Moving analytics from Moneyball into football

Bill Gerrard is a data analyst at AZ Alkmaar and professor of business and sports analytics at Leeds University. He previously worked with Billy Beane at the Oakland A’s and for Saracens Rugby Club. Here he talks to The Set Pieces about the rise of analytics in sport and its use in football…

“It was put to me recently that Moneyball has become a bit of an albatross for analytics. I guess there are two sides to it. Moneyball was a game changer in highlighting the use of data analytics in sport and the possible competitive advantages. The downside of it was that people often forget the context of baseball and underestimate the difficulties of taking a baseball story and moving that into a completely different sport.”

“It’s what I’d call moving from striking and fielding sports – like baseball and cricket, which are reasonably straightforward in terms of analysis – into the invasion-territorial sports, where at the core there’s the tactical coordination of a group of players. When you look at cricket and baseball, the essence of it is a one-to-one contest. Once you move into sports like football and rugby union it becomes much more difficult. That’s the exciting challenge because there is so much more going on that you have to take into account.

I got introduced to the people at AZ through Billy Beane. He recommended me to Robert Eenhoorn, the general manager, when Robert took over in 2014. Robert went to college in the States, excelled in baseball, made it into MLB, but never quite nailed down a starting position. He then went back to Holland and ran their national baseball operation very successfully, winning a world championship. AZ invited him to become their general manager, effectively in charge of the sporting operation and the football side of things. He has a very similar outlook to Billy: not just relying on gut instinct but planning your decision making so you do your due diligence.

Billy and I have been involved for the last couple of years. My responsibility has been working on developing the use of the data to go alongside traditional video analysis. I work alongside the performance analyst at AZ. I analyse the Opta data (from matches) and send off reports looking at the performance of the team as a whole and individual players. I point out what the key strengths of the performance were and, at least in terms of the numbers, what the weaknesses were, to stimulate debate in the coaches’ meeting when they review the game.

The attraction to analytics is what I would call a ‘David strategy’. It’s how an organisation that’s resource-constrained tries to compete with much richer rivals. The teams that are open to a more analytical approach tend to be teams that have to operate within budget constraints. You saw that in Oakland, you see it in leagues that are operating in salary caps. AZ, in budget terms, are quite restricted compared to the likes of PSV, Ajax and Feyenoord.

It’s about having to come up with an alternative. You either give up and just accept your place in the world or you use what resource you do have better. That’s essentially what Moneyball is, using all the evidence and the data to try and compensate for that resource gap and not just relying on gut feeling. It’s forcing people to justify their decisions and their recommendations. It’s a discipline. The role of data analysts is to try and ensure that we make better informed decisions and use what’s available. People saw the film as a story about player recruitment. But I see the use of analytics as much broader than that, as using data to support any coaching decision.

People forget that Moneyball the film, although based on a true story, was not an authentic representation of the truth. For dramatic purposes it highlights conflict between the traditional approach of looking at players and evaluating them intuitively, versus the analytical approach. I always differentiate between the Oakland A’s run by Billy Beane and the Hollywood A’s run by Brad Pitt. There’s no way Oakland would have been as successful as they have if they had operated in the way that the film portrays. They would have been so dysfunctional.

I can see elements in the film – the comment where Billy doesn’t watch games, he goes off to the weights room – that are true. The first time I ever went to Oakland, I sat in Billy’s seat to watch the game and he sent a security guard at the bottom of the 8th to get me. I met up with him as he finished in the weights room. Billy doesn’t want to watch the video and analyse it wrapped up in the emotion of the game. He keeps in touch with the scores, but he’ll analyse the video knowing only the result, and that takes the emotion out of it.

It’s about recognising the biases regarding decision making. If you ask a coach to review a game from memory, they will only remember key points. A player might have had 89 minutes of great performance but if they make one mistake, that will cover the whole assessment of their performance. Coaches are normally very good at remembering things that happen right at the start of the game and approaching half time.

The problem with football is that analytics has made a lot more headway in player recruitment, and less so in the day to day coaching. It only works effectively if everyone’s singing from the same hymn sheet. What I’ve found in a number of football clubs is that you get a separation of powers. The analytics is done in the player recruitment and scouting, but if you don’t have the buy-in from the coaches then inevitably you get tensions. It usually goes hand in hand with the recruitment being controlled by a technical director and not the manager/head coach. Where it works well is when the head coach and technical director see the world in the same way, so the types of players that are being analysed are players they both agree on.

Where you get a dysfunctional outcome is when there is very little input from the head coach in player recruitment. Again, it’s partly the problem of taking a baseball story and putting it into football because in baseball the field coach is trying to get the best out of a group of players individually. In the invasion-territorial sports, there has to be a tactical plan and the players have to be recruited into that tactical plan. As I’ve seen at some clubs, if you’ve got a technical director who’s recruiting on one basis, but a head coach who’s operating on a different one, that’s when it just doesn’t work and it falls apart.

Of course the manager always feels in a weak position because he knows it’s his head on the line if the team doesn’t function. The media often dramatise it in much the same way as the conflicts were dramatised in Moneyball. They build up the story and exacerbate the conflict, but often there’s an element of truth that there’s some disagreement. You’ve got very different views on it. It works a lot better on the continent than it does at times here. Certainly at AZ there’s absolutely no issue. The football operation has three departments: there’s an academy, a scouting department and a first team. The technical director’s role is to ensure the three departments work together.

I still don’t hear of too many clubs that are using analytics to support the coaching staff. I think most of it is still in the academy and scouting departments. Some clubs have invested quite a lot in analytics and some academies have been very progressive. The young coaches coming through are far more adept in the use of computers and data. And the players now are the ‘Prozone generation’, so when they go into coaching they’ll have spent their entire playing career having their data tracked and analysed. Football’s probably a lot further down the line than is often given credit for, but there’s still scope for development.

For more on Bill Gerrard’s views on how sports teams are using wearable technology to improve performance, read his interview with MBNA here.

Moving analytics from Moneyball into football
4.68 (93.64%) 22 votes