10.5 years ago I started writing about football stats while recovering from chemotherapy I had as a result of contracting testicular cancer. I never worked in my mom’s basement, but I do only have one testicle now. It has not further impacted my professional or private life.
10 years ago, I published a blog on a website named www.statsbomb.com. It was a website founded to host work from myself and those I admired doing smarter sports analysis.
Today, StatsBomb is one of the world’s biggest and best sports data companies. We collect 800 games a week during peak season, with customers from Peru to Japan. From its inception, it was a mission - first to educate the football world on what they were missing by not adopting data-based analysis into the sport. And later to develop the data and tools for practitioners to do smarter sports analysis, because the incumbents in the space were (mostly) useless.
It was founded by me, but it’s taken the work of literally thousands of people to make it what it is.
Ten years is a long time. There are a lot of stories to tell.
I better get started.
The Origin Story
Once upon a time, I was a professional sports bettor. I also used to work at a place called PinnacleSports.com, which is now just called Pinnacle. When I joined them at the start of 2007, they had just cut off all their U.S. sports betting business - 70% of their total revenue - in a single day. A man named “Henry” hired a friend and I to develop products to turn their primarily U.S.-facing business into one that the rest of the world wanted to bet at. So we did.
The first time I met one of Pinnacle’s owners, who was actually a former drug dealer and bookie to Hollywood’s brightest stars of the 70s and 80s, was on the island of Cura-
Hold on, hold on… this is the wrong set of stories. I need to fast forward a bit to the relevant era at hand.
So I was working at Pinnacle, but I was diagnosed with testicular cancer, and went on leave until I recovered from surgery and chemotherapy. While miserable and sat at home, I started digging into player stats for soccer/football. I had wanted to do this years earlier, but widely available stats were impossible to find. It was only after WhoScored and Squawka made them more readily available that there was finally enough public data to work with.
In a far earlier era of my life, I was paid as a professional writer. Bored out of my mind, I desperately wanted something to focus on besides my own pain and potential mortality, so I started writing about and publishing my work on stats in football. At first it was a Manchester City website called Bitter and Blue, which was run by Danny Pugsley and where Benjamin Pugsley published his work.
I loved Ben’s stuff. I loved his communication style on Twitter. I loved his humour. I just loved Ben, really, and I hadn’t even heard his gloriously Manc accent yet. Ben is good people.
Back in these days, there were no centralised places to publish analytical football writing. We literally only encountered people’s work when someone told us about it on social media or you got lucky during a Google search. This didn’t make any sense to me - I loved reading new analysis, and wanted to create a centralised hub that made it easy to find the best of it.
I wasn’t always going to be writing about Manchester City with my data analysis, so I needed somewhere else I could publish what had suddenly become a huge passion of mine. So… a website. With a name, and shit.
Like this one.
About the same time I started writing about football stats, I bought an electric guitar. I have always wanted to learn to play guitar for pretty much my entire life, and it seemed like a useful hobby to pickup post-chemo. Years later, my cleaner noticed a very dusty guitar case sitting in the corner.
“Did you ever take that thing out of its box?”
“For about two weeks.”
“What a waste.”
“Eh… that’s one perspective. The other is that I started obsessing about football stats instead, and it got me a job with two football teams and caused me to found StatsBomb.”
Oh god… a name,
I have taken both kudos and endless amounts of flak for “StatsBomb” since day 1. The name is the result of looking for short-ish .com website names that were memorable, weren’t already taken, and that captured something about what the site is about. I made a big list of them and then discussed each with Charlotte Randall: StatsBomb The Company’s co-founder, current Chief Operating Officer, and also my wife. The one I liked best was taken (and has still never been used in the last ten years), but StatsBomb felt like a clear second best to us. It also rhymed with .com, for a nice little mnemonic when spoken aloud.
Certain nameless Twitter community members have complained endlessly about it over the years. (Though not nearly as much as they complained about the term “Expected Goals.”) Sky Sports once told me that if there should ever be “an explosive event in the UK,” they would no longer be able to credit our company for visualisations on air, out of respect for the victims.
On the other side of the coin, investors have lauded it over the years for being memorable in a sea of sameness.
We’re still here, and I’m still spicy about it, so it can’t be that bad.
Those early years saw waves of Ben Pugsley and I publishing large volumes of work. Because so little had been done publicly at this point, it was easy to do surface level analysis on a thousand different topics. There will be other pieces in this 10th anniversary series looking back at contributors and some of the most notable works from that time (Colin Trainor and Marek Kwiatkowski are two who had large early impacts across the football space, while Mike Goodman went toward media), but I was happy to have a diverse group of opinions on the site publishing smart stuff.
We also weren’t the only ones writing things around football analytics. The list of those who went before us is moderate - Charles Reep, Richard Pollard, Garry Gelade (RIP), Sarah Rudd, Ravi Ramineni, Jaeson Rosenfeld, Gabe Dejardins, Chris Anderson, Howard Hamilton, Paul Power, Will Morgan, Paul Riley, Simon Gleave, James Grayson, OG Mark Taylor, Devin Pleuler and a handful of others - but the contemporaries both on and off of StatsBomb seemed to grow fairly quickly. We could disagree about stuff, but it was mostly congenial and helpful. Twitter was more like a collegiate/academic atmosphere then instead of whatever it currently amounts to.
The internet was a different place too. You could get new articles promoted by huge accounts on Twitter that liked your work (Gabriele Marcotti was an early fan). People clicked on blogs/articles in the tens of thousands - especially when retweeted from the official Borussia Dortmund account. It took a massive amount of work, but the website and the Twitter following grew rapidly for what was actually a deeply nerdy, non-mainstream pursuit.
Don’t get me wrong… Opta and Squawka and Whoscored were already fairly huge. They got that way because fans like sports data stats as trivia. Stats trivia and top 5 lists are popcorn and people would consume it endlessly and mindlessly. Our goal was to engage their brains, glimpse into the future, and tell them why people were probably wrong.
We did analytical previews of the Premier League season that were completely new in the space. We also reviewed transfers, and gave actual criticism of moves we thought were inefficient or wastes of money. We produced some of the first work on age curves and per 90 stat distributions in public. We built frameworks for evaluating managers/head coaches on outputs. I did analysis of team tactical styles that seemed to lead to better/outsized success that continue to provide useful signal almost a decade later. And I did a mountain of early work trying to find, analyse, and predict young player and transfer performance across various positions. That last bit is probably what got me the Brentford job.
Oh, and we argued with journalists. And ignorant football fans. And proper football men. All of whom seemed to have very strong opinions about how the nerds were wrong, and stats and data could never be properly applied to football/soccer because it was too free-flowing. It wasn’t like baseball. You couldn’t measure a player’s heart. Coaches and Directors of Football and Scouts and Commentators on TV were all more knowledgeable than we were because they had spent their lives around the game.
That… has proven conclusively not true.
“Watch the matches, nerd.”
My dudes… it was always dudes… I was a professional gambler and hugely successful football trader in the toughest, most liquid gambling market in the world. I did it on 1.5% margins! And I crushed. Because I watched like seven EPL matches and sundry others every single week as part of my job, for a decade. We are not the same.
Fans of the site somehow prodded us into doing a podcast that was incredibly well-reviewed and followed with a fervor that first Pugsleh and I, and later James Yorke and I never fully understood. (Before you ask, there will be two new podcasts out soon.) We liked writing and analysing things. But people wanted to hear us talk about it too? Very confusing.
But the podcast was how we found out about our celebrity fan.
There was one time Sean Ingle (at The Guardian*) had a chance to interview Billy Beane about a wide range of topics, including his interest in football. Some time later, Sean and I were chatting and he said, “Did you know that Billy Beane is a fan of your podcast?”
“No fucking way.”
“Seriously. He told me so when I interviewed him.”
“That… I can’t… What the fuck?!?”
I have met the man in person. He really did love the podcast. It will never cease to blow my mind.
I think the last truly notable thing that the early StatsBomb did was create visualizations. If you look at the 2013 on the site, the only vis present are some Excel bar charts. There are a lot of tables of numbers and screenshots of data dumps, but nothing else that we would become known for. Then in January of 2014, I introduced the first radar charts. I did this after spending about 4-5 months reading about data visualization across a ton of textbooks and figuring out how I wanted to approach the topic.
Those first radars were not good, but without the not good versions, none of the other ones would exist.
I had Nat James - employee number 1 at StatsBomb, the Company - create a Photoshop template of the radars and teach me how to create each new one by hand. Every radar I would spew out on Twitter for months in the future would take me 10-15 minutes of hand entry and Photoshop tinkering. My heartrate just spiked while thinking about it. It was grueling, but also incredibly rewarding. Then I started producing more radars in public and iterating improvements. These were eventually turned into a hacked up radar generator by the incredibly generous Steve Kim**, who made my life so much easier, which in turn allowed me time to do more cool things.
These vis generated responses and interactions in a way nothing else from the website ever did, including starting a decade-long argument across Twitter between true quants, fans, and applied nerds on whether they were useful/good/misleading. I’m not going to rehash them here because I’ve already written tens of thousands of words on the development of the vis, and there are other stories to tell. What I will say is that our stuff is still the prettiest damned version of this visualisation on the internet, I scienced the shit out of them to fix most of the weaknesses, and people bloody love them... To the extent that basically all of our competitors had to create substandard versions (or rip ours off directly - sup DribLab) to satisfy demand.
One of the things that gets missed is how much of the radar work also built foundational information for what data might matter in evaluating different positions. Some of it seems obvious now, but let me stress that this did not exist in public before I did it. It took a tremendous amount of thought and research into what great players of each position actually did that mattered in order to come up with the templates, which then kicked off all the work into the statistical distributions of those outputs, and arguments among various colleagues about whether I was right or not. It also flagged up some glaring deficiencies in the leading data provider’s collection issues, but one does not bite the hand that feeds one when existing on free scraps.
People won’t know this, but toward the end of 2014 Marek Kwiatkowski and I also designed a flexible framework for shot maps that I don’t think anyone has even come close to bettering in the last nine years. Shot maps are hard, especially when you need to scale them from single games into entire seasons, and use the same framework for both players and teams. The ones that exist in StatsBomb IQ now are almost the exact same ones we collaborated on back then, and they are rich, bold, easily usable, information dense, yet still insightful.
If some of this piece seems a bit self-congratulatory, please forgive me. You have to take your wins and credit when they appear. No one else ever gives it to you, and it’s been a long, fascinating ten years.
The Brentford Era
The first paying job I got in football was helping Cesc Fabregas’ team when he was transferring back to England from Barcelona (2014). I put together a statistical dossier on why he would still be pretty awesome and why he deserved some team to pay him all the monays.
He was my favourite player at Arsenal. I was not hoping that he would move back to England to play for and win a title at Chelsea. But I got paid, Cesc got P.A.I.D, it was fun, and he was still really good at the whole midfield thing, and now it’s a good story to tell.
We have also advised various players on their potential moves over the years, including Raheem Sterling, KDB, Rudiger, Bamford and sundry others. It was kind of fun to glimpse behind the scenes on the player side, especially after working for clubs.
In summer of 2014, I was hired to work for Matthew Benham/Smartodds/Brentford. The first project I did there was around designing better set pieces, which helped FC Midtjylland win a few league titles along the way. (Though full credit to Matt and Gianni Vio, who knew there were huge strides to be made there, and to Brian Priske - now of Sparta Prague - for executing this stuff on the pitch.) If you’re interested in learning about set pieces, we produced a public course to teach coaches and analysts that has been lauded as the best of its type. It is not cheap, but goals never are. Get your team to pay for it.
I wasn’t hired to do set piece work, though. Eventually we’d start applying stats and data to player recruitment and styles of play concepts at both teams, with some pretty good success while building a small piece of the foundation for Brentford to eventually leap into the Premier League.
Meanwhile, over at StatsBomb… one of my conditions for working at Brentford was not shuttering and deleting the content on the site, as happened when analysts were hired in so many other sports. I still published occasional pieces of my own, but StatsBomb needed a good pair of hands to keep things moving while my focus was in the actual football world. (And I think Colin Trainor’s focus at this time was at Bournemouth, so there weren’t many people from the first wave left to produce content.)
Enter one James Yorke.
James used to write weekly roundups of stuff on his own personal blog and he is a hugely talented writer who was interested in, if not yet prolific at, the stats side of things. He not only continued to produce new content for StatsBomb, but also kept recruiting new writers to do additional analysis. Many of whom would later go on to work for professional football teams and data companies.
There are a few league titles in that list, plus a few current StatsBomb employees, and at least one medical doctor.
James did a great job, and was instrumental in setting up the website to transition into the next phase of existence: StatsBomb, The Company.
More stories next time.
All the best,
* I admire Sean’s work and think he is an unbelievable writer, but the tone of the coverage around trans people at The Guardian turns my stomach. One of the few regrets I have from running this company and the website was a disagreement I had with our former writer Grace Robertson. Grace used to regularly take Sean and others to task on her public Twitter account while writing for the website. This was a bad look for StatsBomb the business, as The Guardian were potential future customers - hence the disagreement. However, on a personal level Grace was 100% right and it sucked, and I owe her a public apology. I’m sorry.
(**I haven’t heard from Steve in ages, but I hope he is doing well and enjoyed all the Liverpool success in recent seasons.)