Director of Analytics at Opta (Football Data!) - Jonny Whitmore #109

The way we use Statistics and Worked Data in football is quite tunnel visioned in our way. You played a ten yard pass but actually now we can say that you had ten different options. We could see the difficulty and the threat of each one of your options on the ball at that moment. One will be a science at the home team, one will be a science at the waiting. And essentially it's a somewhat computer game where they're clicking onto the screen where the passes start and they have keyboard shortcuts for event types whether it's a pass, interception, tackle, shot. We had I think one and a half thousand applications for the recent data analyst roles. If you have more information on where player positions are, where goalkeeper positions are, the pressure that a player was under as they took that shot then you can really improve that model. We are simulating the remaining tournament 100,000 times. It's a great tool for storytelling and recruitment from that perspective of who's, who are the players that break down the opponent in that context. There's almost too much to dive into in the minute. We've got a backlog of ideas and analysis we're going to do which is a great position to be in. Hello and welcome back to the podcast and today's guest is Johnny Whitmore. So Johnny is currently the director of Analytics at Bob2Sports who of course are a huge player when it comes to football data and football analytics. So Johnny actually joined up to seven and a half years ago as a data analyst and in that time he's progressed up into his current position. So in this conversation we learn more about Johnny's journey and specifically how he transitioned from a financial background into the football analytic space. We also learn a lot more about Opti itself. So I like many of us just know Opti as the data company. So there's a lot more than they actually do which I think a lot of people don't know in terms of how their data is ingrained in many different sectors. So it's really interested to learn about that. We also mentioned different metrics that are used within the world of football analytics and of course we speak about the world cup as we aren't made way through it at this point. So really interesting conversation. Thanks again to Johnny for his time and I hope you enjoyed it as much as I did. Okay so before we bring in today's guest I'm excited to say that this episode is sponsored by DataCamp. Whether you want to learn Python improve your data science skills or get educated on AI, DataCamp literally has hundreds of courses for varying skill levels across loads of different topics. The DataCamp platform is perfect for learning the essential skills that a modern football analyst needs. But more on that later let's get into today's episode. Johnny welcome to the podcast. How are you doing today? Oh good thank you thank you for having me on. No no real pleasure and looking forward to speaking with you. I will I want to start with a disclaimer actually because we're recording this on probably the hot day of the UK. So if you see sweat dripping down one of our brows that's that's why so and we're also recording this kind of as the last round of World Cup fixtures. I just want to put that to we may bring up the World Cup as we're going kind of going through this and obviously this is going to come out maybe a couple of weeks after we're recording. So so Johnny what I'll do first obviously I'll pass it straight to you. So if you want to kind of give a brief introduction who you are for those who don't know what you're currently doing and basically how you kind of got to the position that you're in really. Yeah of course so I'm Johnny with more director of analytics at Opta and stats perform. I've been at the company for about seven and a half years now. So always working in the analytics space with data roles kind of evolved quite a bit over time sort of working a lot more with clubs initially now being doing a lot more work in the media space. But essentially my role is to take the advanced metrics like the traditional kind of expected goals of the past and being able to convince other fans or pundits at Royal Castors media clients journalists of the best use cases and how to use it. So how to tell the best stories with data. So it's kind of a I always say kind of a bit of the handyman within the business in terms of we work with every department we do 10 different jobs every day but essentially it's about translating data and making it easy to understand for the kind of the everyday sportsman. Yeah nice so we will get some of them advanced metrics. So and just just clear up so I know you mentioned kind of Opta stats for just sort of for people that can better understand it because I hope my research is correct here so Opta that's what it was originally stats perform was formed in 2019 when stats merge with perform group which perform group armed optos so that's kind of why some people use them interchangeably so just so if people are watching and they're thinking if whether the different things that's that's kind of just wanted to put that out there. So perfect so the purpose of today then really Johnny is want to learn a bit more about you and your journey in sports in football analytics and kind of it touches on a few topics that we kind of speak on this this channel quite often in terms of people getting into the football space with like a non football degree as you have but also then just kind of shine on opportunities are outside of the club setting so in the commercial kind of field that you're in so so we'll cover that and then we'll kind of touch on some of the metrics as I mentioned and then I'm sure we can't not mention the World Cup as we're made way through it so we'll definitely bring up some World Cup stuff as well. So so you mentioned seven and a half years ish Opta but that's kind of across a few different roles and Johnny you kind of looks like you kind of work your way up into different roles so we will get to that but before we do just kind of curious to ask you how you first got into football analytics because we mentioned that non football degree which many of the guests on this channel have also done a similar path and I will link some down below with a link to Opta as well so Tom Orville, Sam Gregory who you probably kind of know of so yeah just in terms of how you first got into that kind of analytics and what made you want to follow that as a career path really. Yeah of course so I think growing up I always found numbers interesting in a way so maths maths was always my strength so that's what I studied at university so I did a maths and finance degree it leads so it was always kind of one of my strengths being able to analyse things look at things from a kind of numerical side of a point of view but at the same time I was obsessed with sports growing up so watching playing anything I could get near all sports obviously predominantly football but that was kind of my passion on my obsessions so I came out of university I actually did it a placement year during university at Lloyd's Banking Group as a data analyst so I quickly realised that kind of mortgage data and banking data wasn't my passion as much as I could do it and it was quite it was great it was a great technical introduction great kind of office introduction to working a year and in lead as well so that was a brilliant year for me to learn what I wanted to do what I was kind of good at but also what I didn't want to do so came after university my big focus was how do I kind of mix my passion with my my strengths in a way so I saw a lot of people learn the early days you mentioned Tom Warville Sound Gregg are really those guys were very much in the kind of the early analytics days when the officer only had I think two or three data scientists for example and you know I could see that as a target that I wanted to achieve it wasn't big on social media that was still kind of brewing as well so what I did is I essentially taught myself Python and a bit of arm but mostly Python to go on as I just got my hands on at any football data that I could find basically just started trying to prove that I could work with this data prove I could tell stories essentially self-taught not probably not the best code then I think the the bar for that and even the bar for what I can really showcase and yeah sports analytics works completely different these days but I was really fortunate at the back then that a kind of a junior data analyst position that officer like they were they were hired a couple of data analysts at that point Peter McKeever was was in a one around the same sort of time and yeah it was my my goal was essentially to channel the kind of experience and coding practices and just demonstrate as much as possibly could in terms of in terms of that sports data because I knew I could talk about football just from my experience growing up and just understanding the more the tactical side of things and that's the way how we have always looked at football in that way like it's very analytical sort of way and I've always tried to point out things that most people would find boring so yeah I'd say it was a really good timing but also teaching myself the sort of skills that I knew were going to work in industry. Yeah so you mentioned when rather start maybe it ultimately had maybe a couple of data kind of analysts at the time so what kind of kind of do you kind of fall in or would you class yourself Laura's Johnny when we talk about kind of a data analyst a data scientist a data like what and how does that look in the department New York kind of overseeing now up to what's a makeup of it in terms of the different data roles. Yeah so I definitely a data analyst so that's kind of the mix of technical skills in terms of model building but more on the visualization report building story telling side of things so I say back when I joined there were like three others that were kind of data analysts data science hybrid there was a couple more data scientists quickly joined at that sort of time like nearly was in the car and people like that and so they that kind of that's when they branched off so they realised that with a with a few data scientists they needed the data analysts to take off that kind of day-to-day workload of like requests like helping out clients and things like that so back then it was a much smaller team now I have a team of four data analysts is separate and there's probably in terms of football there's probably eight people in that team at the moment it's a strange one there is also we've got innovation teams that have smaller army of data scientists we have you know Patrick Lucy the Chief Data Scientist he has a kind of small group of data scientists as well so obviously there's a lot of work in the background with computer vision so the football only data scientists probably seven or eight mark and and kind of four data analysts data engineers there's another army of data engineers elsewhere but yeah we've definitely specialised and been able to split up the groups a bit more yeah are you all based in the UK then Johnny are you kind of remote like all the kind of data team what what it look like kind of where you spread out everywhere to your list no probably the least in the UK I guess my team, there's myself and John are in the UK, but got ETL who's based in Tenerife, Yash who's based in India, and then the data science team is, well, used to be UK, Netherlands, Spain, Italy, India again, there's Australia, one of the guys has just moved over to London from there, so it's kind of worldwide really. So in terms of that, you've gone from outside of sport, because what I wanted to ask you about that in terms of the skills people learn outside of football, so if you come from the banking industry, you come from the business sector, because I do get a lot of people watch this channel that are in a role outside of football, and they're similar to you, they're not passionate about that, they're passionate about football in terms of they want to apply the skills in something they actually like doing, so do you feel, well, firstly, I can ask you about advice on what those people could do, but do you also feel like people outside of the industry may have a better, they may be able to learn the skills better than if they were to start in a football club almost, in terms of, I'm talking about more data-specific skills, because maybe those industries, and I might be totally wrong, but maybe those industries are much more natural to the finance sector to be doing a lot of data analytics, whereas football has obviously come a long way since you start it, but do you feel like you might get a better learning environment outside of football and then transitioning in, or is that my just kind of wrong with that, you reckon? I definitely think so. I think the way we use statistics and work with data in football is quite tunnel-visioned in a way, because there are certain use cases, there are, I guess, the audience is not going to be that super-quant-level finance industry huge reports, so I think it's almost like simplifying it in some ways, even when you start using the Z-Z-Z scores and things on social media, or to the everyday fan that turns a lot of people off, so I would say, yeah, I think I love it when new people come in from different industries, because they bring a new, like, a fresh perspective, new ideas, I think it's like so productive for all of the teams to be able to learn from new experience in different industries. I think even some of the basic skills, to say, like the core skills, I think for me over the years, like, I'm not working too many industries, but in terms of, like, teaching, so, like, in the past, like, level level teaching bits that I've held with, that's a huge part of my role in terms of translating data, storytelling, the kind of classic things that every data analyst in the world does, but essentially a lot of what, most of what I do every day is teaching somebody, whether it's explaining an error, explaining how a metric works, explaining how they can use it, and it all comes down to the core kind of communication skills. Yeah, definitely. I'm glad you brought that up, because I was going to bring that up myself, so that's good. So, you mentioned when you first kind of wanted to dive into football, then you're like learnt Python, and I'm just trying to want to ask you, data now in Opta, maybe you're more of a leadership role now, but throughout your time, Opta, like, what kind of tools, software, skills are important for an analyst that's working Opta, like, what would you actually use and do day-to-day? I think it's a mix of Python and R, it's been an Opta, in the data analyst team, we use a lot of R for data visualisations, so most of what you see on kind of Dr. Analyst, Dr. Joe, where it's kind of the pitch map visualisations or like scatterplots, those sorts of style, they're all GG plot in R, and a lot of the data science work and engineering is Python, so naturally they are quite transferable, I think, let's say when we hire, we never hire exclusively for one of the skills. If you have one, it's very easy to kind of work between languages, especially like learning on the job, so definitely I think there's that kind of central, and I think, yeah, I've like, yeah, forget the phrase, but like management tools where like, like, GitHub and that sort of thing, that type of thing, SQL, you know, to query the database, fairly kind of low basic skills, I guess, for most data, data, people working with data, but yeah, I think some sort of coding language, I think is somewhat essential within these types of roles, but yeah, it's definitely a mix of these. Okay, so time for a quick shout out to the sponsor of today's video, which is DataGampe. If you want to work in football analytics, you all have heard of Python, Tableau and SQL. You know that they matter, but knowing that you need to learn them and actually knowing where to start are two very different things, and that gap is where most people get stuck. With DataGampe, the friction is completely removed. You don't have to install anything, configure databases or piece together that random tutorials. You're logging in and start learning right away, writing real code directly in the browser from day one. I first came across DataGampe when I had Rob Sutterby on the podcast, a data insights analyst at FIFA. He mentioned that he had used it to upskill himself on certain different data skills, and so I thought if it was good enough for him, it's definitely worth checking out. For football analytics, specifically add point you towards two tracks. With Python as focus, their data analyst with Python track takes you from the fundamentals to advanced data manipulation and visualization. The exact skills you'd use, building player performance models and analyzing match data. For Tableau, their data analyst in Tableau track covers everything from your first dashboard to more advanced storytelling with data. The kind of work clubs actually want to see in your portfolio. Both tracks offer real structure, so you always know what to learn next, which I think is really important if you are fitting this around your current job. Once you complete a track, you can work towards a data campsite certification, which you can put on your LinkedIn, which signals that your learning was structured and practical, not just a YouTube playlist. Link is down below in the description. Now let's get back to today's video. I use less and less these days, but we're going to say it's similar to when I speak to an analyst at working clubs. They work the way up into the head of the department and they're actually doing less analysis and what they actually did to get there. It's interesting how it obviously rolls, evolving stuff. But in terms of being involved in hiring, bringing in a new data analyst out of the job, can we go through that in terms of what kind of things are you looking for when, let's say, you're advertising for a data analyst to join your team, what you're focusing on, what you prioritize and potentially, what maybe people maybe underestimate. There'll be a lot of people watching this, which we probably love a job with you. Just the things that you consider and maybe people on the value. Yeah, of course. I mean, what's interesting as well is that we hired good few years ago now and recently. So within the last 12 months, we hired ETL MYASH. It was really interesting seeing the change in applications between that and maybe three four years ago. Maybe further a long ago. The big thing for me is almost the generic phrase of show you can do the job before you can do it. So we had, I think, probably like one and a half thousand applications for the recent data analyst roles. And as you just imagine, the number of CVs that, like, how long that takes to get through. And the biggest thing for me was almost just scanning CVs, being able to pick out somebody who's worked with sports data. And in that, we really got that many applications. You kind of have to see the people who have already done some work. And that doesn't mean working at a club, doesn't mean working in the industry as already. It's proven that you can do it. So even just to get her repo, obviously some people like to post their stuff on social media, which is in itself brilliant. There were names on there that I recognized the name just from popping up on my feed. And I would recommend that to anybody being able to push your work out there. If no one sees your work, then nobody can, nobody knows your doing good work. So for me, that was the biggest thing showing that you could do it. It doesn't matter really a technical background in terms of education. It's kind of irrelevant as long as you can prove you can do it. And for me, in the data analyst specific role, it was about visualizing that data in particular. So having that kind of unique brand in a way or being able to tell a story in a really clean way. Obviously, now the interesting thing with this round of applications was the creep of AI. And people who being able to distinguish between who's put something into an LLM and got a really nice output and who's able to tell a story with that data themselves, or kind of take that visualization and being able to take to the next step and explain them football terms. So that was interesting from the application stage to the interview stage was differentiating these people, but as I say, a bit of access to data now, the level of stuff that's done by people as a hobby on social media is really impressive. Yeah, there's some really good work out there and they'd like just people just doing it in the spare time and stuff. So yeah, I'm sure the best ones will, the cream rises to the top as they say. So what I'd like to do then, Johnny, is if you can give us a bit of a, just a bit of an overview of a bit more info on what to basically like. What? Because again, if I go back to my time as an analyst, this is going back sort of two many years really, but I just, you know, opt is the stats company and that's basically all I knew them often. And I know obviously you're heavily, I'm just going to read a sentence that took off the website actually. So with over 7.2 petabytes of real time and historical data across 3,900 plus competitions and 20 plus sports. So for those that don't know, petabytes is a big, it's a big one out and there's seven of them there. So yeah, if you can just give a bit of an overview of what opt is and what's the purpose of opt in, where they sit in, we'll just focus on football here. So where they sit in the kind of football analytics ecosystem. If you can. Yeah, of course. So I guess fundamentally we're a data provider, right? So we collect the raw data, so the passes, the shots, the tackles, all of the coordinates. We have tracking data. We kind of had the full package of raw data. But we provide that to a number of different, clients and industries. So we have the big media clients, the likes of Sky and the UK or the athletic from a more digital client. We have hundreds and hundreds of clubs. We work with the federations, the leagues. We work in the betting industry so probably every single major bookmaker takes our data in some form. So we have faster forms of data for their kind of instant goals and being able to help with the markets that they do. But we also supply the markets for shots on target tackles. So it's a wide-ranging role that we do. Automate, yes, say providing the raw data. But we also have products in terms of up to search, up to graphics, up to live. And they're all slightly different purposes in terms of graphics. A lot of the social media things that you see where clubs will post their line-up, they'll post goal maps, sequence maps. That's often done with one of our products where they can overlay their branding and it's all like Fed live via our data. We have the kind of up to search, is that query tool, it's like a research tool. So if you were looking for stats going into the weekend and you were doing the broadcast for the Scotland Brazil game last night and you wanted to find interesting stats, you could go into that tool and kind of create the narratives for particularly historic data. You mentioned some of the numbers there. Obviously the World Cup with batter in 1966. So there's crazy depths of data in a lot of places. And then up to live is our more live in game tool. So get a lot of the pundits, commentators using this tool where we obviously have the live running stats, symptoms of player stats, team stats. But we also have a, almost a live feed of stream of insights. So some of which are automated. So it's the first time our player has done this, this tournament, or the most he's ever done in a World Cup game. We also have an army of researchers, which I can spend another hour talking about. But these guys are the best at what they do in terms of being able to bring out these unique insights and they push them directly into these streams. So I've seen, I've sat with commentators before and they'll just have this stream part of their screen, which tells them that these really cool insights and you'll just see them read them straight out on our word for word. And yeah, we, they often make their commentators and pundits look like they're the best people at their jobs. And sometimes the hard work's in the background. But yeah, as I say, Opta was a company. I think we're naturally global. So when I talk about each of these different products or the different markets, I think we, I can't, I wouldn't know how many clients we have worldwide. But in terms of the data insights, the research team, we have probably over 100 colleagues working kind of live matches, supporting the brand deals, supporting media clients. They work across 20 different countries, 12 different sports, probably 18 languages. So it really are global in sort of every region of the world really. Yeah, that's, I mean, the live aspect is very interesting, obviously you mentioned a couple of the answers to the commentators, but also like I presume that when the bettion, I'll defluxuate in mid game, that's from kind of data that you guys are providing. So, yeah, there's a hell of a lot of the kind of the data is used for. So what does that date, do you have much to do with the actual collection of data joining? Like what can you, can you give us insight into what that looks like on a during a, you know, typical Saturday, Premier League game, all the games that happen, you know, so your Opta collection team are doing some work. So what, what do you look like? How does it kind of come together? Yeah, so I guess two point there is almost the way I explain to people when they don't know who Opta are is that if you see a sports statistic, particularly football, if you search on a search engine, if you've got a live score app, if you've got, you're watching on broadcasts, nearly every one of those numbers is powered somewhere by Opta, or we're feeding those, those kind of those numbers. Yeah, the data collection process, so that's fundamentally what all of this is built on is fascinating. So I've been out to a vario one of our main hubs in Portugal and I've watched it live for I think it was Champions League evening and essentially it's fairly somewhat it's a mix of AI and manualic collected, so it depending on the metric and the type of data we're collecting. But I guess the most interesting one that's been going on for probably, I think, yeah, we recently this year's our 30th year result as well. So that's so it's been going on a long, long time and yeah, so there's typically two data collectors who will one will be assigned to the home team, one will be assigned to the waiting and essentially it's a somewhat computer game where they're clicking onto the screen where the passes start and they have keyboard shortcuts shortcuts for event types, whether it's a pass, interception, tackle, shot, and when there are a shot, for example, where you need more detailed data points, they're able to pause, go into the system, add in all the relevant points that power things like expected goals. These guys have I think it's well over a hundred hours of training before they go into a role like this. Naturally they start on different like lower competitions, for example, where at less prominent in terms of like betting markets or media and they work their way up. So there's a lot of feedback loop in terms of being able to kind of earn like the away through the process and then naturally there's a really extensive post-match process where every single event is reviewed again and double-checked for the coordinates, the data points, there's checkers live so if there's a kind of a melee or you know from a corner, you can flag a moment and then somebody else will go into that check all the angles. We do also have in most competitions multiple angles and camera feeds so when there's contentious events we typically have had seen from more angles than the net the broadcast that has in a way. So we're actually able to make those decisions where as a deflection or kind of minor moments who got the final touch on things. So it's a it's a whole world and and in the final point as well as that there's a separate betting collection like a fast data collection. So not just a betting obviously for live score apps and media as well of we have correspondence in the stadiums to collect the goals and the major moments to suspend the betting markets or get whoever gets the the notification the fastest on the live score apps and things like that. So there's different types of collection there as well. Yeah interesting. So that the point you mentioned about kind of looking at different camera angles Johnny would that be a case of they're not optic like your just like optical itself is just kind of subscribing to different kind of feeds or in the stadium other than what the broadcast has. So you technically have them other views is that it's not kind of optical or not filming it or anything like that. It's just you've got access to the other ones from okay yeah that makes sense. Yeah so it's it's different broadcast feeds it depends on the competition and the relationship we have right so like say FIFA we have access to lots of different FIFA video angles where they give granted access to the likes of the real official official provider in the Premier League so similar situation there they can give us panoramic they can give us the different replay angles so and but primarily it is done from broadcast first in terms of the angles but a lot of the time because of the years of relationships we've had we're able to get more of those angles to be more accurate. Yeah nice so I'm just looking at your LinkedIn profile at the moment Johnny so we're just kind of just to let people know all the roles you've joined as a data analyst then data analyst team lead data analyst team lead and senior data editor data analysis manager and then obviously now director of analytics so some nice little progression there Johnny so can I ask what you're in the current role then what what you actually what you're doing now Johnny like what does an average week look like and you can pick an average week in any time of the year you can do it now during a world cup just so you can get a gauge what you're getting up to really. Yeah of course I guess yes it's changed massively with the years my current role now is more evangelism in a way so I'm kind of the being able to present against conferences join big client deals and almost actors of the football data expert in terms of advising put people of how to use data what data is most effective for them educating them on our new is metrics so like the subdivision suite of metrics that we've got now in terms of blending event data and tracking and so being able to help that so I guess in the last year or so we've done the appearances with with Sky Sports been on the Premier League channel done interviews with like with Flash School and Canal+ two of our partners we did the Peta Check podcast which is brilliant so again almost at talking from a football experts in this conversations I think day to day and a week I still use a bit of our and Python so I'm still creating new graphics say this week in the for the world cup and a lot of we have to be quite reactive to be able to you know fit the later storylines there's been a lot in the last few weeks about hydration breaks so being able to incorporate some of that into our visualizations being able to create reports on the fly to be able to say you know what has momentum died off a bit before and after the breaks as the intensity changed so there is still day to day a lot of coding work but naturally as my role has become a bit more senior there's more meetings and client meetings so in the last few weeks again being in with you calls probably can't name them but a lot of big brands that want to work with data particularly with the world cup and being able to use our data to support campaigns as part of that I think as did some stuff we did in the last year that I can talk about in terms of like x we worked with xbox to design designed the expected Jinks Metrics, Jinks Metric, which was basically a fan design metric to talk about. You're are you costing your team points by watching them versus when you stay at home and do other things than they knew that your team formed better. So a bit of a fun campaign and then I'll always to kind of legitimize it with data and basically assign statistics to yourself or to me to say, you know, on average, if I watch a game, this is how many goals my team scores or concedes or this is how many points or how many times we lost them the year while I was watching a game was either huge Jinks. So yeah, there's a whole kind of underbelly of brand deals that are a massive part of my role now as well. And then naturally there's still the club side of things. So we have again separate departments that work day to day with the clubs and advice. But again, I drew a lot of these conversations to help help a club understand what we can offer them in terms of the types of metrics and where they kind of need future metrics going forwards. Yeah. Is it fair to like has the proportion of stuff that you do in opta like the club side would you say that that's not because he's got small but the other areas have got bigger over time. Like is that kind of a fair assessment like the media side of things and the bet in etc. So football in the club setting maybe takes up less of a percentage and it did 10, 15 years ago or so. I think so. I think in my role particularly more of the more involved with the media side. I think naturally just the size of some of the deals in a way like in terms of like six figure deals or seven figure deals with media partners with betting providers with federations. Whereas the club space I guess there's more deals but like lower like revenue. But I say we do have a whole pro services department that are working on that stuff every day. Being able to advise clubs. I know they're working with some of the national teams at the World Cup and they did the same during AFKON. So this kind of like a it's still it's definitely not something that we do less of in a way. I guess we probably do more of the media of other kind of areas these days than we did in the past. Yeah. So obviously you mentioned you're speaking to kind of you're in these meetings with potential partners and different companies but who else who you who you report into Johnny within up to then who's your boss and who else you kind of conversing with day today other than your team of day-traveless and stuff like that. Yeah. So as a day-traveless team we kind of work for everybody. So my boss is the head of data insights. So that team that has kind of 100 plus colleagues across the world who are the day-to-day researchers at their helping media clients and such during live games pre-game live game post-match but naturally I work a lot with the sales guys marketing products likes of Yeng's Melvank in the product team who's building out the latest kind of AI metrics working close with the data science team. So it's kind of a bit of everyone I would say I say handy men at the business in a way but I mean the data insights world I think as said some of the previous roles you mentioned I was also part of the opt-or-analyst team for a while as well. So being able to help them I was providing some explainer articles building out the visualizations for a lot of the work that they were doing. I'm not much of a writer myself so I kind of a no-mer strenghts are in numbers so I can I leave that to the experts that we have in that team but that's been a fantastic project in the last I think five years or so now yeah those guys getting millions and millions of views every month and I think the World Cup is about to break every record they have in terms of engagement. Yeah yeah nice so just well we're on the kind of football sector then so what it so you mentioned it earlier Optovision Johnny can you just kind of tell us what what that is and how is that used by clubs and or is that used by everyone again like the organizations. Yeah so Optovision is the combination of tracking data and event data so being able to say at the point of a pass where was everybody else on the pitch or the point of a shot exactly where we'll give a player's worth so naturally that enriches all of that content so you played a 10 yard pass but actually now we can say that you had 10 different options we could say the difficulty and the threat of each one of your options on the ball at that moment being able to say didn't break multiple lines what pressure were you under what kind of phase how many runs were being made at the point that you made that so it kind of just paints that full picture naturally it's it's more of a pro pro product I think in a way because they want to go deeper into the data deep and do individual moments assess more the decision making of players I think the media this a bit of pick up in the media a lot of the live score apps again because they want to do more with they want to be the most advanced in terms of data but I think media is always naturally going to take a bit longer to pick this stuff up it's you see it a lot you know expected goals of the past in various metrics the the uptakers slightly slower but again they still want in the conversations with them they want to be seen to be the most advanced users you speak it depends a lot obviously there's a various variation in pundits we want to use this data and who don't but it's interesting when you speak to the likes of say sky BBC any big broadcaster there's a there's kind of a chain of understanding that needs to happen which is why a lot of this comes down to storytelling and teaching and education because ultimately that person at the end of the chain say it's a Jamie Carrebar on Monday that football they need to be the one that buys into this metric and understands it and helps them tell the story they want to tell and if it falls apart anywhere in that chain if I'm speaking to it but that's that sky's at sky and they see value in it they then have to cover the next person that is value in it so it has to kind of go the whole way down the chain and that's why I always say stories are the best way to talk about football data if you can't boil it down to the simple conversation that you'd have with your friend or family member as I always call it the friend test or the big Steve in the pub test if you can't talk to them and tell them what a metrics doing then you kind of got the wrong audience to start off so yeah optimism just almost paints the full picture is not really anything we can't measure now in terms of if you have a football or tactical opinion I don't think I think every single opinion isn't we're now able to put a data point behind it now which I think is pretty it is almost too much to dive into the minute we're we've got a backlog of ideas and analysis we're going to do which is a great position to be in yeah definitely so you met you actually mentioned there like player decision-making and I don't know if you put your own spot because I didn't actually bring this I didn't tell you about this before but we've got a found a video what you did with the training ground guru a few years back and it was around kind of decision-making with the passes and it kind of covered expected receiver and expected past completion and expected threat and how those kind of fit together to see whether that pass was the right decision for the player to pass it up or the player for example one can you kind of remember what you about that but also if you can then can you potentially just build on what I've just said then in terms of what what kind of data goes into trying to quantify decision-making as a as a player on the pitch yeah I think that's still the thing which I'm solving terms of like what's that single metric that's going to have the most power so things like expected passing this difficulty so essentially the the difficulty the likelihood your pass is going to be completed or expected threat the likelihood you're going to get shot within the next sort of 10 seconds those matches have existed for quite a long time now in various clubs using quite extensively a bit harder to push that into the media space yeah for me again as you said it's like what I think I tried to do in that presentation the time was just talk about who what types of decisions players make so how often do you go for the safest option how often you go for the most threatening option how often when you're available is your teammate passed you so like I try I think it's why I always used to look for was kind of the messy the messy test at the time it's like almost I think almost almost every time that messy was available to receive a pass so like in a just in a good position his teammates passed at him and he was just so far ahead of that metric and almost just like relays that trust and it's something that you kind of understand watching the game right you your best players everybody just tries to get the ball to your best players like Limee and your Mal at the the World Cup as soon as he's on the pitch the players channels down that right right hand side or people always trying to get to them so I guess decision making out from a data point of view is difficult in some ways because the best decision is not always the statistically best decision so you have you always have to wait that angle of you know a player playing backwards late in the game in the context of being you know being a bit safer with possession is it's probably the right decision in those moments so as being able to pick up the smaller bits or smaller bits within the context so decision making in which game state do you play safer when you're when you're winning versus when you're losing losing when do you take risks and then naturally I think it from a recruitment perspective is it then gets interesting as well of who are the risk takers so traditional metrics probably fall short in terms of being able to say you pass completion there's a lot of this World Cup players sent back to who played every single pass and completed them but how many of those passes were easy or difficult or what kind of mediums so being able to ban that up in a more understandable way and then give an extra context is always super interesting in there. Is it past studying? That's what they call it these days, isn't it? Like players just literally playing little side side with passes, but the risk aspect is really interesting because if you think of like a central midfielder, let's say one in four passes is going to make it through a great chance, but three of them are going to get accepted and it looks like well, they've only made one pass successful, but that one pass was a result of a more higher position on the pitch and maybe the score from that. So yeah, it is a balance of that and I'm sure you kind of understand that as well, but in terms of what the fans might see and what people are not involved in, it might not kind of appreciate that kind of risk-free one, how important it is to, because we probably see less players now taking risks in my opinion, you know, when you just watch football, it does see a lot of it is more safer, like, all kind of playable, start and we'll kind of go and switch it, whereas you kind of miss that excitement where you might not happen all the time, but when it does, we get a great through ball and then we see a goal or whatever, so yeah, that's interesting. Yeah, I think we, one of the guys from the analyst, he actually went to see Bruno Fernandez and didn't interview him and he loved the advanced metrics side of things, because naturally he's somebody who would appear at the top of the metrics for like losing possession because his player takes his risks and I think this is a great way of showing that's what they're doing and maybe that's why they're losing the ball and I think the other one that kind of lays into all of this as well is that football is going, there's a lot more low blocks, you see in it at the World Cup in particular, like I think the intensity of the game has dropped significantly, so it sprints and high intense movements that are massively down this tournament compared to previous and you see the low block is almost everywhere, and now with these types of metrics again, you can say that, you know, how many passes were against the low block and incredibly boring and safe, even if they were in the final third and who are the players who were actually making a difference during the low block who were trying those line-breaking passes who were attempting those and more threatening more difficult passes, so it's a great tool for storytelling and recruitment from that perspective of like who's who are the players that break down dear opponent in that context and I think again that kind of comes with tracking because to be able to overlay that context on every pass. Yeah, yeah, because definitely because like a side-runs pass between two-centre-half is much different when you get impressed when you're, when I say against the low block and there's no one anywhere near you, so it's yeah that context is key, so okay so a quick interruption for anybody with an interest in video analysis, so I did mention at the start of this video that this episode is sponsored by InPlayOnline, so I wanted to spend just one minute going over exactly what that actually is and why it might be useful to you today. Okay, so InPlayOnline is a cloud-based video analysis software, but unlike other analysis software it's not going to cost you an absolute fortune, so prices start from as little as £10 per month and that's not including the discount code that I've put in the video description down below along with a link to take you to the website to find out more information, but as a really quick overview it will allow you to basically break down your match or training video, so if you are an analyst, a coach or a scout and you're wanting to do more video analysis, the InPlayOnline would be a great option for you to check out, so not only can you tag the videos with various different button types, so a tag of button, a label or even a start stop button, you can then add a bit more detail and depth to the clips themselves, so that includes adding comments, you can even have voice notes and also do drawings or illustrates over your clips to make them more interactive, as well as that you've got video playlists which can be a great way to download clips to store and share them offline, but also store databases of clips in terms of maybe best practice clips, or if you're a scout for example you might have a player, it's for a particular player that you are monitoring, so there's a whole lot more to it, but like I said you can click the link in the video description to find out more details, so that's it for the introduction, so let's get back to the episode. We'll get onto the World Cup, they're actually Johnny, so firstly people that are watching the games, you mentioned visuals earlier on in the conversation, Johnny, but I've noticed one, the momentum that's coming up on the screen, are they momentum charts, is that something that would be using optadata, and if so, or either if it is or not, how is that built, what kind of stuff goes into building one of those momentum charts? Yeah, so this tournament we are with the betting providers, so any bet is officially settled by optadata, but FIFA themselves have their FIFA training data where they provide that for the media partners, so the momentum is an example of FIFA's one of FIFA's broadcast metrics, but it's something that we designed actually about five years ago with the in partnership with the Premier League and Football Datesco, so I remember being in a meeting with Rob Bateman, my boss at the time, and we were discussing, I think it's part of the Oracle Archivist's recent sponsorship partner for the Premier League, so they were looking for the best metrics or new innovative metrics that they wanted to show on broadcast, and so there's a conversation again, as we talked about already, the storytelling, the fan first football approach where we were talking about what do people talk about during a game, and momentum was the word that just kept coming back up, and it was like, well, TMA had the momentum, and that all changed when this happened, and there was no way to show that as a metric, that we would have always had the traditional box scores where shots and targets, possession, and goals, and actually XG has come into that now and tells kind of adds an extra layer of insight there, but we thought momentum was a really key metric, so at the time we built possession value, which is very similar to expected threat, the current things model that a lot of people have seen or aren't on social media and the paper, so we had a way of measuring threat, and we wanted to essentially visualise that, it was the real challenge, and being able to say almost every minute how that was swinging, and essentially that was something we designed to say within the minute, each minute, and obviously a bit of decay from the previous minutes, who was on top, or who created the most dangerous moments, so it was funny, as I say, five years ago we designed the concept, and built the visualization, we've kind of pushed it fairly extensively on the opto analyst, on the social posts, it's only the last couple of years where the mainstream media has really picked this up, so the likes of the BBC I think introduced it of the Premier League season last year, which is obviously a big step for somebody more kind of traditional with BBC, you've really seen a live score up now that doesn't have some sort of momentum tracker, and I think it's a great way, especially the World Cup being able to pick up the next day, see the games from the front on the last night, and almost one of the first things I do when I go on an app is to see how the game played out, you can see that whoever had an momentum in the first quarter, if we were to call it that, the hydration breaks, where the team could just completely dominant, and they still drew the game, it's just that next level of insight, but naturally for the mainstream media in a way that that just takes a little bit of time to keep buying, it's almost the same timeline as we have with expected goals, so yeah, in terms of the creation that was almost a room of football experts, and naturally you get some, you get market research, you take this to market, test the concepts, get some feedback, you know how big thing for us now is testing this across different generations, so younger audience versus older audiences, there's not much use of having kind of 30 plus two-year-old white men in a room telling people what they think is going to be successful in the market, so you just have to get the full perspective, proper market research, and then yeah, testing that it would market, and then kind of getting that kind of easy feedback. Yeah, because it has been a talking point, especially with the advert/hydration breaks, so whatever you want to call them, but like how momentum does swing, and maybe unfortunately as well, for like, say the lesser teams, maybe they are on top, and then they're calling this break, and then that kind of takes a sting out of the game, you see them all having kind of effectively a tactics break, like, so it's a bit unfortunate, and so I would say that probably the recent data already in terms of looking at how the momentum does change, but it'd be interesting to see that over there when it work was finished in terms of the full data set in terms of how much of an effect, and how much the momentum did swing after these hydration breaks and things like that, so a couple more, because you shared a couple of interesting stats with me, Johnny, again on the World Cup, that were quite kind of the first since 1966, I think you mentioned, so, and I think they were both from the same games, Spain and Cape Verde, so firstly we'll go on the, play it didn't touch a ball for half an hour, is that, is that, if I was a legend, I would definitely be checking that and thinking I've made a mistake, but yeah, it's wild, that's crazy. I'm sure that was checked a few times, and since I was a Spain striker, MacGal, a year as well, and in the first 30 minutes he didn't touch the ball, and that was the first time since 1966 that that's ever happened, and then the second one was Cape Verde only, baby, I think their possession was about 20%, but they only committed one foul, which is the lowest the teams ever done in a World Cup game back to 66 again, and again this comes back to this, this team of researchers that I mentioned, so I think we're working with, probably it's kind of joins in with what opt to do as well, so we're working with I think over 70 different media clients, probably triple that in terms of people taking raw data, but our researchers are working with kind of 70 odd media clients, probably 30 of those are more broadcasters, so you can see, we kind of plug into the football storytelling EQ system in so many different ways, so those facts were great examples because they became kind of viral content on social media, so there's probably 30 different outlets plus that we're using those stats. that both stats were actually posed to the Spain manager after the game. I suspect probably used them prior interviews as well. Fire on social media, you get the UK ballcasters mentioning those metrics at half time. It extends out into the wider media space. So like the rest is football of Gary Linnaker and the team, they use the Optifax almost every episode and I say, well, I picked out that game as it happened and those were two brilliant stats because they probably went the most fired at the time. But you see every single game that there's people breaking records, you know, messy breaking records, you're belling and playing at multiple world curbsies. There's almost every game that's some sort of insight and you know, arguably I think Optifax in the past, people probably don't realise that we are pushing those facts. So each individual media partner and yeah, well, Castor will have their own stats teams, but they also feed them facts. So one of our core products is like free match live support and post match facts. So we provide like a fact sheet for commentators, pundits that say, you know, the last time this happened or this, this has never happened in a world cup game and such. So we have to be reactive and as I say, those guys are working globally. So whether it's a forecaster in China, the US, South America, Australia or Europe, but they're kind of all over the place. Yeah, nice. Okay, so if you've made it this far into the video, you must be enjoying the content. Either that or you've simply fallen asleep. So whichever one of those it is, you won't mind me asking you for a quick favour. So if you remember right at the start of the video, I didn't ask you to subscribe to the channel if you haven't already done so. So now would be a good time to do that if you haven't got around to it. So the stats are showing only 30% of regular viewers are subscribed. So there's a hell of a lot of people out there that are watching the videos and they've not yet got around to hit that button, but I'll let you off because you can do it right now in just one second. So as well as subscribing, if you are enjoying this particular episode, then it would be great if you could give it a thumbs up even if you also shared it to a friend, let's say, if you think they would also find it useful. And then if you wanted to get involved by starting a discussion or asking a question, you can also leave a comment down below the video on both YouTube or Spotify. So that's enough for the introduction. Let's get back to the conversation. Now, so another kind of welcome related thing was there. So I think you mentioned it as well. We're recording this the day after Scotland, Brazil. So obviously Scotland finished third up. What do they get? Three points, obviously. So they're waiting to kind of see whether they can qualify as one of the third places and obviously they didn't help concede in three. But then I think it was an opt to tweet maybe. Obviously a way to give them a percentage of being able to qualify as one of those the best place finishes. And I see the last I've seen it was around 24 25%. So I'm just curious to actually do any what kind of what, how is that figure aroused upon a presuming it's just kind of simulating all of the remaining games. And because there's a lot still to happen, but each one of them has so many different outcomes. So it's quite hard to get your head around whether Scotland have a 20% chance or a 50% chance. So yeah, if you can just kind of riff on that a little bit. Yeah, so I guess it's the opt as super computer as it's been famously or infamously named this last year. So it was quite prominent in the Premier League at the end of the season. But yeah, essentially it's a model built on team strengths and then works on the simulations of the remaining tournament. So naturally something like the World Cup, especially with this kind of added element of third place qualification when so many teams go through. It's quite hard to conceptualise all of the different outcomes and permutations of results. So what we do is say for the Scotland predictions, we are simulating the remaining tournament 100,000 times based on the team strengths that we have. So we, depending on what what every other group happens. And we basically take out of those 100,000 simulations, we can see how many of those did Scotland qualify for the next round and they're able to take that as a percentage. So the model works live as well. So we're able to get that after every goal, every record. So as soon as the finishes scored its goal last night, we would be able to update that simulation and see their changing probabilities. So it's a really, it's a simple tool in a way that people just what people are trying to work out. So if you're looking at the table after last night and trying to see what are the best third place finishes you've got think was itself South Africa, South Korea last night as well and that would have damaged Scotland's chances as well. So every single result in every game is going to change and naturally as well those predictions will change almost every day. As more results come in and fewer permutations can happen, we simulate the rest of the tournament. So it's a great way of getting in or there's a single number to tell that story. Yeah, nice. So as I mentioned, this is probably going to go out kind of in a couple of weeks. So even if Scotland make it through, they'll be out by then. So I love that it's not looking good now. All right. Okay, so let's talk some more kind of metrics. I've mentioned a couple as you've been speaking, Johnny, but we'll start just brush over a little bit in terms of one that everyone will know, expected goals in terms of what I wanted to ask you specifically about this, Johnny really is. Well, firstly, can you just kind of recap what the optimal or what calculates expected goal? What goes into it just so people can kind of get up to speed on that? Yeah, so expected goals is measuring the quality of a chance. So up until the moment that the ball is touched for that shot, we're able to account for all the factors that happened. So what type of assist was it? Was it a through ball or cross? Was it headed? Was it shot with the foot? Was it after a counter-attacks and actually probably going to have slightly more space? We're able to determine the positions of teammates and oppositions, the goalkeeper position. So there are huge range of factors. I think probably more than 30 factors power this model. And essentially, we're bringing out one number at the end, which is the probability of that chance being scored. So as they say, it's been hugely popular metric. Everyone I meet that I tell about what I do for a job. That's the first question people have is usually I don't believe in expected goals, can be convinced me, but there is almost every ballcaster life score out everywhere you google a score now. How it tends to have expected goals. So what do you think people get? I'm talking when I say people, I mean, maybe more so than the fan rather than the analyst. Like what do people get wrong? Why do people? Is it does it need re-explaining almost like people? Because I think a lot of people do misunderstand it and they think, well, expect because it does show you on like match in the day I was three tour and I expected we should have won the game kind of thing or so yeah, like what do people get wrong about it basically? Yeah, I think I think people use it as a silver bullet almost. I think it part of the blame goes down to the name. I think expected goals is more is to I don't know to explicit in saying that there should have been a goal or shouldn't have been a goal. And that's not really what it's trying to show. I think people using it in that way, for example, that a team should have won the game. It's not necessarily saying that. Like you could screw a goal in the first minute, defend the rest of the game and naturally you're probably going to have a lower expected goals value because you don't need to go and create chances. For me, it's just another metric in the list of stats. The same way that you would all believe in shots and target and you wouldn't question that metric. It's almost just like another level to that. So it's almost saying like, how good were those chances? If I had 10 shots in the box and you had 10 shots from the halfway line, I don't think any football fan would argue that you had the better chances. And XG is just a way of putting a number on that. So I think people overweight it. I think all it is is saying that that's how good that's how how that's how good those chances were and kind of nothing more. It doesn't necessarily need to imply that they should have won the game. And I think that's where people go wrong. People will treat it as if it's comparable with goals. When I would argue it's more comparable in the kind of contextual metrics of shots and what happened in the game. I think as well because it's kind of each shot is given a value with those values added up and it's like, well, is one sitter from a yard out the same as I don't know 10 shots from 30 yards out. Like, you know, it's kind of when you add them up like that, it does kind of lead to the being some, as you mentioned, kind of misunderstanding really. But yeah, I think the way you describe it into the chance quality, like quality of chances created, that sounds like how it should be described basically. So yeah, that's that's useful. I think with the with a model like that, Johnny, is it something that like, can you build on that model? So like, let's say are you thinking actually we can make this more accurate by adding this bit of data into it or we can we can, like, is it something that does get iterated over the years or is it the same as it has been for or is it, is it got to a point where there's really not much else you can add at this point? I think it can always be improved. I think fundamentally it's measuring the quality of a chance. So if you start collecting more data points that can inform that model, then then we would always, we train the model, we would use the latest data as well. Because that's another thing that 10 years ago, football, as it was, is not football as it is today. So you have to update that training data set to say, you know, the millions of shots that we're using to to calibrate this model. We have to kind of update that as things go along. So I think the biggest thing in the last maybe five plus years is the more of an introduction of tracking data. So naturally, if you have more information on where player positions are, where goalkeeper positions are, the pressure that a player was under as they took that that shot, then you can you can really improve that model. So I think it's this is going to be more improvements in the future. I think it already is a very accurate model. It takes into account most of the key factors. But I think as the raw data gets better, I think you can always improve a model. Yeah. And then another one I want to ask you again, one you've you haven't mentioned a possession value is you say in possession value? Is that the one that you're saying is similar to expected threat effectively? Is that okay? So so just again, for those that don't know what possession value is, can you give a simplified version of that in terms of how it's calculated and what is wise valuable really? Yeah. So possession value expected threat kind of fairly into general names in a way. There what they're essentially doing is is measuring the quality, measuring, looking at every single action and determining how likely that is to increase your team score in a goal. So shot expected goals purely focuses on shots. So how likely to score from that shot, whereas possession value are expected for looks at every single action. So did you carry the ball 50 yards, take it past four players and then suddenly you're in a best position then when you started. And then what we're able to do is say the start of that move, you were less than one percent chance of getting a goal in the next 10 seconds. But by the end of it, you were 60% likely to get a shot or a goal in those next 10 seconds. So we're able to value every single action. So not just those kind of outcome bias kind of end actions, where it's a shot or assist. So it's really nice for evaluating players who are involved more in the buildup or play that kind of pass before the pass. And it's a great way of like, well out of date now, but the kind of shavvy in the Esther roles at Barcelona, where they match, they maybe weren't always top of the chance creation. I mean, it's a do often work, but then being able to value players like that in terms of the kind of buildup chain and saying that actually it was their action. Three passes back the increased the probability of Barcelona scoring by 12%. Yeah. So would that then give us like a score and number to that player? So then would a player finish a game with a possession value score? Is that always a B for each different action? How does it look? Yeah. Both. I think the probably the reason it's not gone so mainstream is quite hard to aggregate that number because you've talked a lot of different actions that increase and decrease by different percentages. So it's more intuitive on this action increase the chance of scoring by 10% whereas if you add 13, 15 of those actions up, the number starts to become harder to explain. And I think this becomes less fan friendly. It's more of a me, sorry, more of a pro club metric. So I'd say most clubs are using metrics like expected for a possession value because you can dive a bit deeper into what a player is doing. And obviously you can split that up as well. So you'd be able to say you added that much value with your passing or you added that much value with your dribbling or you added that much value with your kind of defensive work and stopping the opposition threat. So you're able to kind of measure players in different ways with that metric. Yeah. Yeah, I suppose you could have a player which is creating chances and there are creative player, but then once you actually look down at the numbers, they might be create every time the dribble, they might be rubbish at that. And they might lose that, but every time they actually play a pass, then that's when the better. So you can start to tell to do that more basically. So yeah, that's cool. So I did want to ask you about kind of new metrics, Johnny. So what what goes like, do you sit down in a room and think, actually, we could, this would be, you mentioned a little bit about the kind of momentum where you're in a room. You kind of discussing what, but in terms of like the next expected, whatever, like if there is a new metric, like how does that start? Like is it just an idea that one of the analysts might, and then it's kind of spoke about and kind of tested or. And actually, on the back of that, what is like, what would be the most latest metric that maybe you've, has been created? I don't know if that's a, if you can answer that. Yeah, so I think there are lots of different ways the new metric comes around. So as you say, there is the kind of, you brainstorming sessions that you can come up with these new metrics, but some of it's market led. So it could be clubs. It could be media clients asking for this or somebody asking a question that says, how do I measure what message it's did or how do I measure what's happening in this game? And sometimes there isn't a metric yet. And that's the kind of demand, the demand creates that metric in a way. So people ask in the questions and you're realizing there's a gap somewhere that you need to build out. So it comes from different angles. It then ultimately goes into kind of that brainstorming, these modeling sort of session. And there's always a balance of how complicated do you make it or do you name it? There are lots of different ways you can go. Who's your target market? So if it's clubs, guess the naming is slightly less important than if it was front-facing in the media. I think the most prominent metrics that are new at the moment are the off-ball movements. So I really love this type of framework and there's been a few versions of it in the market as well. Being able to quantify everything a player is doing well, they don't have the ball. And it's obviously huge, huge part of the game. But also it's really intuitive. So being able to talk about a fallback overlapping the wing-air or a strike over and in behind, like I'm not necessarily talking about data when I say those phrases. And I think it's been a really easy one to introduce to people because people understand it from a full concept and then being able to add numbers so that is just part of that process. So obviously in the background there's hugely complicated models and computer vision, like technologies that are building this stuff and each of these runs probably has 10, 15 metrics associated with it in terms of that expected threat, the difficulty of the run, the space creation, the coordinates and all the other elements to it. So there's different levels to that. But I would say that that off-ball movement is definitely the thing that's becoming more and more popular now. And it's something I really like because it's really intuitive. Yeah. Yeah, no, that is really, it's also with the kind of tracking data. I see, you could play it only on the book. I don't know, the data on the book, like obviously off the ball a lot more than they're on the ball. So there's a hell of a lot of untap kind of stuff you can find out there. So yeah, that is about 3% at the time. That's the number that goes around. Yeah, wow. Yeah, so definitely there's loads you could. Yeah. Because it's such a dynamic game in the fact that one run can pull a defender out, which then in turn pulls another defender out and then it moves create space on the other side of the pitch or what I'm saying. So yeah, it's very, very interesting and lots of, it's almost that over and you think about what you could actually look at, you know what I mean? And I feel like in the modern game, like the physicality and the movements are more and more important. It's like pressing and being able to make those movements to drag is just so much more tactical and so much more intense the game now. So those movements are so much more important to the result of the game now. Yeah, definitely. So I'll kind of always kind of really ask people, Johnny, to come on to try and look ahead if you can in terms of is there any kind of things that you would expect to see or where you might see football and things going and maybe also like what are we getting if anything? What are we still getting wrong? So like obviously lots happened from when you first go into football analytics to now, like we've obviously lots of change lots of new metrics, et cetera. We've come a long way, but are we, is there certain areas where you think we're not quite there on this or we could maybe dive and maybe is that off the ball stuff, but like in terms of where you'd like to see it go and what people can maybe look out for looking ahead in football analytics. I think there's a few different things. There's the skeletal tracking data side of things so that is being able to pinpoint your body movements and not just a 2D representation moving around the field, which is kind of tracking data, is it? Yeah. Is in the most part today there is skeletal about, but I think that becoming more, like more prominent, I think you've seen it the World Cup of BBC have a brilliant, brilliant kind of tool on their website where they, you can put yourself in the position of any player in that kind of VR world and see what happens. But I think from a analytics perspective, there's that opens up a lot of like, attention in terms of general kind of movements, kind of off players, being able to look at the body pose, scanning, there's kind of a whole new level of metric that I think that can come out of that space. I think there's always going to be the challenge of what is interesting and what's to be there's going to come out of that and naturally everyone has kind of different movements. But I think there's that, there's the kind of augmented reality side of things in media, being able to like, superimpose those metrics onto the screen as it has a game has been played, kind of overlaying data and statistics. There's kind of another point that probably fits and spoils those brackets with the kind of what if scenarios, so I've talked about decision making that you could almost evaluate an exact moment and say, you know, as this player took a shot here, what were their four-over-options that they could have made the pass or if they took made a pass to this player, could they have also, which could be instead of switched the player over there and we've got models in the company where we're actually simulating plays that didn't happen. So we can actually not just say that, because we have one pass, what happened, we could then play out the next five, ten seconds and say, if they paid it out there, we can predict all the movements of the players and suggest that the most likely outcome was them playing a pass across the goal. So you can kind of, you can see from that almost the simulations that could appear for all that if you did this, if this was a different player in that situation, if you subbed in that player, if you didn't do that and obviously with AI, those possibilities are massive now. And I think the thing with AI, the minute, is that it's probably not necessarily going to be huge, like, unexpected innovations, but I think what AI does is it makes everything faster, it makes everything more accessible. Things that would have been built out in a couple of years can take weeks now. I think some of the basic kind of level analysis is done in seconds. You see people on social media building out the road apps, their own websites, their own interfaces that again would have been six months, 12 months projects that are talking days and weeks. I think that translates into the club space, the media space where people who may be not even used data to that extent so we can now kind of, the floor is lifted in a way that there's a whole level of technical integration that you don't need anymore that it's almost done for you. So I think that flows into every work process in terms of collecting, there'll be more more data points collected quicker and the applications are endless as well. So on AI that joined you think AI will create more or less analysts? Will it be a tool that does replace analysts or will it be a tool that you just allow analysts to do much more and much quicker? What's your thoughts on that? Tough question maybe, I don't know. Yeah, I think it will create less analysts. I think that it shouldn't. I think it should empower analysts to do more. But I think from a company level you probably see analyst dropping. But I suspect there'll be a short term drop followed by a kind of a climb back again because ultimately if you can do more and two of these more and you still going to need these technical people to control some of this work. I guess not. You can social media is a good example of them and it where people are spinning up apps and things but you still need that next level of analysis and application otherwise that there's no point having the best app or visualization in the world if no one's using it and no one's able to show the use cases of it. So yeah, I think the sound reality is short term we might lose some kind of technical roles but I think that'll pick back up again as people realise that there's people are the people who are going to get the most value out of these their kind of hellends and things. Yeah, to be honest I do actually agree, Johnny, so it's like a balance of well we can, hey I can do that role or do 8% but then you think well what could that person do with like if you get more of them people in which they've got the skills and give them AI then they can hell of a lot more can be achieved. So yeah, I kind of agree with that. So just finally then, Johnny, so you've touched, you have mentioned advice for analysts in terms of what should focus on communication skills, storytelling, putting the work out there showing they can do with that. But just finally on kind of the advice aspect is what should people and maybe with AI in the mind, what should people be thinking about now, what should they be learning if they are wanting to get into football analytics and not quite in there yet, what should they be concentrating on now within the next sort of 12 months or so? Yeah, I think for me it's about different opportunities yourself now. So as I say, I suspect if we've put an advert for data analysts next week, we'd probably get 2000 applications but I'd say probably 30% of those would look the same and I think that's the thing, the bars now, everybody can create a good visualisation, everybody can pull in the data, it's now how we can be specific with that, how can you add the application, the analysis, what problems are you solving? I think that's the biggest thing social media now is I see so many graphics or radars or scatter plots that are kind of just put out there and there's no story to them or there's no, like they're not solving a problem, you're showing something, but you're not really saying what that means and what that solves, do you be solving a recruitment problem? Are you telling a story that's going to be great on with a broadcaster? Are you solving the way that you can beat this opposition? And most of the time people aren't doing that, I think the sum is that next level of like, okay, you can do it and the technical bar, a lot of people can do it now with the likes of the LLMs but now prove that you can use it in a way. So there's always going to be that, you know, learn our own Python, tell better stories, that's kind of the known thing now, but I think the next thing is that application stage. Yeah, especially if you're going against 15,000 to 2,000 people, the majority of them will all have the same skills as you so it's not just about what you can do, it's what you can show to stand out, basically see a good, good answer. So that's crazy, getting that many applications for a job as well, by the way, that's a lot of people. So I always ask Johnny, if anyone, when people come on, do they have any recommendations for any books, any podcasts, it doesn't have to be football, it can literally be anything. So anything that, are you a reader, by the way, Johnny, do you have, do you read much or anything at all? I was having to think about this one because I'm not much of a reader. The last book I read was probably Ian Rankin, the John Rebors books, they're kind of like Scottish detective on a holiday last summer maybe, but for me, I would probably point more to like a couple of newsletters that I read regularly, like shorter content. So Mark Thompson, does a brilliant newsletter called Get Goldside. Like I love that one. I've still got a backlog of newsletters to read and I get the time, but another one is Foot Bizz from Ed Mallion, who used to be one of the top people at the Athletic in the UK. That's a great kind of overview of the football world from a more of a top level than the analytics level. So I'd say those two of my kind of go to newsletters at the minute, say sadly not much of a traditional reader, but that's fine. I'll be low, that's interesting. And then I can't let you go without asking you who's going to win, and not who the opt-supercomputer thinks Johnny, who does Johnny think's going to win the world call? For Rant, for me, I think I would have said Spain a few weeks ago, but I'm less going to be, I think the players at France have just unlocked any defence, and I think that's the difference. There's going to be a lot of low blocks, a lot of tight games, and I think it's going to rely on those kind of moments of magic from the forward players, and I think they just have so many options in that sort of thing. So sad to say, France, I think. Yeah, and Bapay for France is an absolute beast in it, at least. Yeah, to be honest, that is one thing that I'm enjoyed about this world, couple of the top players are actually, like, they're all performing, like, Messi and Bapay and Harlan, they're all banging goals, and so it's good to everyone wants to see that. So, perfect, well, Johnny, listen, really appreciate your time. Thanks very much for that. Really interesting. Is anything else you want to give a shout out to before we wrap it up? No, that was brilliant, I just want to say thank you for having me on. It was a lot of fun. Nice, appreciate that. Have a good rest of the day, Johnny, and I'll catch you soon.