do it using proprietary models and proprietary data will be very much at a disadvantage.
MEGAN: 16:41	Yeah. That's super interesting. And I think a trend we're seeing outside of just football too of as more applications of AI are available, as more people in the business start to understand how they can use machine learning and invest in it, then those models and those predictions start to weave themselves into the everyday business conversations. Which is super powerful thinking about starting out in the analytics journey being all about, "Okay. What did the data show us? What happened in this game?" Going from that to, "What's going to happen in the next game? What's going to happen next season?" Shifting that conversation, I just feel like, is a huge leap forward.
RICHARD: 17:28	Completely agree. Think in football, as a positive and for the right reasons, we've got more metrics than ever that are descriptive, but we haven't really yet as an industry put our finger on which of the metrics that matter most, which are the most predictive of things we care about, and which are the most predictable. And I think as we get to grips as an industry with ML and in particular using it to kind of evaluate future player performance, so getting to grips with which are the metrics that matter, will be a huge part of that.
MEGAN: 18:04	Totally. Yeah. Like what you said about the metrics that matter because that's also a challenge I think across the industry is once you have a lot of data, you can use that to make endless metrics. And some companies do, and some places do where there's a million dashboards, and then you start to lose value because you start to lose sight of what's actually important here. We have all this data telling us all these things, and we can make a million dashboards. But I guess that's a little bit where the data literacy comes in, where strategy comes in, of where do we focus our time? Where do we get the most value and the most insight? And what data do we need to clear out of the way? What's not as important? [laughter]
RICHARD: 18:47	Yeah. Absolutely. Because we talked a little bit earlier on about the evolution of tracking data and how we're coming to a point where more and more clubs are able to process it and able to manipulate it. From tracking data, we will be able to derive an infinite number of metrics, and different companies will derive different metrics and talk about them in different ways and put different emphasis on them. And as the end user as the decision maker, to what extent are you able to cut through the noise and find the signal? To what extent are you able to dismiss the 95% and really focus your attention on the 5% of metrics that really do matter in terms of affecting whatever it is we're interested in at that point? Is it player performance tomorrow? Is it player performance in five years' time? Is it inflation in the transfer market? Is it the valuation of players? There's a load of different uses for this, but the commonality across all of them is we'll need to be able to hone-in very quickly and very accurately on which are the metrics that really matter.
MEGAN: 19:54	Yeah. Definitely. Okay. So I wanted to close out by asking, for our listeners who are interested in sports analytics, I know it's a really popular topic. I mean, who doesn't love sports and a lot of our listeners love data too, so. [laughter] I'm wondering what your advice is for getting into this field. Pun intended. [laughter]
RICHARD: 20:15	So I suppose I'll talk about my route to it a little bit because it's what I know best. So I never set out to get involved in sports analytics. Sports analytics probably wasn't even really a thing when I did set out. All I ever really focused on was improving my understanding of what happened in football. And I did that a number of different ways. I followed the game, and I watched the game. Obviously, I started to follow the industry more broadly and understand what was happening at a macro level. In that regard, I coached and I scouted, and I still do coach. But really, I guess where it became clear to me that there was value in analytics and value in data was around 10 years ago when the event data that was used still is used started making its way more and more into the public domain. And at that point, I was someone who was, let's say, numerate but had no background in computer science or kind of big data analysis or anything similar, but then I started to teach myself about what could be done with data, how that could help me better understand the things that I was interested in, and how I might get myself to a certain level, let's say, in being able to self-serve in that regard, and built the insights that I thought would be useful to myself and others. But the great thing about sport really is that, in some ways, there's no barriers to getting into it. It might be more challenging to earn a living out of sports for a certain level of, let's say, data proficiency than it will be in other industries. There will be a greater scarcity of jobs, and those jobs won't necessarily pay quite as well, but if sports is your passion, there'll always be a sports team or kind of a sports franchise who is willing to let you go and help if what you want to do is contribute and learn. I think because sports is everywhere, and it's a passion for so many people and a hobby for so many people, there'll always be someone who's willing to let you contribute and willing to support your learning. And if you find them and pursue it from there, then who knows what route it might take you?
MEGAN: 22:27	Yeah. And I feel like I've seen that with other data analysts at companies all throughout the industry, where you work your way into it with that domain knowledge. Maybe you're in HR or finance or whatever. You get to know the processes, what kinds of data there is. You get to have that domain knowledge, and then you can learn the tools. And yeah, something that I love about Alteryx is it is very learnable. Tools like that where, once you understand the data, the jump to learn the tools and start applying analytics is not as big. It's really just understanding the data and the industry and everything.
RICHARD: 23:14	Yeah. No. Couldn't agree more. For me, from personal experience, kind of Alteryx was my gateway from understanding what others were doing in terms of let's call it football analytics to being able to replicate some of those things, and in some cases, go beyond them myself. And as someone who doesn't have a coding background, but knew what I wanted to know, and was able to apply the logic that you get of the canvas and the tools, then for me, it was a natural tool to use, and I still do use it very heavily.
MEGAN: 23:48	That's awesome. And one last thing. You mentioned that you coach football as well. What do you love most about doing that? What keeps you doing that on top of everything else?
RICHARD: 24:01	I guess, probably for myself, like many others, I didn't get into football for numbers, like expected goals and pass completion percentages, and that kind of thing. I got into it because it's a great way to meet people. It's a great experience. It's hugely enjoyable. For I think young people, the social aspect of it's really important, regardless of background. So I still coach the local under sevens team, and the kids enjoy it, and it's nice to see them develop. It's nice to see them have fun. And I think football's given me a huge amount over the years, so to kind of to help other people get something from it and hopefully help set them up for life in other regards is a nice way to give something back, I think. [music]
MEGAN: 24:47	Well, it's been so fun to talk to you today.
RICHARD: 24:50	Thank you, Megan. It's been my absolute pleasure. And kind of thanks once again to Alteryx and continuing the partner over here in the UK for supporting the publication, for supporting the report.
MEGAN: 25:05	Thanks so much for listening. Be sure to check out the report, Analytics in Football: The State of Play, in our show notes at community.alteryx.com/podcast. And to continue the sports analytics conversation, tune in for Alteryx Fanalytics, where we showcase analytic insights and applications across the most watched sports and leagues in the world, including F1, the NBA, NFL, Premier League, and the PGA Tour. See you next time.

MEGAN: 00:04 [music] Imagine you're playing your favorite sport, and location data is captured at 25 frames per second. Now imagine that you can use that data along with technology like machine learning to improve your performance. When you think about it, that's actually a lot of data capture, and the analytics opportunities are vast. So what are the metrics that really matter? And how can analytics help sports teams succeed? I'm Megan Dibble, data journalist for the Alteryx Community, and today we're talking with Richard Battle from Left Field Football Consulting based out of Liverpool in Northwest England. Richard is a consultant in football working across strategy and analytics. RICHARD: 00:44 When I say football, perhaps I should clarify for our audience, I'm talking about soccer as some might refer to it, but I've worked in football for 10, 15 years now with clubs, leagues, federations, and investors. So very happy to come and talk to you guys about some of that today. MEGAN: 01:00 With support from Alteryx and Alteryx partner Continuum, Richard helped drive Analytics in Football: The State of Play, which is a deep dive report into the building blocks of analytic strategy at football clubs. Fun fact, I also helped a bit with this report. In this episode, you'll hear how football clubs think about data literacy, analytics maturity, AI, and spatial analysis. And as a bonus, if you love when the analytics world collides with sports, be sure to check out Alteryx Fanalytics for riveting, in-depth looks at the world's most popular sports. Links to Richard's report and Alteryx Fanalytics will be in our show notes at community.alteryx.com/podcast. Let's get started. So I'm curious what your role as a football consultant looks like. What kinds of things do you advise on? RICHARD: 01:54 So there's quite a wide variety. So I suppose my background going back a long time is in finance and accountancy. Much more recently, I've worked in football on topics such as player development, squad development, and high performance. So my work can really be across any of those strands. And my focus is very much on what I'll call the football side of football. So over recent years, the last couple of decades, football has become an industry, as well as a game. That industry creates lots of work in terms of marketing, in terms of fun engagement, all those things that go around operating a professional sports club or legal franchise. But my focus within football is very much on the game itself and on the disciplines which support player development and team performance. MEGAN: 02:49 Okay. Awesome. Yeah. And I mentioned the analytics report. So you work with football data, and I'm curious to know what kinds of data do football clubs have and use. RICHARD: 03:03 Good question. The report focuses on different kinds of data. So there are lots of data types which football clubs get given based on the league in which they play. And that can include event data, which tells you things like who kicked the ball, how they kicked it, where they were when they kicked it, and so on. You can also get what's called tracking data, which tracks the position of every player and the ball at a rate of kind of 25 or 50 times per second and gives you X, Y, and Z coordinates of the location around the pitch, which is kind of a more recent development in the football data industry and has the potential to give a much greater and deeper and richer insight into what it is that goes on the pitch than event data historically has done. So those are probably two of the main types of data that football clubs use, but equally, there are tons of other kinds. So it could be related to player contracts, player salaries, player injuries, player transfers. All different elements really, which help clubs and teams, and other organizations to better understand A, what goes on the pitch, and B what the market for talent in the industry looks like around the pitch. MEGAN: 04:18 Gotcha. Yeah, the data you mentioned about player location on the field, what was the frequency? RICHARD: 04:26 So most commonly is 25 frames per second. MEGAN: 04:30 That's a lot of data. Do clubs ever have trouble analyzing that much data or visualizing it? RICHARD: 04:37 So it is a lot of data. It's where you go from thousands of rows per game in event data to millions of rows per game. And yeah, as a result, there are fewer clubs, let's say, which are fully competent in processing and manipulating, and extracting value from that data. The industry is moving along its analytics journey, let's say, to a point where more clubs are becoming tracking literate, but event data is still the staple for the vast majority. I think it's a question not just to volumes of data, but also then of the skill sets needed to process and understand that data and the types of insights it gives you. I think event data is much more intuitive. You can create shot counts, you can create pass counts, you can create derivatives of those metrics. Whereas, tracking data is essentially a series of points which move in different ways through an area over time. And I guess in the world of Alteryx, people are very competent in talking about the world of points and lines and polygons, and football really is based on all those same principles, even if we use slightly different language to describe it. MEGAN: 05:50 Yeah. You mentioned all this tracking data that football clubs have. So how do they end up analyzing that? What tools do they use? What tools have you used? RICHARD: 06:02 So for me personally, I'm a big fan of using Alteryx for that data. I guess my understanding of Alteryx is that, many years ago, it was initially created for use cases around spatial data. Megan, you'll correct me if I'm wrong. But I think that that was where the product started. And those spatial use cases clearly weren't football, but actually, you can repurpose almost all of the spatial tools to help you analyze tracking data if the spatial tools are built around points and lines and polygons, then so is the game of football. And much as we might use different words to describe them - for points, we'll say players; for lines, we'll say units; for polygons, we'll say teams or areas of space, areas on the pitch - the fundamental principles are the same. So for me, where I have worked with tracking data, I've always turned to Alteryx to do so. And it's been ideal, to be honest. MEGAN: 07:00 Yeah. Yeah. So geospatial data. That's cool. I know Maddie had an episode about that a couple back. I think location data's really interesting. And that makes sense what you're saying about, it's very clear who's making the shots, and lots of the metrics are very obvious. Whereas could be kind of just this mine of data, all of this location data, how do you get insights out of that? That's definitely further along on the analytics journey where you would get to that as an organization. RICHARD: 07:29 Exactly. It takes a maturity and a skill set both in terms of kind of process and infrastructure and also appetite from the organization to invest in-- MEGAN: 07:40 True. RICHARD: 07:40 --deriving insights from that data. MEGAN: 07:43 Yeah. Totally. So I'm also wondering about what kinds of technologies clubs are using to bridge that gap between where they're at now and what their future goals for analytics are. RICHARD: 07:57 So it's a really interesting question. And there's some quite different approaches, get taken across the industry. I mean, each club will be at a different place in its analytics journey. They will have started at different times. They will be going at different speeds. And not everyone's journey, even then, will look the same. They will have different priorities, different circumstances, and take different directions. But what we really see is, I guess, a fundamental choice for these organizations, which is presumably the same as the choice that companies in other industries make. Which is, do they want to use kind of the established industry BI tools, or do they want to create their own customized frontends in which to deliver the insights that they're building? And we see a fairly even split across the industry. The BI tools that get used are the ones you might expect. As is the case for people who are creating their own front ends, they do that using the same kind of languages and programming that you would expect them to that are the industry standards. Beyond what's used to visualize the insights, we haven't yet seen massive tech adoption within the industry, in my view. Solutions like ETL tools, like AutoML, like kind of enterprise level, no code or low code app development I think all have a good distance yet to run. And I think we'll see much-increased adoption of those over the next three to five years within football clubs. MEGAN: 09:37 Gotcha. So what do you think right now holds back that adoption? RICHARD: 09:42 I think it's a couple of things. I think firstly, cost or let's say perceived cost. There's a lot of good work getting done within the industry by analysts or people who have perhaps been the first through the door so to speak in a club's data science or analytics department. Budgets for those departments tend to go more on data than they do on technology. I think technology is a harder sell at a business case level. And part of the reason for that is this is a journey that not many people in football have been on before. There's no template for what good looks like. There's no template for the right way to do things. So clubs are really discovering as they go - as of course, are many companies in other industries - as they go through their own processes of digital transformation. MEGAN: 10:34 Yeah. Totally. So yeah, for the report that you wrote, you ended up talking to 27 football clubs about their analytics departments. And I'm curious to know what trends you saw in the current state of football analytics and there were any surprising conversations that you had preparing for that. RICHARD: 10:55 So I think probably one trend which was ubiquitous, if you like, was wherever there are analysts who are being able to do good work combining data and technology to create insights that they are not running out of valuable work to do. So there's definitely scope for more insight to come when the resources are put in place to facilitate that. And I'm very much of the view that, as that insight comes, so will the return on investment from a business point of view. So that insight will be important in improving decision-making across all facets of a football club. I think beyond that, we will see an expansion in the departments themselves in terms of analytics departments or data science departments within football clubs. I think the number of stuff in those will grow. Equally, I think we'll see an evolution in the competence and the skill sets, and the data literacy of people in other departments who are increasingly able to self-serve in terms of analytics, who are more competent in coding or in BI tools or in using data to inform day-to-day practice than their predecessors have been just because that's the way they've grown up. They've grown up with greater availability of data and greater availability of the tools to analyze it. So I really think we'll see not just increased value come out of centralized departments, but increased value come out of all the disciplines around a training ground. Whether that be performance analysis, sports science, recruitment, coaching. Think all of those disciplines will become increasingly self-sufficient. MEGAN: 12:48 Awesome. Did you say one of the disciplines is sports science? RICHARD: 12:52 I did, yeah. MEGAN: 12:53 What does that mean? I'm just curious. RICHARD: 12:55 So that includes many different things, like performance nutrition and conditioning, performance psychology, those kind of elements which can contribute to player development and elite performance. MEGAN: 13:09 Oh, that's really cool. So then my next question I wanted to ask was, well, it was really because AI, artificial intelligence, is a hot topic right now in the industry and data, and even in all industries I've been seeing a lot about it on LinkedIn. So I'm curious to know what kinds of applications of AI or machine learning do you see in sports. RICHARD: 13:35 It's a good question. I share your view that you can't move these days for seeing something new about AI. MEGAN: 13:43 Yeah. For real. [laughter] RICHARD: 13:44 It Is used to cover kind of all manner of vagaries, but also some very valuable stuff. So where is it used? Where can it be used in football? I guess you can break it down into two different areas. There are AI solutions where football clubs might benefit from acquiring a service from a third party. So, for example, the tracking data we talked about earlier was once only available where you had access to a stadium with cameras in it, and those cameras would capture the data. Computer vision now allows tracking data to be derived from broadcast feeds, which makes it much more widely available, and which will increase the scope of what it can be used for, but also kind of will act as a catalyst for clubs upskilling themselves to be able to use it. MEGAN: 14:37 Yeah. That's a good point. RICHARD: 14:38 Equally, I guess, a very important area of AI is machine learning. And I think the majority of clubs I spoke to as part of the project had at least one machine learning model in production. I think soon that will be all clubs. Much as clubs might often buy the outputs of suppliers' models, they will also build their own kind of proprietary models, which will be particularly important in the area of player recruitment and predicting future player performance. Whether that be for fully grown players playing in the first teams or whether it be for young players coming through clubs' academy to be able to talk with greater certainty and greater precision about what those players might go on to achieve in the future in footballing terms and also what they might be able to generate for the club in transfer fee terms is going to be hugely valuable. And I guess to do that, as is the case in any industry, to build and productionize good quality models, you've got to have a certain quality of data processes and data infrastructure and data governance underpinning it all. And for clubs who have laid those foundations already as part of what they've done to date, they will be in a very good position to go on a build those models and take advantage of AI in that regard. For clubs whose infrastructures isn't quite so robust, there might be a bit of retrofitting, let's say, to be able to strengthen it to the point where they can generate those kinds of insights. Because I see a time very soon where it will be normal, completely normalized, that the output of machine learning models is integrated into conversations around player recruitment across all the top leagues in Europe. And clubs who aren't able to do that and aren't able to do it using proprietary models and proprietary data will be very much at a disadvantage. MEGAN: 16:41 Yeah. That's super interesting. And I think a trend we're seeing outside of just football too of as more applications of AI are available, as more people in the business start to understand how they can use machine learning and invest in it, then those models and those predictions start to weave themselves into the everyday business conversations. Which is super powerful thinking about starting out in the analytics journey being all about, "Okay. What did the data show us? What happened in this game?" Going from that to, "What's going to happen in the next game? What's going to happen next season?" Shifting that conversation, I just feel like, is a huge leap forward. RICHARD: 17:28 Completely agree. Think in football, as a positive and for the right reasons, we've got more metrics than ever that are descriptive, but we haven't really yet as an industry put our finger on which of the metrics that matter most, which are the most predictive of things we care about, and which are the most predictable. And I think as we get to grips as an industry with ML and in particular using it to kind of evaluate future player performance, so getting to grips with which are the metrics that matter, will be a huge part of that. MEGAN: 18:04 Totally. Yeah. Like what you said about the metrics that matter because that's also a challenge I think across the industry is once you have a lot of data, you can use that to make endless metrics. And some companies do, and some places do where there's a million dashboards, and then you start to lose value because you start to lose sight of what's actually important here. We have all this data telling us all these things, and we can make a million dashboards. But I guess that's a little bit where the data literacy comes in, where strategy comes in, of where do we focus our time? Where do we get the most value and the most insight? And what data do we need to clear out of the way? What's not as important? [laughter] RICHARD: 18:47 Yeah. Absolutely. Because we talked a little bit earlier on about the evolution of tracking data and how we're coming to a point where more and more clubs are able to process it and able to manipulate it. From tracking data, we will be able to derive an infinite number of metrics, and different companies will derive different metrics and talk about them in different ways and put different emphasis on them. And as the end user as the decision maker, to what extent are you able to cut through the noise and find the signal? To what extent are you able to dismiss the 95% and really focus your attention on the 5% of metrics that really do matter in terms of affecting whatever it is we're interested in at that point? Is it player performance tomorrow? Is it player performance in five years' time? Is it inflation in the transfer market? Is it the valuation of players? There's a load of different uses for this, but the commonality across all of them is we'll need to be able to hone-in very quickly and very accurately on which are the metrics that really matter. MEGAN: 19:54 Yeah. Definitely. Okay. So I wanted to close out by asking, for our listeners who are interested in sports analytics, I know it's a really popular topic. I mean, who doesn't love sports and a lot of our listeners love data too, so. [laughter] I'm wondering what your advice is for getting into this field. Pun intended. [laughter] RICHARD: 20:15 So I suppose I'll talk about my route to it a little bit because it's what I know best. So I never set out to get involved in sports analytics. Sports analytics probably wasn't even really a thing when I did set out. All I ever really focused on was improving my understanding of what happened in football. And I did that a number of different ways. I followed the game, and I watched the game. Obviously, I started to follow the industry more broadly and understand what was happening at a macro level. In that regard, I coached and I scouted, and I still do coach. But really, I guess where it became clear to me that there was value in analytics and value in data was around 10 years ago when the event data that was used still is used started making its way more and more into the public domain. And at that point, I was someone who was, let's say, numerate but had no background in computer science or kind of big data analysis or anything similar, but then I started to teach myself about what could be done with data, how that could help me better understand the things that I was interested in, and how I might get myself to a certain level, let's say, in being able to self-serve in that regard, and built the insights that I thought would be useful to myself and others. But the great thing about sport really is that, in some ways, there's no barriers to getting into it. It might be more challenging to earn a living out of sports for a certain level of, let's say, data proficiency than it will be in other industries. There will be a greater scarcity of jobs, and those jobs won't necessarily pay quite as well, but if sports is your passion, there'll always be a sports team or kind of a sports franchise who is willing to let you go and help if what you want to do is contribute and learn. I think because sports is everywhere, and it's a passion for so many people and a hobby for so many people, there'll always be someone who's willing to let you contribute and willing to support your learning. And if you find them and pursue it from there, then who knows what route it might take you? MEGAN: 22:27 Yeah. And I feel like I've seen that with other data analysts at companies all throughout the industry, where you work your way into it with that domain knowledge. Maybe you're in HR or finance or whatever. You get to know the processes, what kinds of data there is. You get to have that domain knowledge, and then you can learn the tools. And yeah, something that I love about Alteryx is it is very learnable. Tools like that where, once you understand the data, the jump to learn the tools and start applying analytics is not as big. It's really just understanding the data and the industry and everything. RICHARD: 23:14 Yeah. No. Couldn't agree more. For me, from personal experience, kind of Alteryx was my gateway from understanding what others were doing in terms of let's call it football analytics to being able to replicate some of those things, and in some cases, go beyond them myself. And as someone who doesn't have a coding background, but knew what I wanted to know, and was able to apply the logic that you get of the canvas and the tools, then for me, it was a natural tool to use, and I still do use it very heavily. MEGAN: 23:48 That's awesome. And one last thing. You mentioned that you coach football as well. What do you love most about doing that? What keeps you doing that on top of everything else? RICHARD: 24:01 I guess, probably for myself, like many others, I didn't get into football for numbers, like expected goals and pass completion percentages, and that kind of thing. I got into it because it's a great way to meet people. It's a great experience. It's hugely enjoyable. For I think young people, the social aspect of it's really important, regardless of background. So I still coach the local under sevens team, and the kids enjoy it, and it's nice to see them develop. It's nice to see them have fun. And I think football's given me a huge amount over the years, so to kind of to help other people get something from it and hopefully help set them up for life in other regards is a nice way to give something back, I think. [music] MEGAN: 24:47 Well, it's been so fun to talk to you today. RICHARD: 24:50 Thank you, Megan. It's been my absolute pleasure. And kind of thanks once again to Alteryx and continuing the partner over here in the UK for supporting the publication, for supporting the report. MEGAN: 25:05 Thanks so much for listening. Be sure to check out the report, Analytics in Football: The State of Play, in our show notes at community.alteryx.com/podcast. And to continue the sports analytics conversation, tune in for Alteryx Fanalytics, where we showcase analytic insights and applications across the most watched sports and leagues in the world, including F1, the NBA, NFL, Premier League, and the PGA Tour. See you next time.

This episode was produced by Maddie Johannsen (@MaddieJ), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.

Alter Everything

Episode Guide

132: Scoring big with analytics in football

Panelists

Topics

Transcript

This episode was produced by Maddie Johannsen (@MaddieJ), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.