Can you really win the Premier League with analytics?

One book claims to have the answer

Hey, everyone! This a pretty nerdy look into football analytics over the last decade and change. I know some of you love that stuff and read the newsletter for it. But if you’re craving the exact opposite kind of Grace Robertson football writing, I wrote something for Versus’ collaboration with Gay Times on writing about the sport and being transgender, and I hope people check it out. Anyway, onto the meat of the newsletter.

“We know by now this is not a very important metric, but some teams still pay attention to it”

Book review!

I think I’ve done this once before, but then this book came along that demanded to be discussed. Jürgen Klopp has now left Liverpool and, with Pep Guardiola rumoured to be nearing the end of his Manchester City stay, we’re coming to the end of a Premier League era. Few are better placed to understand the Liverpool side of that story than the author of this book. I do want to review the other major new football book, The Pep Revolution by Martí Perarnau, some time soon to understand how things played out for City, maybe in a month or so? We’ll see how it goes.

How to Win the Premier League by Ian Graham, Liverpool’s Director of Research from 2012-23, has a title that wants to grab you. Its subtitle “The inside story of football’s data revolution” is probably more apt, as Graham is upfront about his role in all of this. “Building a successful football club is a team effort”, Graham explains early on in the book. “My data analysis team increased our chances of success and identified future stars for Liverpool but the theoretical edge it gave only became a reality thanks to the hard work and talent of our owners and my colleagues”.

If there’s a single lesson to take away from this book, it’s that buy in is everything. He starts by talking of his time at Tottenham Hotspur, where he found it harder and harder to have an impact as Harry Redknapp took more control of transfer decisions, then onto his early years at Liverpool, when Brendan Rodgers was in the dugout. It’s probably not a huge shock that Redknapp isn’t a data evangelist, and Rodgers’ clashes with the “transfer committee” are no secret. A big part of this book’s value is that it gives concrete proof to rumours and whispers we heard throughout Graham’s time at Anfield. If you were paying close attention to the rumour mill during this era, though, much of the book won’t surprise you.

What might surprise you is someone’s data scepticism. “All this data analysis in football is bullshit, isn’t it?”, Graham was told by Michael Edwards on the very first day they met. Yes, Michael Edwards, the “laptop guru” who you probably know for championing a data-led approach as Liverpool’s Sporting Director. Edwards wasn’t convinced by the very basic data that Graham had access to in 2009 and wasn’t shy about challenging the validity of those numbers. Edwards comes across incredibly well in the book as someone who can bring together lots of disciplines and act as the ultimate arbiter. Graham and Edwards are obviously friends, so this is not a neutral perspective, but it’s about as ringing an endorsement for Liverpool’s current CEO of Football as you’ll ever find.

Graham is much more positive about the Jürgen Klopp era, where the manager bought more readily into a Sporting Director structure with greater emphasis placed on analytics. He talks at length about all the things that went right in this era, the crucial players signed and the pieces of good fortune that went Liverpool’s way. None of this is news, exactly, but it’s nice to see how these things can work properly.

And these things aren’t wildly out of line with public analysis at the time. Anyone with a modicum of analytics knowledge could’ve told Liverpool to make some of these decisions. I know this because I can’t code, dropped maths at school when I was 16, and did my undergraduate degree in film studies, and when Liverpool were rumoured to want a winger in early 2017, I remember thinking “why don’t they just go back for Mohamed Salah? His stats are amazing!” James Yorke at StatsBomb wrote an article raving about the deal at the time. A lot of what Liverpool did was very much in line with what the stats nerds online were saying back then, and I think the community can take a degree of credit in that.

It’s astounding, though, how few football clubs were listening. We could sit here all day naming big money transfers that were obviously bad deals if you paid attention to the numbers, but football as a whole did not. They’re far from the only offenders here, but Graham cites Barcelona as an example. The Barça Innovation Hub is cutting edge when it comes to the quality of research being done, and has been for many years now. “We help seekers build the future of sport”, says the website. Just not the future of sport inside FC Barcelona itself. Barça have some of the smartest data scientists in the game and put essentially none of it to use at their own club. The conferences and research papers might make for a bit of good PR, but that same research team could’ve told the club not to sign Philippe Coutinho or Antoine Griezmann. Instead, they were completely ignored when it actually mattered, as is the case at most clubs. Graham and his colleagues weren’t doing impossible things. They were simply the people who got listened to.

The most important skill, then, far more so than the quality of the data or modelling, is knowing how to talk to “football people”. Graham was able to build a rapport with Klopp because they got off on the right foot. We’ve heard the story many times now of that first meeting, where Graham pointed out how much Borussia Dortmund underperformed their xG in Klopp’s final season and the manager enthusiastically nodded along. “All that stuff where you kiss his arse”, Graham’s girlfriend told him of his planned presentation, “put that at the start”. Graham had the buy-in: Edwards and Fenway Sports Group were certainly game, while he found a way to talk to the Proper Football Men. Many of his smart colleagues at other clubs just didn’t have that.

But I do have some quibbles with the book. The first is: what, if anything, went wrong in the end? Graham’s account largely ends with Liverpool winning the Premier League title in 2020, three years before he left the club. Reports claim that Edwards and Klopp’s relationship soured in that time, and that the recruitment team disliked certain players bought who Klopp pushed for, with Thiago Alcântara and Darwin Núñez thought to be the biggest flashpoints. Since the book confirms how much reporting from 2012-2020 was accurate, we at least have to consider that this information is also true. Graham doesn’t mention any of it. What did he think during those later years, when Edwards decided to leave and Klopp seemed to take a greater lead on transfers? What happened there? He won’t tell us.

Another question is more important to the field of analytics. Graham loved Naby Keïta and pushed aggressively for Liverpool to sign him. “When he played”, Graham claims, “Keïta played to the level that I hoped he would. He just didn’t play very much”. A lot of this is down to injury problems, but that’s not the whole story. There was no point during Keïta’s five years at Anfield when, if everyone was fit, he would be first choice for a crucial game against a top opponent. He was, in Graham’s telling, “happy to risk losing possession in order to create a scoring chance, but our coaches prioritised the midfield’s defensive contributions over their attacking ones”.

If I could ask Graham anything, it would be this: was Klopp wrong to leave Keïta on the bench? Would Liverpool have won more games if the Guinean started over Jordan Henderson or Gini Wijnaldum? Graham talks a lot about managers being far too risk-averse, putting too much emphasis on defence over attack. He loved watching Keïta “risk possession again and again by attempting a killer pass”, which he says “maximised our chances of success”. Klopp clearly did not love it. Keïta often frustrated me with how easily he could lose the ball in midfield, and I assume Klopp felt the same way, prioritising conservative passers in that part of the pitch. Graham seems to think that’s wrong.

In reality, Graham was in no position to tell Klopp which team to pick. A German manager might accept others making transfer decisions, but not picking the lineup and tactics. Most of Graham’s work seemed to centre on transfers. As he says in the book, and as is common sense, by far the most important thing for winning football matches is having the best players. So this is pretty important. But is analytics so recruitment-focused because that’s where it makes the biggest difference, or because that’s the easier part of a football club to penetrate?

Klopp isn’t a data expert. He famously said that “no playmaker in the world can be as good as a good gegenpressing situation”, but he doesn’t know that. He hasn’t run the numbers on it and found it to be statistically significant. He just thinks it’s true because he watched lots of football matches where it appeared to be the case. It’s a hunch. Pep Guardiola will tell you that Manchester City play a certain way because that’s the style he grew up watching, and thus it’s what he “feels”. I can’t think of a single manager who plays a certain style of football because the data said so. Graham spends a lot of time in the book talking through his numbers-backed theories about football. Were Liverpool wasting huge edges not exploiting this? If analytics can simply help teams make better transfers, then Graham made a mistake advocating that Liverpool sign a midfielder who takes lots of risks in possession. But I’m pretty sure he doesn’t think that’s all it can do, and it’s a tension that the book only implies.

I certainly don’t say this to criticise Graham. I think he’s done brilliant work in his field, and his book was a fascinating read. But as I read along, I did start to wonder: to what end is all this work being done? He frames analytics as an edge that clubs of lesser resources can exploit. “Smart teams like Brentford, Brighton and Liverpool have proven that when competing against the ever-greater spending power of petrochemical football”, he tells us, “innovation and intelligence can be used to fight back”.

Now, obviously Liverpool have a lot more money than Brentford or Brighton. Their story is of one of England’s richest clubs briefly out-competing sides who spent slightly more. We’re not talking about Leicester here. But even if you agree with Graham, what’s the end game? Let’s say a bunch of smart teams keep massively overperforming, then what happens? Everyone will heavily embrace analytics. Hypothetically, let’s say all the clubs in the land fully adopt the methods Graham advocates here. So who’s going to win the league, then? The team with the most money, stupid. It’s an edge right now because teams are leaving value on the table. But if everyone exploits it, analytics will mean football gets decided by money more, not less.

During the period this book covers, there was a general push from the “analytics community” to see more people adopt its methods. It was considered a win when clubs were making smart data decisions. In the last couple of years, I think I’ve realised that I just don’t care about that anymore. I’m interested in analytics for the fun of understanding the game better, not to see clubs get more efficient at turning money into points and help John Henry with his long-term vision of suppressing player wages. Even as I genuinely find the field fascinating, I have no investment in seeing football clubs get smarter about analytics. It just makes it harder and more opaque to understand what was going on, which is why football data was more fun in the 2010s.

I really did like reading How to Win the Premier League a lot. It’s a book that covers a lot of what we already knew, but told by an absolute authority in the field. I think it’s a must-read if you’re interested in this stuff. It just raises some questions about football analytics in my own head, and I’m not sure I like all the answers.