Click here for Liverpool FC personalised gift shop

Page 1 of 4 1234 LastLast
Results 1 to 10 of 34

Thread: Meet William Spearman, Liverpool’s secret weapon

  1. #1
    Join Date
    Jun 2014
    Posts
    6,119

    Meet William Spearman, Liverpool’s secret weapon

    Today in the Athletic, by Simon Hughes

    Quietly, there is a new arms race in the Premier League and those trying to find their place in the rush acknowledge that it is being led by the champions.

    Figures at other clubs have described William Spearman as “Liverpool’s secret weapon”, a researcher whose name appears whenever articles about the use of data are written — though he has never received the same level of focus as Michael Edwards, their sporting director, or Ian Graham, the chief data analyst.

    His only public appearance was accidental.

    When someone filmed Jurgen Klopp dancing at the club’s Christmas party in December 2018 after Manchester United had been beaten 3-1 at Anfield earlier that day, Spearman was in the background, performing some impressive moves of his own.

    Spearman is a scientist. Colleagues at Melwood occasionally call him “the American boffin”, though the term is an affectionate one. Most of his focus is on recruitment and that means he is close to the scouts, with whom he shares an office. Beyond social conversation in the canteen over lunch, he rarely speaks to the players — who roughly know what he does but not always exactly.

    Edwards and Klopp prefer it that way. Spearman’s findings can be complex. While Edwards’s background means he is comfortable looking at numbers across spreadsheets, Spearman’s information needs to be packaged in a way that inspires a clear mind for those who are less familiar with the analysis. That is where the analysis team come in, converting what he has discovered into videos which get presented at team meetings.

    When Klopp speaks about the support at Liverpool “blowing my mind”, he is referring immediately to materials brought to him by Greg Mathieson and James French, who study opponents, as well as Harrison Kingston and Mark Leyland, the post-game analysers. They are active on a match day, watching what is happening live and via monitors. Their remit allows them to operate in dressing rooms.

    Spearman is usually sat in the stands. He is right at the start of the production line, setting the programmes and tools that deliver the data from which conclusions are formed. As much as 60 per cent of his time is spent on long-term innovations, the sort of blueprints he’d want to publish on if he were working in an academic setting — as he once did at Harvard University.

    After that, 30 per cent of his time is spent collaborating on short-term or bespoke projects with coaches and analysts. This could be a piece of focused work over a concentrated three-week period or a few days each week over a few months where the idea goes through a refining process. The remaining 10 per cent is recurrent work: post-match reporting, usually geared towards potential new signings.

    When Spearman moved to England from Texas in 2017, he decided to rent a house a short distance from Melwood because he was unsure how hard it would be to get a UK driver’s licence. He reports to Graham, who lives and works from his home in Cheltenham, some 140 miles from Liverpool’s training ground. Though Graham travels up to Merseyside a couple of times a month, it is easy enough for the tech-savvy research team to maintain contact via the Slack messaging facility, which is relentlessly used throughout any working day and proved to be very helpful during the pandemic, especially as Spearman had returned to the US while Britain was in lockdown.

    Graham’s Gloucestershire address isn’t far from that of Tim Waskett, another figure employed by Liverpool who keeps a low profile. An astrophysicist, he has been a software developer and statistical researcher since 2012 and during the week he and Graham work from each other’s houses.

    There is a team of six researchers at the club. Waskett was Graham’s first hire and he was essentially brought in to implement the data models that Graham did not feel like implementing himself. While Graham is the ideas man, Waskett is a programmer. In his previous job with performance analysis firm Hudl, Spearman had to clear a lot of data and synchronise certain events into his models but at Liverpool, Waskett handles a lot of that.

    Dafydd Steele, like Graham, was born in Wales and the former junior chess champion joined Liverpool in 2013 from energy company Petronas with the remit of looking for answers to very specific questions. Mark Howlett, the systems administrator, manages Liverpool’s database while Mark Stevenson is a software developer who focuses on front-end visualisation— preparing data findings for other departments. Aside from the data analysts and the scouts, the researchers also create tools for the club’s fitness and conditioning unit led by Andreas Kornmayer.

    Spearman was hired because of his understanding of player tracking data, having created a statistical tool at Hudl which measured levels of control across a football pitch. Tracking data is where the ball and players are tracked, usually at a rate of once every 25th of a second, using in-ground cameras. This data complements the events recorded by the likes of Opta, which show every ball touch.

    The tool was of interest for Liverpool especially because of its capacity to measure distances between player and ball and realistic expectations on the seizure of possession. The ability for a player to press, after all, is something Klopp values very highly.

    Liverpool have since become one of the very few clubs to invest early in broadcast tracking data, which is a fairly recent innovation in the area of football analytics where positional data on the players and the ball are captured not using in-ground cameras but from broadcast footage of matches. Their partnership with SkillCorner means they have access to a machine that monitors the movement of every player.

    In the meantime, Spearman has helped to develop a model that means Liverpool’s analysts are now able to estimate the probability of a goal being scored from anywhere on a pitch within the next 15 seconds. This tool is used to evaluate player performance after all Premier League matches, not just Liverpool’s.

    “We can analyse a lot of players all around the world using the ball touch event data,” Waskett said when he spoke at a Royal Institute lecture in December 2019. “It gives us really good information on which players are doing well and who we might be able to sign in the future.”

    College Station is a US college town built originally on the back of the earliest Texan railroads, located somewhere in the middle of a triangle created by the big cities of Dallas, Houston and Austin. Education is the primary source of employment there and in the most recent census was responsible for more than 16,000 jobs.

    Mark Spearman studied at Texas A&M (Agricultural & Mechanical) University, graduating in 1986 with a PhD in industrial engineering. Nearly two decades later, he returned to the campus as a department head carrying more experience in academia, having initially worked as an associate professor at Northwestern University near Chicago, before securing a more permanent role at the Georgia Institute of Technology in Atlanta.

    He would realise after less than a year, however, that he did not enjoy the politics that came with university leadership and returned to consultancy work with his own company, Factory Physics — a name taken from one of his earlier textbooks published while at Northwestern.

    Spearman’s wife, Blair, was a biologist. Maybe it should be no surprise that at least one of their three children emerged as scientifically minded. Jacob was born first and Rebekah last, though she preferred the classics and is currently completing a PhD at the University of Chicago, where she also works as an instructor.

    William Spearman was born in Chicago but moved to College Station when he was six because of his father Mark’s work. In 2005, he enrolled at the University of Dallas, where he studied science, specialising in physics. Following a year-long scholarship in Geneva, he returned to the US and, at Harvard, completed a PhD in high-energy particle physics where his thesis measured the mass and width of the energy field Higgs boson. Though some reports have made a lot of his role in such a far-reaching investigation, The Athletic understands he was one of as many as 5,000 researchers working on the project.

    Like his father, Spearman stepped away from academia and into industry, in 2015 joining Hudl, a performance analysis tool manufacturer used by lots of American sports teams. Though Spearman watched American football, he was not a sportsman himself and his competitive interests at college in Dallas extended only as far as the frisbee club.

    He became interested in what the rest of the world calls “football” at Hudl because of the flowing nature of the game and the complexity of the data created. Though analytics had long since established itself in decision-making processes in baseball, Spearman felt it was easier to crunch numbers in America’s most popular pastime because of the fixed number of hitting outcomes: first base, second base, third base, home run or out.

    This, he thought, made modelling very simple. While American football was somewhere in between, basketball was the closest to football in terms of analytics but remained much more straightforward to assess because there are so many point opportunities since on a court that would fit in a football pitch’s penalty area a team was always within one pass of scoring.

    There was the threat of the counter-attack in football, but the pitch was much bigger than a basketball court and the probability of a long pass leading directly to a goal was much lower. He could see that in football there was an increased number of phases in play and, as in ice hockey, outcomes were naturally influenced by the use of space.

    Crucially, five years ago, he could see that data was still in its infancy in football. Not every club had embraced its use. In his mind, so much about the game remained unexplained. Even though he was one of the thousands of students working on the Higgs boson at Harvard, the experience gave him a taste for the cutting edge in a field that had not been explored before. Analysing something only a small number of people across the world were studying excited him. Others might have had access to the same data but he liked to think he looked at it differently.

    Inside his first year at Hudl, he had designed the control model that alerted Liverpool.

    When it was made public in 2016, Graham and Steele wanted to meet him and after that, the trio kept in touch, mainly via email, until the following year, when Graham asked Spearman whether he’d be interested in joining a club.

    Though he had established a home in College Station where he was able to work for Hudl remotely and the lifestyle change in Britain daunted him, he could not resist the temptation to try something he considered groundbreaking.

    Researchers at football clubs tend not to advertise their work, because it might alert a rival to a new development.

    This explains why there are challenges detailing exactly which club is leading.

    Spearman is the sort of person who finds innovation addictive. Certainly, Liverpool cannot afford to rest on their laurels and make the sort of mistakes other clubs make. Liverpool’s owners are rich but do not have the wealth of Manchester City or Chelsea’s owners. So they have to be smarter and better in other ways.

    Chelsea, supposedly, are building a promising analytics team. City, meanwhile, recently made three new hires and are working more closely with scientists at the University of Manchester. Elsewhere in the Premier League, Spearman apparently respects the work of Mladen Sormaz, his contemporary at Leicester City, as well as Sarah Rudd at Arsenal, who transferred over from StatDNA when the club acquired the start-up seven years ago.

    Rudd and Barcelona’s Javier Fernandez are considered the leading researchers in football, though respected figures inside the game think Spearman is actually ahead of them at the moment because he is being listened to more both at ownership and management levels.

    When Spearman joined the club three years ago, one of his biggest fears was that he wouldn’t be taken seriously, or that it would take years before there was any sort of buy-in from significant figures. His friends at other clubs have struggled to regain credibility after presenting seemingly half-baked ideas but he’s never felt that way at Liverpool because he has always been clear that his first version of any potential tool might sound far-fetched. There have been lots of instances where Spearman’s models have said something about a player that hasn’t correlated with the opinions of a scout.

    Those who have worked with him say they have only had encouraging conversations. As much as the scouts learn from him, he learns more about the realities of football from them.

    As one source told The Athletic, “Liverpool are ahead because they’ve got the balance right between the football people and the programmers.”

  2. #2
    Join Date
    Jun 2014
    Posts
    6,119
    If you find sports analytics interesting here he is in action

    https://youtu.be/X9PrwPyolyU

  3. #3
    Quote Originally Posted by 3underpar View Post
    If you find sports analytics interesting here he is in action

    https://youtu.be/X9PrwPyolyU
    I came into football with Bill Shankly as my hero.
    He must be wrestling in his grave.
    What has happened to the game.
    It is not just football but everywhere - data analysts.
    I took a course a few years back with the Open University on data analysis.
    It was quite brilliant but it taught me that you can portray any picture you want with certain graphs.
    The right targetted audience will believe the data.
    Depending on your audience?
    Data analysis is only a branch of mathematics. I quite love some branches of mathematics.
    Time will tell if the trends portrayed come true.
    If more and more influential people believe such data then perhaps it is a self-fulfilling prophecy.
    For myself I fear for the World in the hands of data analysts. They will make the graph match the buyer's wishes.

  4. #4
    Join Date
    Jun 2014
    Posts
    6,119
    Quote Originally Posted by scientificred View Post
    I came into football with Bill Shankly as my hero.
    He must be wrestling in his grave.
    What has happened to the game.
    It is not just football but everywhere - data analysts.
    I took a course a few years back with the Open University on data analysis.
    It was quite brilliant but it taught me that you can portray any picture you want with certain graphs.
    The right targetted audience will believe the data.
    Depending on your audience?
    Data analysis is only a branch of mathematics. I quite love some branches of mathematics.
    Time will tell if the trends portrayed come true.
    If more and more influential people believe such data then perhaps it is a self-fulfilling prophecy.
    For myself I fear for the World in the hands of data analysts. They will make the graph match the buyer's wishes.
    Agree with much of this. I’m in education and data analysis is taking over in my field too. Still it is cool to see new ways of rating players and predicting positions to get into for scoring chances for example. Obviously the boot has to kick the ball into the net still.

  5. #5
    Quote Originally Posted by 3underpar View Post
    Agree with much of this. I’m in education and data analysis is taking over in my field too. Still it is cool to see new ways of rating players and predicting positions to get into for scoring chances for example. Obviously the boot has to kick the ball into the net still.
    Great post 3underpar
    From my studies of data analysis there are the following crude steps
    1. identify someone interested in your data analytics
    2. propose a likely outcome
    3. agree a contract
    4. gather raw data
    6. clean raw data from rogue inputs
    6. analyse data
    7. present data to client
    8. if client does not like presentation go back to step 4
    9. client is happy with presentation
    10. you get paid

    Big business these days

  6. #6
    Quote Originally Posted by scientificred View Post
    Great post 3underpar
    From my studies of data analysis there are the following crude steps
    1. identify someone interested in your data analytics
    2. propose a likely outcome
    3. agree a contract
    4. gather raw data
    6. clean raw data from rogue inputs
    6. analyse data
    7. present data to client
    8. if client does not like presentation go back to step 4
    9. client is happy with presentation
    10. you get paid

    Big business these days
    For point 4 3underpar4
    I find myself always asking
    Who is collecting this data? Are they bonafide? Are the people they are asking bonafide? Who checks them nationally/internationally.
    Even then.. for example watching a 90 minute football match it is pretty hard to concentrate on that whole 90 minutes focussed on one individual even with modern TV coverage
    What constitutes a pass or tackle good/bad.
    Yet these statistics are abounding. If it is to be believed then there must be manifold checks at many levels by different departments.
    Is that down to one persons opinion or is it further verified and who is that person(s)?
    e.g. regarding heatgrams what technolgy is involved there - do players have sensors on their bodies?
    With the ensuing computer modelling of players there must also be an algorithm to decide how important there position/influence was with regards to their sense of position and importance regarding play. How important is that algorithm and does that algorithm dictate how good a player is. I find it all a little bit spooky. I am not against all technology. I am sure it has place but I lament the art of a true footballing scout like Geoff Twentyman.

  7. #7
    Join Date
    Nov 2009
    Posts
    15,143
    It's all very subjective, but I've never understood resistance to analytics and data.

    They're there simply to make our understanding of a medium even better.

    A friend of mine who is religious says the Northern Lights are sent from God - I (quite scientifically-oriented) on the other hand understand how the Northern Lights work and thus they don't hold a comparative "spiritual" resonance, yet they are no less beautiful to my eyes and data (for me anyway) doesn't take away any of the magic of how we came to win the League.
    Your hobbies are rollerblading and you're also a bit of a rat-hound? Steel Wool
    Sid knows he's crazy and he likes it. Balinkay

  8. #8
    Quote Originally Posted by Insidious View Post
    It's all very subjective, but I've never understood resistance to analytics and data.

    They're there simply to make our understanding of a medium even better.

    A friend of mine who is religious says the Northern Lights are sent from God - I (quite scientifically-oriented) on the other hand understand how the Northern Lights work and thus they don't hold a comparative "spiritual" resonance, yet they are no less beautiful to my eyes and data (for me anyway) doesn't take away any of the magic of how we came to win the League.
    You should question it always.
    What has religion got to do with it?
    Going off topic.
    I don't recall such data being mentioned in the Bible.
    Personally speaking I would hope we are all viewed as something more than our data statistics.

  9. #9
    Join Date
    Nov 2009
    Posts
    15,143
    Quote Originally Posted by scientificred View Post
    What has religion got to do with it?
    Was trying to communicate an analogy of sorts in terms of (some) people not liking it when the "mystery" is taken out of things, ie "you simply can't explain how X happened" - and that was the first example of a somewhat comparative real-life situation I've been in conversationally that came to mind.

    Clearly failed miserably, never mind
    Your hobbies are rollerblading and you're also a bit of a rat-hound? Steel Wool
    Sid knows he's crazy and he likes it. Balinkay

  10. #10
    Join Date
    Jun 2014
    Posts
    6,119
    I would say that like most things a balanced approach where the data driven model is used along with traditional scouting would be best. As a manager you’d want a mix of player recommendations unless you believed all the analytics is snake oil. My guess is managers that reject analytics won’t be around much longer. Older managers might not understand all of it, not do I, but if it produces results on the field than doubters turn into believers as they say.

Bookmarks

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •