I am always (pleasantly) surprised by the people who visit the site and I enjoy it even more when comments are left. Last week I wrote a piece calling on the software suppliers to improve their usability/visualization of information. A few hours after publishing the piece I was left a comment by Kevin Goodfellow from Sports Data Hub. I had never spoken to Kevin before but was delighted with his feedback, despite the fact that he poured cold water on my hope for a one size fits all approach.
I was intrigued by Kevin’s comments so decided to email him with a retort. He promptly replied and I couldn’t argue with his view point – which effectively was that it is too much to ask one company to do all the stages of the analysis process. I had never thought about breaking the analysis process into such distinct steps before.
You should also check out this video of Kevin presenting at the MIT Sloan Sports Analytic Conference where he outlines his thoughts.
A New Model
Some people like models and others don’t think much of them. I find they help you compartmentalize what are sometimes difficult concepts. This model allows you break the 3/4 sections of the analysis process into the box that it belongs in and think about them independently. We all do this whether we think about in this way or not. We all Collect – Manage – Analyse – Present information & video. And certainly over the last few days I have been thinking more and more about these functions independently rather than looking for a one-size-fits-all approach.
4 Step Process
I have added in a 4th step called Data Collection but either way the concept is the same. In fact as I began to look at all the tasks involved in each step I realised that it is likely that no software will ever do all 4, and in fact if a company tried to do everything it would likely be too big and cumbersome and would severely limit your scope to adapt it to your own needs. Also thinking about the steps seprately allows you change one or two as your budget changes. I will give a brief overview of each section.
I would include the analysis software and 3rd party sources in this. Software like Prozone, Sportstec, Dartfish & Longomatch and many other would fall into this category. I would also include data collection sources from mobile devices like the IPad and IPhone in this category. The last option to consider in this step is data from 3rd part services; companies like Opta & Prozone in the UK would fall under this category but also now many sports produce stats that are freely available on websites.
Note: Think about how important it is that the collection is as easy as possible as well as how the collection method fits into the other steps of the process.
Kevin listed data management as the most important step in the process and I can’t disagree with him at all. In fact if you get this bit right it makes life so much easier. Data Management can often get overlooked as it takes time and it is certainly not the fun part of the job. Nobody is going to be down the pub on a Saturday night telling stories of Data Management! But getting this step right is crucial. There is nothing more unproductive as working your way through excel sheet after excel sheet trying to cut and paste data from 20 different sources.
I am currently working on quite a big project involving a huge quantity of data and the management of this data is proving the hardest bit to crack. Most people will use Excel but maybe for the bigger analysis departments they should look more at databases.
Note: Spend time on this part of the process it will make the next 2 steps much easier and ultimately save you a huge amount of time and will mean you can work more efficiently.
This is the statistical analysis we do on data. At a beginner level you are probably looking at some totals, averages and trends over the last 6 games. At a more advanced level you might be calculating standard deviations, variance or conducting regression analysis.
Note Regardless of the level of analysis you do it is important that the way you collect and manage the data makes life easy here. If you have collected data from different sources manage it in different ways – it makes real statistical analysis very difficult and cumbersome.
This is the exciting bit – and the piece of the puzzle that the players and coaches will see most. Certainly for video analysis you are mostly going to use the same software you used during the collection process. That is all fine but it might be worth looking at other methods (software or hardware) that can make this process more coach/player friendly. i.e. Think of the IPad as visualization tool. I was also intrigues to learn that there is software out there specifically aimed at improving your visualization. They take the raw feed of your data and give you loads of different options and templates to present your software differently.
Note: Data Visualization is the fun part of the job and it is the ‘front of house’ element to your job, but as Kevin said in his presentation if you get the management part wrong the rest becomes difficult, cumbersome and time consuming to manage.
This model has really got me thinking over the last few days – rather than looking for a tool that can do all of these maybe we are better of treating the tools we use for their speciality. As analysts a large part of our job is finding the best solution for each piece of the jigsaw – one that meets our budget and our expectations and managing that process as best we can.