Looking back on five years in soccer analytics: Part three

[This is the third and final part of my perspectives on Soccermetrics on its 5th anniversary.]

As I said at the end of part two, football analytics haven’t stood still over the last two years.  In fact, a lot of people have gotten involved from a variety of backgrounds and perspectives.  I’ve heard about some of them from SoccerAnalysts.com, but there are a couple of other sites that also aggregate contributions (StatsBomb is one, I’m sure there are others).  There are so many blogs on football analytics it’s hard to keep track of them even with an aggregator!  Their posts run the gamut from very simple analyses based on summary data to some composite metrics using simple or detailed data.  A number of analysts have been picked up by football analytics firms after having proved their worth with a limited dataset.  Some analysts and their works have made their way to print, electronic, and broadcast media in the US, Europe, and elsewhere.

Access to data remains and will remain a significant challenge to analysts; the more granular the dataset, the more likely it is proprietary and of course the more likely it is to be expensive.  The recourses are to scrape data, whether from a website or a video broadcast (both open the door to legal ramifications, especially in Europe), or to write on behalf of organizations that either produce the data (Opta Sports) or can afford to purchase large amounts of them (Bloomberg Sports).  The implication of the latter practice is that sports data companies have a low-risk approach for evaluating prospective talent, and a number of recent hires have been independent writers on their respective company platforms.  I expect this trend to continue as sport data companies build more analytical teams within their organizations.  Blogging collectives or cooperatives might be an alternative approach in which a group of analysts shares the costs of data and presents and discusses their analyses together. (StatsBomb could work like this, but I have no idea.)  There are a number of hurdles that will have to be overcome, from the technical, such as data formats and schemas, to the social, such as established rules of membership, sharing analysis procedures, or branding issues on the part of sports data companies.

If we accept that analytics have progressed in the last couple of years, and I can be convinced of that, why does it appear that so little of that knowledge has been transferred to clubs?  If all you had to go by the state of football analytics were recent panelist statements at the SSAC or the Leaders Sports Summit, you would leave thinking that people within the sport are puzzled with how to make sense from the temporal and spatial dimensions of in-match data, and people outside of the sport have ideas but little way of examining them due to lack of access to data.  Chris Anderson made a remark at a forum last year about the apparent stagnation in analytics at club level and the risk that analytics has of being shunted off as a fad.

(Incidentally, I did a poor job of paraphrasing the original quote during a recent podcast.  As a result, a number of people took offense to what they felt was a claim that analytics had not advanced in the past year. I apologize for contributing to the misinterpretation and resulting hurt feelings.)

From time to time one will read pieces from club insiders about the use of data in football clubs either on the training ground or between matches (a recent piece by Richard Whittall on injury analytics comes to mind), and the news that Arsenal purchased the services of an analytics firm indicates that some clubs find utility in analytics.  At the same time the news of job advertisements at Premier League clubs for performance analysts willing to work long hours at great financial cost for no pay seems to indicate that some clubs don’t really see much value in analytics.

When I started out in analytics five years ago it appeared that the clubs’ interest and aptitude for advanced analytics in football was greater than that of average fans.  I don’t think that the interest has waned or the aptitude has declined, but I do think that clubs are asking if analytics ultimately mean anything to them.  Short-term mentality is part of professional sport, and the immediate pressure to either win a championship, win promotion, or avoid relegation will always be objective #1, #2, and #3.  To that end, I don’t see football clubs being the source of innovation in football analytics.  But that behavior is not different from the way that baseball analytics grew and matured with independent grassroots analysts before maverick personnel at MLB clubs took a chance and the rest of the league followed suit.  I think the innovation will first come from the use of analytics in the media, the trends and patterns that are seen by football bettors and traders, and the nuggets of knowledge found and collected by independent bloggers, analysts, and researchers, and then flow to playing staffs and executive boards of football clubs as concepts and practices mature.

I think it would be great if more of the research in soccer analytics were published in archival journals like the Journal of Quantitative Analysis in Sports or the Journal of Sports Economics, to give two examples of publications.  I am showing my bias toward academic-type work, and I understand that a number of people in the soccer analytics community don’t want to write that type of work, but I believe that there is a lot of fundamental knowledge in the way we quantitatively understand football that would be of interest to the rest of the sport analytics community. (I’ll just say that I have no connection with the journals except that I know the founding editor of JQAS and one of my papers was published there.) The same sentiment applies for conferences like MIT Sloan Sports Analytics Conference and New England Symposium on Statistics in Sports.  I am hoping that work of similar quality is presented at Opta’s Analytics Forum next week.

So ultimately, what does all this mean for Soccermetrics?  I’ve recently found an outside job so that I am no longer dependent on Soccermetrics for my livelihood.  But I am taking a long view, and as long as I am around, Soccermetrics will continue to be around.  The focus of the organization is back where it belongs — activities that advance and mature quantitative data analysis in football, and informing the public about the same through the blog, the Newsletter, and the Podcast series.  Over the next few weeks the website will be reworked to reflect this change in focus.  I want to be a little more open, a little more accessible, a little more collaborative, a little more proactive.

If opportunities for short- or long-term consulting or paid projects arise from serious entities, I’ll consider them carefully if I’m convinced that the request is serious.  Same thing goes for research collaborations.  At some point in the future I will write some specifics on what I mean by “serious” offers.  While I am simply overawed by the number of CVs I receive every week — more than 100 since I started Soccermetrics as a startup! — I do not anticipate having open positions other than contract jobs of limited scope and duration at some unknown point in the future.

As always, I thank you — first of all, for slogging through 3,500 words over these three parts, second, for visiting this site and telling others about it.  I’ve always said that I expected the site to be me clapping with one hand, but it’s grabbed onto something real within a lot of people.  I am certain that I will forget someone if I state a list of people, but to all those who visit, to those who have supported Soccermetrics, and to those who have advised, invested, and otherwise believed in me, thank you, thank you, my sincere thanks to you.