All the top teams won and (nearly) all the bottom teams lost, so there is praise for Ashley Young, Aston Villa, Jordan Henderson and Jonas Gutierrez. Not Di Maria, though...
Never mind the long goodbye, should he actually get back into this excellent Liverpool side playing with pace? Mostly we talk Manchester United, though. It's good...
During the transfer window on Twitter, it was striking that there were so many people claiming extensive knowledge of targets and potential targets and suggestions for targets for their clubs. The realisation that often I had only the vaguest notion or no notion at all of who they were talking about was unsettling for this long-in-the-tooth football fan. I was amazed at the apparent depth and breadth of knowledge. How could people know so many players? Sure, lots of people watch Spanish, German or Italian football - but Russian, Ukrainian etc? What was going on here?
Then the generational realisation dawned on me - these people were most probably gamers of the Football Manager ilk. Forget about the ITKs, these were the KIAs - Know It Alls - who appeared to know every player in Europe, who had a rather inflated opinion of the worth of their opinion, and who often got annoyed that their clubs didn't seem to be following their suggestions on Twitter.
In a way it was easy to brush them off as people who should probably get out more, but the 'knowledge' they appeared to possess raised some rather interesting questions in my mind. The best of the football management simulation games clearly have powerful databases underpinning their game play. So, do actual football clubs use that information or similar? And if so, what clubs and how?
For answers, I turned to Lee Mooney - a data analyst, business intelligence expert and writer for The Tomkins Times website. Lee has been working in and writing on this area for some time and provides excellent insight into the power of data and football...
Paul: Games like Football Manager, for example, are ultimately just that, games, and thus aimed at that market - but are there 'professional' databases aimed specifically at the football industry?
Lee: Yes indeed. There are a number of established player recruitment solutions on the market. Each offers a subtly different perspective and/or coverage of particular territories. There is also a growing number of 'boutique' propositions, like my own, which are entering the market inspired by the 'big data' idea.
Paul: Back to the football management-type games for a minute. How is their data compiled? Who populates their databases?
Lee: I'm not an expert on this to be honest. But, from what I've learned, I understand the game creators employ a network of enthusiasts to compile and maintain their player databases. They might have an individual that 'owns' a specific team or league within a country.
Paul: Are those who compile the data 'football experts'? That is, can the expertise of those feeding into the database be trusted to provide a realistic picture of a player?
Lee: Again, I'm no expert. I know of one person that has had one of these roles. He currently works for a football data company. The question of 'trust' is a tricky one. They can certainly be trusted to collect data that is 'good enough' to fuel a hugely popular game. Thinking about it, the game data is the product of decades of evolution, investment and crowd-sourced validation by thousands of dedicated enthusiasts - so I'd bet it's pretty good. But, being 'good' doesn't mean it's the 'right' data to inform multi-million pound decisions - at least not in isolation.
Paul: You made the following remark about the Football Manager database in a piece for the The Tomkins Times: 'The Football Manager database, especially in the hands of someone who knows how to manipulate and integrate data, is a hugely powerful resource.' When you say in the 'hands of someone who knows how to manipulate and integrate data', can you give us an example of what they might be able to do over, say, an experienced gamer putting together his virtual squad?
Lee: I can certainly offer some insight as to what I might do. Each published game (for example, Football Manager from its first iteration to its latest) is effectively a time capsule. If you were to bring all of those 'snapshots' together you'd have - I suspect - the deepest and cleanest structured player database in the world. If we assume that the data is consistent, even if it's 'wrong', then we can start to create useful content from that consistency - not enough to drive a recruitment decision, but perhaps enough to add another cog to that machine. For example, you could analyse hundreds of thousands of individual careers to understand how perceptions about a player changed over time. Did that 'exciting youngster' go on to have a long and successful career or not? Another useful application would be to use recursion and regression to perform 'find a player like X' analysis. Each player is just a sequence of numbers. These numbers can be correlated to create a co-efficient between 0 and 1 (1 representing a perfect match). So, you could pick your 'dream team' from the last ten years - the perfect player for each role at the peak of their powers - and then process the database to find the players that mostly closely fit that profile today (or are most likely to fit that profile in a couple of years).
Paul: Do you think football clubs are well-placed to exploit the potential of such data? Are they, for example, staffing their scouting departments with that, at least partially, in mind?
Lee: I've been fortunate enough to meet with several clubs in the last year. With the exception of one club, I'm yet to meet any with a 'data scientist' type on their permanent payroll. The capability is available to clubs, but universities, consultancies and dedicated enthusiasts usually enable it on a temporary basis (and often free of charge). It's also worth noting that any potential 'explorer' would require legal consent and support from the vendor to do any of the things I've described.
Paul: Would such data-mining only be of interest to certain types of clubs - clubs like Liverpool or Newcastle, looking for hidden gems and value, rather than say Chelsea, City, Real Madrid who operate in a very different market?
Lee: Not at all. I think every club should be interested in the kinds of recruitment analysis methods that I've been writing about. Every year, clubs will make a small number of decisions that will allocate huge proportions of their annual revenue on transfer fees and wages. These are epic decisions with the potential to cripple or propel a club's development. The very biggest clubs will have a much smaller talent pool to monitor and are perhaps less interested in 'growth'. However, big clubs have their own challenges. What if Gareth Bale really is a 'home bird' and ends up having an emotional breakdown living in Spain? How much of that world-record transfer fee are Real Madrid likely to recover?
Paul: How can clubs like Newcastle, Liverpool etc. exploit such data - and exploit imperfections in the transfer market, as some put it?
Lee: If success in football is all about money, then the clubs without it have a few options for generating it - all of these have their roots in the football 'product' that they produce - which is about assembling a team that wins games in a style that engages fans beyond your immediate locality. All of this needs to be done whilst managing costs and the club's exposure to risk. Effective analysis of the 'right data' has the potential to inform all of these elements. Introducing a little rationale to a world that is seemingly so emotional can only be a good thing (providing the work is good quality and decision-makers take notice).
Paul: Suppose for a minute you were chairman of a League One or League Two team, would it be worth investing in someone who could manipulate this data for you? Is there value there for clubs at that level?
Lee: I think there is, but cost is a factor. I suspect smaller clubs would not fund such a capability. After all, the costs of people and data don't change because you're a smaller club. Rugby League and Rugby Union offers some great examples of how to do data-driven innovation on a budget, so I suspect the challenge is more one of culture than finances. If I was the chairman of any club, at any level, I'd commit to developing this kind of capability - but I'd do so iteratively and by starting small (with each iteration being expected to pay for itself and earn its place at the decision-making table). For me, it's really not about 'one best way' - but about blending the different perspectives: performance, technical, financial, statistical (and so on) that exist within a football club.
Paul: Finally, and only half-jokingly, what would a computer gaming Know It All on Twitter need to do to make the jump from a football management game to a real-life position at a football club where their 'in depth' player knowledge might be listened to?
Lee: I don't think they can - at least not without something more than their acute knowledge of a game database! Games like Football Manager are cheap yet powerful resources. Few professional tools will match the coverage of the Football Manager database, for instance. Yet, alone, it's really just a great list of players - not the actionable information that football's decision-makers really need. I can connect to a Facebook data stream and list 'people', but can I use that same data to tell you which one we should hire for a particular job and how much their salary should be? Unless we do something useful'with that data, something that enriches it, the answer would be 'no'. This is where people aspiring to operate in football need to play. If you can produce insights that are genuinely actionable by a scout or club director, with a robust underlying rationale that they can understand, then eventually...and it could take some time...doors will start to open for you.
Paul Little - follow him on Twitter