Luke Bornn

Luke Bornn: When we last spoke, Toulouse would have just missed promotion on an away goals tie-breaker. And so since then, most of my time has been spent on the running and management of Toulouse, and then a smaller part on sort of Zelus, which is the more technical analytics piece.

And Zelus, we sold to Teamworks last August, so sort of sort of full cycle. And then Toulouse and we were briefly involved in Milan as well. But Toulouse, we also divested from that a few months ago. 

What’s fascinating is that when we last spoke, I was deep in the trenches with those clubs and with Zelus. Now everything we’ve in a very different state now, certainly with the clubs.

The origins of the Zelus Soccer Group, now Teamworks Intelligence Soccer Group, was as the analytics arm for Toulouse. It was essentially Toulouse and AZ Alkmaar’s shared analytics group – and essentially shared because my partner, Billy Beane, was involved with both clubs.

That was the origins of it. Then, at some point, we made the decision to roll that into Zelus, alongside the work that my co-founder, Doug Fearing, had done. It started as our own internal analytics capabilities for Toulouse and Alkmaar and then we packaged it up commercially within Zelus.

Analytics was the driver of everything we did (at Toulouse). We had a lot of success, which certainly made things easier. The first year, which was right before we talked last time, we had really hit our goals financially. We had brought our payroll down drastically. When we inherited the team, we were over 30 million-year payroll. We got to under 10, which means a huge firesale on players.

Only in year two did we really start to get rolling where we won the League, we set a League record for goal scored, and the next year won the League Cup. Truthfully, winning the cup is always maybe more luck than skill, but I’ll take it.

The funny thing is, if you ask the players what the highlight of all that is, they will tell you when we beat Liverpool at home in the Europa League, despite lifting a trophy. 

My role was really at the ownership level. So pre-Toulouse, doing diligence on all the clubs we were looking at, and then bringing in Damien as our Chairman and Sporting Director, bringing in analysts, all that type of stuff, and then really putting in place processes and systems to keep everyone aligned and on track.

We were really fortunate that we had Damien on the ground in Toulouse, obviously an incredibly talented executive, and a huge part of our success, so that certainly made life easy for us there. 

The way I tend to think about it is that Damien really handled everything day to day, so that we didn’t have to think about any of the nuts and bolts. Our focus was really on larger time scales, thinking about squad management and from maybe a monthly or multiple year, long-time horizon.

And certainly Damien was involved at that level as well, but he was the one that would handle all of the day-to-day operations.

It’s not necessarily that the data was overly prescriptive. From the data, there are certain things that we know to be true or observe to be true. And there are other things that, ‘Oh, maybe this is an edge,’ or ‘maybe it’s not.’ And then there are things that are just noise. 

We just really focused on the handful of things that are going to really create advantage and leave the rest up to the coaching staff to give that freedom. The equivalent maybe in basketball. When I used to work in the NBA, you might tell your coach, ‘Hey, we really don’t want to take mid-range shots. These bad long two-point shots, those are really ineffective.’

And then also let’s just make sure our best free throw shooter is the one taking technical fouls. Other than that, you know, do what you want to do. That’s kind of how we operate. It’s, ‘let’s make sure that we’re doing the things that we know to be true, and we know to add value. And then, you know, otherwise we want to make sure that we’re giving people the freedom to do their jobs.

I think every club that uses data uses it differently. If you look in baseball, at the Dodgers or the Tampa Bay Rays, for example, they have analytics staffs that number into the thirties, forties. They’re really exploiting every single possible edge you could get.

At Toulouse, in some sense we were operating at 80% efficiency with the data, essentially realising that we can get a lot of edge by just doing the simple things really well.

If I were to describe what we did, it was really focusing on using best practices with data to make decisions around squad management and the roster and a little bit of playing style.

And then every other area of the organisation is just making sure that we’re following best practices and executing in the leanest way that we possibly can. The part of Toulouse that’s probably less well known is that we were break even or profitable the whole time.

Essentially, what that requires is both efficiency in the transfer market and in your squad management, but also making sure that you’re efficiently deploying resources in the Academy, in the stadium, throughout.

The way we operated is essentially to to figure out, hey, what’s the best practice in each functional unit of the club? And what’s the leanest way we can execute on that? We ran a really lean ship and were able to have the success we did in a really financially sustainable way.

I had very little involvement in the commercial side. I’d be involved in those meetings, maybe, but I shouldn’t take credit for that. We partnered with the folks at Redbird Capital Partners at the time, which is now called Otro Capital, but we essentially bifurcated it such that we ran the football side and they had the the commercial side.

And then we had good people on the ground that did a lot of the legwork. But yeah, I know nothing about ticketing or or sponsorship deals. That is not my area of expertise.

I think it’s natural to think of a club that’s being run, ‘with data’ as being this pure robot. Ultimately, you’re still dealing with humans. Humans are really good at certain things. They’re really good at empathy. They’re really good at the man management piece, the human piece

Where humans struggle is on certain decision-making tasks, particularly when it comes to talent identification and so on. In that case, we know that statistical models are just better, at least in certain cases, certainly for certain types of players with certain amounts of data.

It’s really about what are the areas where data can remove some of those cognitive biases that are inherent in human decision-making. We’re not robo gm’ing this club, right? It’s still humans. 

A lot of it is using the data where the data is good and then using the people where the people are good. And then, as much as possible, having clear processes in place around that, to make sure that you’re continuing down that path.

Because it’s super easy if the statistical model misses on one player to drift back to the old way, or or just become a bit robotic when you’re used to making data too much and realise, no, these are humans; these players, many of them are kids, they’re 18, 19, 20 years old. They’re not ruthless professionals, they’re human beings, and you have to account for that when you’re managing them.

If you look at the history of what makes humans good at things – whether it’s a doctor being good at reading an X-ray – you need a couple of things. You essentially need to have the same thing happen on repeat. If you’re thinking of the doctor example, you need to see X-rays repeatedly, to make a decision and then have really quick feedback on whether you’re correct or not. 

The places where we get good are the things we do on repeat where we get immediate feedback. Think of a musician. They’re practicing their piece of music – I got these guitars here. If I’m practicing a piece and I misplaced my finger slightly, I get immediate feedback. That oh, that note did not ring out, or Hey, I hit the wrong note, or I got the timing. I played it slightly earlier, slightly late.

The reason you can become so proficient at that is you can do it on repeat, in this sort of homogeneous way, where you can play the same things, or very similar things, and with slight variation, and you get immediate feedback on that variation.

In contrast to that, if you think about talent identification in sports, you’re making a decision about hundreds of players, where you’re doing sort of thorough evaluation.

And you’re essentially trying to say, ‘Will this player be good in five years’ for young kids. And so you don’t get feedback in any meaningful amount of time.

So it’s really hard to become good at this. And if you think, ‘Oh, but hold on. I actually, I saw a bunch of these players that are good now five years ago.’ The problem is that we humans inherently like misremember their own past right.

We tend to have a lot of hindsight bias. We tend to think that we were better decision makers than we were, we tend to think we were smarter than we were at the time, and that leads to biases and overconfidence in our own abilities. 

Statistical models don’t have those problems. There’s a reason that bookies these days rely on statistical models. They’re not using humans to say, ‘Oh, yeah, you know, Liverpool’s going to beat Manchester United by this much.’ That’s not at all what they’re doing right. They know these statistical models are just better.

Damien kind of got the rough end of this from a handful of his transfers that he did in Liverpool, where early on somebody saw that they didn’t work out, or whatever. But if you look in hindsight at the set of decisions that he made, you know, had a great long-term impact on the team and set them up for success in the future.

So the timescales are challenging. I had dinner last night with some of the staff from the Cleveland Guardians, the baseball team, and they were saying for the last decade they have overperformed their payroll.

But, despite that, there’s still uncertainty over whether they’re actually a good front office or not. You’re operating on long time horizons and in football you’re really talking about bringing in five, eight, 10 players a season.

There is just a lot of randomness in those outcomes. You could have an incredible Sporting Director come in and and miss on two or three high-profile players and they could still be a great Sporting Director. 

It’s really about saying, ‘Hey, can we really look at the process that they’re using, and how they’re making decisions, rather than the actual decisions.’ So if they’re really using best practices and making decisions in a really sound way, in a repeatable way. 

Then, yeah, they’re going to have some misses due to randomness, but I would rather have that than someone who’s operating off the cuff. I’d rather go for the thing that that we know to be repeatable.

I’m sure you can think of tons of examples where people say, ‘This guy’s the greatest coach’, and then he gets poached to a big team and his old club brings in a new coach and the old team does just as well. 

I think we have this real tendency to attribute all success and failure to the coach. It’s actually more common in England than elsewhere in Europe. I’m not sure why. The coach certainly matters, but you have 11 players out on the pitch, and for the most part, once the kick-off happens, the coach has very little influence. 

He might scream as much as he can from the sidelines, but there’s very little that they can influence. So it’s, you know, it’s largely the players at that point. So I’m not sure why. Maybe it’s because they’re the most visible, maybe it’s because they’re the one that does the press conference afterwards.

But we tend to attribute a disproportionate amount of good or bad to the coaches. In reality. It’s there’s a lot more that comes from player skill, and so on than I think we oftentimes get at.

That’s why it’s challenging to identify coaches, to find good ones. It’s very hard to find coaches that add value, predictively, to your club, because the biggest driver of a team is player skill.

So it can be hard to disentangle that. And to have enough data or information to know a coach is really good, you need to have seen them win with multiple different unique sets of players.

You need to basically go saying, ‘Hey, this guy had success with with one squad and then a totally different squad three years later and a totally different squad.’ He can maintain success, right? And by the time that happens, they become unattainable financially. And so it’s a really interesting challenge. 

We went through a couple of coaches at Toulouse and our focus was really on going back to process versus results, finding someone who’s really aligned with our way of thinking. 

A lot of coaches will say, ‘I play a very particular style and I want to recruit to that style.’ And there’s nothing wrong with that approach, that’s a super valid way to go about things. 

But the way that we did recruitment, our approach is very different. We say, ‘Hey, we’re going to work to build a squad that is balanced, but not specifically focused on a particular style.’

We might play a very ball possession on the ground style, but if all of a sudden we have an opportunity to acquire an incredible target man at great value, then we might want to alter our style a little bi. And we make sure that we had a coach that had that tactical flexibility. 

So instead of saying, ‘Hey, here’s my rigid tactics, and we need to get the specific exact players for this tactical system,’ it’s saying, ‘Here’s a bucket of player and it’s then up to the coach to decide what’s the best tactical system to execute on that. 

We’re not giving you 20 left backs, we’re giving you a balanced squad. That requires a lot more flexibility and tactical intelligence. And that’s how we thought about coach recruitment – as well as a coach that’s going to be easy to work with going to work within the club system.

You’ve got to hire someone with tactical flexibility. So, if you bring in someone who’s only ever played a 4-4-2 and happen to bring them into a system where you have great attacking wing-backs, and they force them into a defensive full-back style, so you need someone that really has that tactical flexibility. 

There are certainly pros to recruiting to a specific style – you can really get that great alignment. But there are two big flaws. 

When you do that, you’re constraining your possible squad. Imagine you’re playing a very particular style and you say, ‘I need players of very particular archetypes that fit within that style.’

As a result, you’re drastically reducing set of players that you’re interested in. And that means, if, for whatever reason, those players are really expensive relative to their contributions, you essentially are getting really poor value.

Given that you have a fixed budget, you’re actually going to get lower quality players than if you maybe got players that are outside of that tactical system.

For a given budget, you might be in a situation where you have to decide, ‘Would I rather have worse players that fit the system perfectly, or better players that are slight mismatch?’

And that’s essentially the decision that’s going on. 

The second piece of why this is problematic – and you see this over and over again at the big clubs – is where they recruit to a given manager’s style and then they get rid of the manager and bring in a new manager with a different style. That manager wants to change things drastically. 

This is not the coach’s fault, right? It’s the club’s fault for bringing in managers, on repeat, with different styles and different types of players. Then you end up with these huge budgets. That’s obviously a recipe for failure.

If you go back to my experience years and years ago, in the NBA, there was a time when – actually it’s still true now, largely – that three-point shooters like Steph Curry, let’s say 2015 to 2020, there wasn’t the talent there. Those positions became extremely expensive.

Having shooters was very, very expensive. In terms of return on investment, you were often better saying, ‘Okay, let’s just have a mediocre shooter, and let’s put other players around them.’

So I heard a story in baseball where you know, going into the trade deadline and saying, we we are the weakest team in the League in our relief pitching. So this is like, you know, the pitchers that come in after the starting pitcher.

Sometimes it’s better to improve on your strengths than try and tackle those weaknesses. It’s really a function of what’s the return on that investment?

From a fan, that’s really hard to understand, because you just see the end player come in and you think, ‘Hold on! They have the worst left back in the League, why in the world did they go buy a right winger? This makes no sense.

Let’s say they were going to spend an extra five million. It’s possible that they they made the decision that that five million was was better deployed in terms of improvement of the team at that position, taking them from having an average right wing to a great right winger. That was a better return on investment than going from a bad left back to a mediocre left back, for example.

Certainly, if you look right now, everyone has sort of copied the the pep and Klopp style. Football’s become a little hard to watch.

And so the teams that really deviate from that like… what comes to mind is the Barnnsley team from a couple of years ago, that played the craziest long bell style I think I’ve ever seen, and had a lot of success with it. Zagging when others are zagging is valuable, but only in so much as when you’re going in a different direction, not in ways that are less efficient.

As an example, every team is moving towards taking more 3 point shots. If you say, Oh, we’re going to not take 3 point shots to be differentiated. The problem is that’s a very inefficient strategy, not taking 3 point shots.

It’d be like a a football club saying, No one takes short corners anymore, we’re gonna start taking short corners again. Well, that’s a bad idea. Short corners are very inefficient.

Where you want to be orthogonal to others, in terms of being differentiated – you just want to do it in a way which isn’t making you worse.

Actually be one of the reasons why clubs that get promoted can struggle, because they often have a squad that’s built to dominate in the Lower League, and then they go up, and all of a sudden those players they’re attacking players thathad the ball constantly in their lower league. Now never touch the ball.

I never had a formal role at Alkmaar. They worked with us at Zelus again, Teamworks, and Billy Beane was always an advisor there. My role was I became close to them, almost like a colleague and friend.

But I had no management role there, even though I got a lot of respect for those guys. I think over the last decade, it’s probably been the best-run team in Europe, in terms of how they’re operating relative to their resources.

It’s an incredibly well-run club. Robert Eenhorn was the CEO there for a decade, he actually just departed and thought it’s time for a new challenge, but I have incredible respect for what they’ve accomplished there, where they’ve punched way above their weight. They’ve done it in a really financially responsible way, continually making decisions in a really process-driven efficient manner. A really, really impressive club.

They placed really strategic bets and were really focused on the handful of things that matter. One of the things that clubs do wrong is just do way too many things. They’ll spend time and resources on the latest sports science gizmo and gadget and they’ll go down some rabbit hole of a newfangled sleep thing. What they (Alkmaar) did well is really focus ruthlessly on the core things that drive value and then very selectively invest in innovation.

I think this is actually true across sports. I think teams tend to do too much. They tend to be distracted, because the truth is, most of your value comes in terms of building a team.  I’m really thinking from the Sporting Director perspective – there’s probably 10 to 15 decisions in a year that drive 95% of the value. 

And maybe 10 of those are things around recruitment decisions. The other five are maybe major decisions around the tactical style you’re going to push, you’re gonna stress, or whatever it might be.

And so if you’re getting distracted with all these other things, It can be be a detriment to those bigger decisions. When I was at the Sacramento Kings, I remember them making a decision around a free agent that was a hundred million dollar decision they made in 30 minutes.

Then another time spending four weeks on what was essentially a fringe player who played for the team for like 2 weeks. I think this is backwards.

Making the main thing the main thing, right? There’s this other more subtle problem there that you get into too. From a recruitment perspective, you’ve got your analytics group and your player evaluations. You’ve got these great statistical models valuing players – awesome. 

Then we have our medical staff and our psychologist, and we maybe bring in these outside people to assess a player’s cognitive capacity – whatever.

Then you start doing all these different things and bring everyone inoto the room. And you say, ‘Let’s make a decision.’

I’ve seen this multiple times, where you might have your analytics group and your scouting group represented by one person and the analytics person in the room might represent a team of 10 or, if you’re using someone like Zelus, you’re representing dozens of people, work that’s been heavily back tested, that’s been used and validated over a decade.

Maybe you have a head of scouting there that represents a scouting group of dozens, synthesising their knowledge. And then you have a psychologist, who asks these players 20 questions and is making a judgment from there and the Sporting Director is giving sort of equal weight to all of these things. 

These things can actually sabotage the decision-making process, where you miss-weight them, where you’re putting too much importance on these unvalidated, untested ideas.

This happens a lot in sports science – a ton in sports science! We know the things that really matter are load management, which, if I were to sign really briefly would be, ‘Make sure the players work really hard on Wednesday and barely move at all on a Friday.’

I’m exaggerating a bit, but that simple weekly periodisation has been really shown to be effective.

And if you’re if you then say, ‘Okay, we’re going to try this new machine where we’re looking at creating ankle stability. And we’re going to try this other gadget and this new fancy workout,’ and whatever

The real limit is player time and attention. Any time you’re doing these unproven, untested things, you’re taking time and attention away from the things that  create competitive advantage. 

That’s why, as an organisation, you need to be ruthless about paring it down and really focusing on keeping the main thing the main thing.

Teah, he’s certainly been involved in the ownership group while he was also running the A’s, so it was a part-time thing. He’s been interested in football for 10 years, maybe 15. 

I think fell in love with the game years and years ago and certainly played a very strategic role over the last few years. There’s all these things that go on on an ownership level that aren’t really top of mind for the media, for the fans, things like investor relations and budgeting and all that kind of stuff. 

A lot of that happens behind the scenes and he added an incredible amount of value in all those areas.

Zelus has sort of been running in parallel to this. It started off as a spin off of Toulouse’s analytics group in a way. At the time we merged in with with Teamworks, back in August, we were about 75 employees with a pretty broad coverage across all the major sports. So a big group and basketball, baseball, American football, cricket etc.

After that acquisition, I continue to be an advisor there. I actually like really love what they do, I love their vision, I just think they’re solving a real problem in sports. 

I think there are a lot of companies on the periphery of sports that are just leeches and I but I think Teamworks is actually creating a lot of positive value for clubs and for the ecosystem. Plus, I really like the people, so I’ve been really happy to stay involved there.

If you think about it from the club perspective, clubs should not be software development companies, they should not have to build out their own information systems and their own messaging platforms and their own athlete management tools and their own data engineering back-ends and their own sort of core data infrastructure, and so on. 

Functionally, every team is doing the same thing on repeat. Clubs don’t get value from having better data pipelines. They don’t get better competitive advantage from having yet another expected goals model or some possession value model. 

They don’t get the value from building out these foundational things that everyone has. Where they get value is on the execution piece, is using this stuff to make better decisions and ultimately win on the pitch. 

If you look at the ecosystem 10, 15 years ago, maybe even further back, the sports scientists were using Excel to manage this data. And at some point we realised this is a huge time and effort, let’s use some of these athlete management systems. Smartabase came along, and it was like, yeah, this solves a lot of problems? 

And for a tiny fraction of the amount of time and money that we’re spending building internally, we just get this great tool, and we can just spend our time on what matters, which is on the execution side of it. That’s a very small example, but that applies much more broadly to making the main thing the main thing.

If you’re spending your time thinking about data engineering decisions and the micro nuances of your communication platform and how your video is transmitted and stuff, you’re not doing the things where you’re really adding value. And so where teamworks comes in is to say, ‘Hey, look, we, we’re going to do all these things.’

I’ll build all this software, this sort of operating system ecosystem for for what clubs need. And it’s messaging, it’s athlete management, it’s nutrition monitoring, it’s analytics. It sort of does it all in a way which gets you 80 or 90% of the way there for what every club wants or needs, all the sort of foundational core things that every team needs and and more.

And really let you focus on the communication, the integration, the execution, which is really where the value creation is. So, instead of spending your time again recreating the wheel, it sort of says they say, here’s the wheel. Now do what you really do, which is sort of get it rolling.

When we were Zelus, we had this very exclusive model of just working with a couple of teams. And it’s still in some sense self-constraining, because the thing of what we’ve done at Zelus we’ve really made sure that we are the best of the best.

So we have the biggest football analytics group in the world and I firmly believe we have the best data engineering infrastructure, we have the best models, the best player valuations, the best player projections, the best translations across leagues, understanding how players move up and down as they move up and down the pyramid – all these kinds of things.

And that’s been developed over decades, well, over 100-person years of work has gone into this, roughly, 10 people for 10 years is the ballpark.

Because of this, it’s expensive to build and because we’re at the very top of the ecosystem, there’s a cost that prices out clubs. For the majority of clubs, where you’re just like, ‘Oh, I just want to have a little bit of analytics, and I want to sort of dabble. I’m not going to make it a key part of what I do.’

You shouldn’t be spending all that much on analytics. Right? Go get like an inexpensive data set and back to making the main thing the main thing. 

If you don’t have a clear strategy on how you’re going to use data and analytics to make better decisions, don’t waste money on it. Spend your money where you know you can get value.

If you know how to get value from an additional physiotherapist, get another physiotherapist. Don’t spend a bunch of money on data and an analyst that you’re going to stick in the corner and ignore. That’s a total waste of time and money, right? 

But conversely, if you’re a club, that sort of is making data a core part of your decision-making process and making it be a driver of the way you make decisions, you want the best of the best. So I think it’s inherently limited, because the number of clubs that fall into that latter bucket is very small still.

Most clubs that are doing data analytics, they have an analytics person, but stick them in a corner and treat them like a houseplant. They often have them just because the owner says, ‘Oh, we need to do Moneyball.’

They’re not making use of them in any meaningful way. So, for those clubs, you shouldn’t be wasting resources and time. But for the handful of clubs that are really using it as a driver of competitive advantage, then the Zelus-Teamworks model starts to really make sense.

One of the things that I think clubs do wrong is they think of their performance group and their analytics group ia different. The best clubs have these things fully integrated. 

If you want to be in a spot where you can very naturally go from, ‘Hey, here’s how this player’s fatigue is impacting their performance on the pitch,’ or how their athletic measures on the pitch are impacting their decision-making on the ball, for that, you need full integration of these two groups. So I think it can have broad impact. 

But, again, I think a lot of clubs go in and say, ‘We need analytics to be smarter,’ without a clear directive on exactly how it’s going to make them smarter. 

Ultimately, if you’re not changing decisions or improving decisions, you’re not doing anything. Really, the club should think about what are the set of decision points that we’re going to improve and how are we going to improve them? 

Maybe it’s, ‘We’re going to be more efficient in the transfer market and have lower fail rate on our transfers in.’ And we’re going to be smarter about contract management.

A lot of clubs don’t even have ways to visualise and see their contracts. The fact you can’t at one glance see how all of your players are doing, how much time is left on their contracts, the options, all these types of things in a really clear simple visual way. That should be table stakes. But it’s for some reason it’s not. 

In terms of analytics impact, I think there is a chance for for broad impact. Teams should probably focus on where they know they can actually make that impact or where they can actually create the opportunity to for data to have that impact. 

The other thing I will say is that there’s certain areas where the data just doesn’t exist, or where it is not good enough. In football, we have very limited financial data. A lot of the data, even around biokinematics and so on, is very limited or unclear. At this moment we don’t get good data or any data at the lower tiers and young kids, so it can be really hard to sort of look at the projection of youth.

Maybe long term, it’s absolutely reasonable to think we can create value in our Academy from data, but for the most part, at the moment, it’s pretty limited.

The way Teamworks Intelligence – the analytics part of Teamworks – does across all the sports is this realisation that if you buy data in football – StatsBomb, Opta, Skill Corner and others. A lot of times slubs get this data and then they realise, ‘Oh, the raw data is almost useless in and of itself.’

So then I’ve got to hire a bunch of data scientists to build a bunch of things on top of it. The raw data in and of itself is very limited utility.

This is why tracking data for the longest time they did things like counting sprints and how far did a player go? These things are useless, but it’s easy to calculate. It’s what comes out of the raw data. 

For the longest time, the analytical layer on top of this tracking data has just been, ‘Let’s just count how far a player went,’ which is the simplest thing you can do, and not terribly insightful.

But, if all of a sudden, you have this whole suite of statistical machine learning models that sit on top of this, you can actually start to turn all this raw data, this tremendous amount of data that’s being collected now, whether it’s it’s event tracking data, kinematic data, financial data, etc, pulling that all together and creating actionable intelligence and insights from it. 

Before you had simple raw data. Now you have player valuations, breakdowns of player skill to say, instead of just, ‘This player passes from here to here,’  to ‘This player is really good at this type of pass, they’re a really good decision maker. They make the right decision. When they pass, they execute on those passes, they’re really efficient at X, Y and Z and they’re very inefficient at these other things. 

We’re not a data collector. Our whole ethos is taking that raw data and taking the models that we’ve built with this huge team over the last decade to turn it into really actionable intelligence.

If I were to boil it down to the absolute, simplest thing, I would say, ‘We take this data and we tell you which players are good and which players are bad.’

The AI piece is interesting. Mostly when people think about AI these days, we’re thinking about large language models, fundamentally around text. And I think there have been some interesting things there, like Twelve has used it, where they’re essentially taking standard analytical models and outputs and then using these tools to turn it into like interpretable human language, which is really good, because we spend a ton of time thinking about data visualisation, but it’s so much more natural for a human to just read, ‘this player is really good at this and has these attributes’ rather than  trying to extract that out from some data visualisation or table of numbers. 

So on the AI piece, that’s sort of where we are now. This idea of artificial intelligence running a team, we are a long way away from that.

But the machine learning piece, that is essentially what we do and what a lot of these sort of analytics companies are doing – building models, ideally in an interpretable and decomposable way, so you can understand why the model says what it says. 

People talk about the future. I think in a way the future is there. Where we hopefully see some evolution is on the execution side. If you looked at the clubs that are using data heavily – the Liverpools of the world and the Brentfords and the Brightons – I bet if you ask them, would they be better off keep making great decisions with worse data or having better data but a worse process, they would choose the better process and worse data all day long. 

And so we’re at a point now, like at Teamworks, where we have these incredible models, and our partners who are executing well on them are having a tremendous amount of success. But you can also see there are clubs out there that are spending a tremendous amount of data that are not, and it comes down to just not having the processes in place to use these, or in some cases just completely ignoring them, having big analytics experiments and then just completely ignoring them. 

So if I were to say what’s going to be the biggest evolution, I actually am not that fascinated about the technical evolution, because I think we’re in a spot now where on a technical perspective, we’re so far beyond what teams are actually executing on. 

The thing I hope to see is a big change in how teams execute on that data. And that’s going to take a while, honestly, because a lot of the folks that are running football clubs right now don’t have that expertise. They don’t know how to use an analytics department, the output of analytics department to improve their decision making, so it’s probably going to require a combination of, you know, changing the personnel, upskilling people that are currently in the jobs, to really understand this, to be able to think probabilistically.

There’s not an MLB front office in the world that doesn’t think probabilistically about every decision and that hasn’t read Thinking Fast and Slow to deeply understand cognitive biases and behavioural economics.

I could probably count on one hand the amount of football clubs in the world that truly think probabilistically and could tell you what hindsight bias is or outcome bias is.

We’re a long way away from front offices having those skills. If I were to say the one thing it would be that – having the horsepower in the front office to actually execute on these things.

I think there are going to be a lot of teams out there that spend a lot of money on this data and then get no value from it. But that said, there will probably one or two who put in the effort and have the systems in place and will actually get a tremendous amount. 

I think the technical evolution is multiple steps ahead of where the clubs are, but it is changing. What I tend to see is, it doesn’t happen as a change where everyone’s just getting a little bit smarter all the time. What tends to happen is a given club, maybe has a new owner, or whatever, and starts from scratch and goes heavy into it. 

The way this evolution is happening, it’s not like everyone’s getting a little smarter every year. It tends to be an entire club flips to being smarter and more data driven.

A good example is Como, where the previous ownership had no data analytics at all. New ownership comes in, they invest heavily in analytics, and all of a sudden they’re making tons of smart decisions. So it’s individual clubs taking that step and in the aggregate it looks like we’re getting slowly more analytical. But in reality it’s individual clubs taking that step. And the next year, maybe another club has that evolution, and that’s how it tends to look.

A big step now is to focus on that decision making, as you say. 

Those people can add a lot of value, especially if they’re empowered, where the Sporting Director feels strongly that person is is going to be a key part of our decision making and drive decision making. Or maybe you have an owner that says, ‘Look, this is critical, and we’re not gonna make any decision without this,’ then that can be a good way to go, for sure.

I really like what Teamworks is doing. I like the I like the product. I like the people I like the vision. So, planning to continue on working with them, helping them execute on that vision in the years to come.

We’ve been out of Toulouse for about the last 4 or 5 months. We’ve been out of Milan for over a year. Looking at going back and doing this again. I think we had a lot of success at Toulouse, and it was a lot of fun. Take this to the next level, replicate it, and ultimately take it one step further.