I accidentally stumbled into the world of entrepreneurship and the internet in the late 1990s. We created businesses that relied on data and upended a marketing orthodoxy made famous by John Wanamaker when he said: “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” In that new digital realm we measured our marketing spend in real time, to the penny, and were able to calculate the return on that spend immediately. This precision mentality, coupled with genuinely meritocratic cultures, created workplaces that have changed the business landscape beyond recognition. The culture pioneered at places such as lastminute.com and match.com are now commonplace but 20 years ago were unique.
When we came to football ownership 18 months ago I was interested to understand how a similar focus on the relative importance of culture and data science could improve Grimsby Town. Having read around the subject, in my view the holy trinity of books on data and football are: The Numbers Game, by Chris Anderson and David Sally; Soccernomics, by Simon Kuper and Stefan Szymanski; and Football Hackers, by Christoph Biermann. Moneyball by Michael Lewis tells the story of the data strategy that helped the Oakland Athletics to success in 2002 through recruiting “undervalued” players. Although I hoped this would be true for football, Anderson and Sally conclude that the similarity between baseball and football is weak. In football, there are too many variables. They write: “We have examined tens of thousands of European league and cup games over the course of a hundred years … and we have come to the conclusion that football is basically a 50/50 game. Half of it is luck and half of it is skill.”
Although there has been massive progress in the use of analytics in football in the past 20 years, there is an exaggerated sense of how data can be a short cut to success. The data-led, reductive view of the game underestimates the role culture and luck play. In collecting data on players and the games we get a mathematical insight into performance but it tells us little about the complex interaction between personalities and character, or the animating nature of the bonds between a club, players and fans. Anderson and Sally are spot on when they say: “A football team was as much a sociological phenomenon as a statistical one.”
Highlighting the incompleteness of our understanding, Biermann’s book gives examples from behavioural science that echo the work of Daniel Kahneman. His most useful insight was the reminder that cognitive biases exist in football for fans and owners: “As a low-scoring game, it is in football’s nature to produce singular events that get lodged in the memory. In people’s perception it’s a game of decisive moments.” It is why we overvalue goalscorers as opposed to defenders even though a goal stopped is clearly as important as one scored. We tend to extrapolate these singular moments into heuristics and often draw the wrong conclusions.
This is where data can be most helpful because it helps us to slow down our thinking and hopefully draw better conclusions. “Outcome bias” is a good example because, as Biermann writes: “If the outcome is positive, we assume that the plan or pre-game decisions were right. If the result is negative, however, we are quick to believe that his tactics must have been bad.” This came vividly to life for us in our first season with Grimsby when we had a run of one win in 11 games. It sounds paradoxical but there was very little difference between those performances and our promotion‑winning form later in the season, except obviously the results. The data tallied with our subjective view: we weren’t playing badly. We just needed time and a bit of luck to improve. And we did.
The insight I found most useful from Soccernomics is how relative spending, over the long haul, is the best indicator of success. The important but often missed part of this insight is that the sample data looks back over decades and the inference breaks down if we think in one-off campaigns. “In a single season the correlation between salaries and league position is weaker than over the long term,” the book states. “That’s because in such a short period, luck plays a big role in performance.” For any single season “clubs’ wage spending explains only about 70% of the variation in league position”. On any given day any team has a chance to win. Look at Grimsby v Hollywood-backed Wrexham last season in the playoffs or at Liverpool, who beat the league champions then lost against the team at the bottom of the table in successive weeks this month.
We believe there is a minimal level of data capability that an organisation needs and are investing in three areas as we look to retain our place in the EFL. We have a performance coach who will use analytics to enhance fitness and recovery. We have, a game analyst who will look at our performances and those of our opponents to formulate strategy and assess our performances versus our results. This allows us to think strategically about our opposition and more objectively about how much our results reflect our effort. Finally, we have hired a head of recruitment who has a background in data science and analysis to create a more efficient way to narrow the funnel of players that our management team need to look at. Objective criteria will be critical in assessing where we can improve physically but judgments of character are as important and left to the management team.
I have come to the same conclusion about football that I came to about business and life. As Anderson and Sally state: “There is no secret recipe for success locked in the numbers. There is no winning formula. There is no right answer to football. But there is a way of making sure we are asking the right questions.” You need data and information to help you ask the right questions but in the end hard work, togetherness and luck will be part of the answer.