11 Questions with @thecomeonman

The pursuit of wisdom in any walk of life quickly reveals that what you think you know is not nearly enough to get you to where you want to go. As I'm starting out in my football scouting journey I challenged myself to reach out to those already working in various roles in football to answer a short list of questions. My goal wasn't to get answers but relevant perspectives on the game within the game.

Here is @thecomeonman:

How did your first opportunity in football come about?

My first and only opportunity has been through The Scouted Hub. I'd responded to @VillaAnalytics' call for some help with some visualisations on Twitter late one night. With a time difference of quite a few hours between us and my phone out of action, that was quite an exercise in coordination let alone the visualisations themselves. I've since helped TSH with some analyses and visualisations once in a while for some of their projects.

The bulk of my work is mostly for my own amusement and is either on github, at https://github.com/thecomeonman, or my blog, https://thecomeonman.blogspot.com/search/label/Data. I'd definitely like to do more work professionally and hope I get to work full time in the industry some day!

What attracted you to scouting/analytics? What’s more intriguing now names or numbers?

I'm quite a sports buff and enjoy playing multiple sports, in addition to which I also enjoy playing fantasy football, FIFA, and have been working on data science and data visualisation related problems since 2010. That set of hobbies is a potent mix of algorithms, data, and sports and led to football analytics becoming a serious hobby in the last couple of years or so.

I'm more intrigued by numbers. I feel like names will come and go but numbers and the logic to process them are more permanent, more fundamental in nature, and also more scalable than names. I wish I was better at names than I am though. Data is still scarce and often confusing or misunderstood whereas names can be checked and can be trusted far more than numbers.

Who/what is the first player/concept you "found"? What caught your eye?

I'll skip the little bit of fantasy football work I'd done much earlier and consider my spatial passing similarity model as my first concept - https://thecomeonman.github.io/SpatialSimilaritiesBetweenPlayers/. Looking at all the flexibility with which various player roles were being interpreted by various managers around the world, I thought there was a need to quantify what formation a team was actually playing in. In that process, I ended up finding more uses for that model including in identifying player roles, identifying similar players, and identifying team strategies and published a bunch of posts on that methodology.

Who/what is the player/concept you "missed" on? What did you learn from it?

xT, something devised by @Karun1710. If you haven't heard of it you should read https://karun.in/blog/expected-threat.html. 

I was planning to follow up my spatial passing similarity work with a model very similar to xT. Halfway into it I started fretting too much about the finer details - how the model didn't incorporate so many other things that affect football games, how some assumptions the model makes didn't hold true in football, how I should probably use tracking data instead of events data, etc. and gave up and even considered adding disclaimers to the passing similarity model. What I overlooked was that even if those details weren't satisfactorily addressed, it would still be a very useful model when trying to get high level trends. A lot of these questions don't have satisfactory answers from Karun's implementation of xT either but it still finds high level patterns and that's useful enough.

My learning from this was that getting bogged down with theory and problems that you can't solve is a sure shot way for a project to not end. My advice to anybody else in a similar situation would be to simplify things, make reasonable assumptions, see if the results make sense and if they do then put it out a v1 and keep working on solving the niggles that are still bothering you.

If you could start over what skill would you build on first?

 

Building good looking interactive visualisations. I feel a significant part of what gets people interested in your work is how cool it looks. 

The unfortunate flip side to this is that more interesting / detailed / rigorous content without fancy visualisations often doesn't get the eyeball it deserves too. This may be a social media thing, which is where my observation is mainly from, or an outcome of working with an industry which is not very comfortable with numbers yet.

Do you see player development as more of an art or a science? Is development on the club or the player? Why?

I'm gloriously unequipped to answer either of these question but they feel like the sorts to which the answer would be - the right mix of both is the best. That's what my answer would have been if you'd asked me the more general question about employees and employers.

What is your favorite sports moment? Why?

I'm struggling to think of an overall favourite but in the recent past, it is probably Barcelona's 6-1 win over PSG. To be 3-0 down to begin with, and then being in the 87th minute of the game and still needing 3 goals to qualify and then actually doing it was unbelievable!

What coach/player/team inspires you? Why?

Can I switch to tennis for this question? I've been a huge admirer of Roger Federer  - the grace with which he plays, the longevity of his career which must need lots of discipline, commitment, and motivation, his demeanor on and off the court, how he is able to have a family life despite the demands his career must place on him, and all that he has achieved, these are some of things that I find incredible about him.

What advice would you give to someone wanting to get into media/analytics?

For those getting into analytics - don't try to do complicated stuff just because you can do complicated stuff. But if you only know how to use a hammer, and it could be a simple k-means hammer or a complex deep learning hammer, every problem will look like a nail. With the sort of tools available today, you could teach a monkey how to build some model. The difference that you make is in understanding what inputs would help, whether the nature of the model is compatible with the problem you're trying to solve, whether the working of the model makes sense, whether the outputs make sense, what can be improved, and so on.

And, of course, presentation matters.

What is your favorite app/tool to use (fun only!!)?

I'll pick Pocket, it's an app that you can store your online reading list on. There is so much interesting content that people publish, football or not, that I'm always in catching up mode and I would have struggled to keep up without that app. It's been a faithful companion on many long journeys and family functions. If you're being very strict on the fun only condition then I would probably pick one of the Stick Sports games like Stick Tennis or Stick Cricket.

What other sport/hobby/discipline do you feel improves your work as an analyst? Why?

Playing the sport and also FIFA definitely helps me understand some aspects of the game better, the strengths of the players better, and most importantly engage with the sport beyond just watching games.