This is a post I’ve been writing for a while, where we take a step away from the detailed work and take a 10,000 foot view of data in sports. I want to outline some key points about using data within a sporting environment. Again , I would like to reiterate this is simply my own opinion and not the only way to do things. If you have ideas that you feel are worth mentioning here I would love to hear them as well!
Fast and Slow
Sporting environments can be very fast-paced chaotic places where staff are under pressure each day to perform their tasks. As such it is very easy to become so focused on the short term that we leave out looking at the bigger picture. That’s where we need to either have the ability or enough staff (if you’re lucky enough!) to both produce those daily reports or dashboards but also keep one eye on long terms trends, dig a bit deeper into the data, along with identifying areas of future research/analysis. (See here for a nice article on this concept).
Anyone who has spent hours working on THAT perfect Excel spreadsheet or RScript, knows the selfish pleasure of being able to press refresh/run and watch the magic happen. Yet this part of the analysis is ultimately meaningless if we are unable to both draw insights from the data and communicate those insights to the people that want to or need to hear them. Often it can be difficult to draw meaningful insights early on and they come more as gut feelings than definitive statements but having conversations with more experienced or knowledgeable staff can help turn these gut feelings into actionable insights (or you can shot down but that will never change no matter the experience level!)
Following on from drawing insights from your data is how can we communicate those insights? Regardless of the sporting environment this is often a tricky issue as we can be working with staff who do not have the same understanding of the various metrics we are reporting upon and can fail to see the relevance (if present) of them. Similarly, many sport coaches may have relied upon techniques or training programs without much data collection present early in their careers and can be hesitant to include it later on. In these instances it’s easy to fall into the trap of believing everyone else is at fault for not seeing the value you can provide whereas the reality is if the data your collecting and analysing isn’t providing value, the onus is on you to work towards changing this. Often this can start with small conversations about anything besides the data to let people become comfortable speaking with you in general and not view you as ‘just‘ the data person. When they are comfortable speaking with you the next step is to meet them on their level by asking their opinion or ideas how to improve your practices so it can start to have a degree of impact upon daily practice. Never underestimate the value that a coach’s or an athlete’s own intuition can provide over and above that of ‘the data’.
Fail Fast, Fail Early
Continuing on from the earlier theme of THAT perfect excel sheet, who has suffered the soul destroying experience than can come following the final reveal of your analytical masterpiece spreadsheet where blank expressions do nothing to hide the complete lack of interest. Regular conversations with the data’s end user are key when building your analysis model. Speak to them throughout the building process about whats of interest to them and how best to show the data to them, even show them samples to get early feedback. By all means include what you feel needs to be present but never forget that any analysis offers little if it cannot be communicated to the end user. Along with the theme of fail fast, fail early comes the idea of “Iterate, iterate, iterate”, where we are constantly in a cycle of progression and revision. Each step we make, we have conversations about, everyone feeds into the development process to guide the next iteration.
Similar to the idea that the analysis is meaningless if we cannot communicate what it is telling us, we can also fall into the trap of becoming robotic in our analysis. Following each step on the analysis path without really thinking about why we do it a certain way, thinking about the benefit and meaning of the analysis. We need to clearly understand why we are doing something before diving head first into a swamp of copy/paste, refresh, pivot tables etc. This is often a trap interns can fall into, or perhaps are let fall into, where they become too focused on carrying out the work they fail to see the the reasons and meaning behind it (Yes, I did everything to avoid “Start with Why’ here ;)).
It’s very easy these days to buy into the reliability of all our data sources or analyses. Most companies will employ sales people to give you a well tested spiel about their product and how trustworthy it is. However if we are going to use these products to make informed decisions about our athletes I feel there is an onus on us as practitioners to ensure reliability of said methods. Personally, I consider myself an optimistic skeptic when it comes to the issue of reliability. By this I mean I believe your product or analysis to do as you say but I want to take a look under the hood myself.
How many of you have sat there and either listened to or spoken about an athletes acute/chronic ratio being too high without knowing the reliability of the metric in question?
- If we take High Speed Running and the common AC ratio upper limit of 1.2, how accurate are GPS units at recording high speed movement? Well most would seem to suggest somewhere around OK to good reliability (See here). So if we have good reliability, meaning there are some reliability issues present, is an AC ratio of 1.25 really cause for widespread concern here? Or is the version of the ratio you are using even appropriate for your sport/environment/data?
Don’t be afraid to question commonly held beliefs, just be rational in your thought process!
More to Data than Load Management
As practitioners we can often be consumed with that which can be measured and analysed, shying away from areas we cannot quantify objectively. This can lead to focusing on the measurable areas with ‘Load Management” being a main culprit here. While there are vast columns and samples of data analytics in sport science around the concept of load management and the reasons behind the large interest in this area are often written about, it’s important that we remember as practitioners we are ultimately preparing athletes to perform. Many performance sports require athletes to walk a fine line between appropriate workloads and competition-ready. Can we as practitioners move to a place where our load management approach is designed to optimise performance through assessing sport demands and training outcomes rather than focusing on injury rates/prevention at the expense of performance preparation.
Not Everything Can Be Measured
In the same theme as the previous point, while striving to provide objective data to support much of our decision making is a positive, it’s important to recognise that when dealing with such multi-faceted areas such as performance or injury not all aspects can be quantified. We must recognise the need to step away from numbers and charts and use our own experience to see when athletes need to be pulled back a bit or pushed harder. Often brief conversations had early in the morning can be as beneficial, if not more, at gauging an athletes condition than the variety of tests available to us. Similarly when a group of athletes that are normally quite loud and talkative turn up for training in a quiet, lethargic state it can tell you more than any readiness assessment.
Hopefully I have given you some food for thought on the the use of data in a sports environment, I would be really keen other other peoples thoughts on what are common pitfalls in this area or even examples of where you feel data is being used in a really positive manner.