Is Performance Analysis drowning in raw useless data?: A response
(Lost blog 1/3: December 2012)
Note: Following a loss of data during VPA migration, I have re-housed the original blogs here. Unfortunately, the abundance of of accompanying comments, which substantiated the blogs are not available, however it forms a useful reflective exercise preceding #UCSIA15 #TVON.
Jason Lear posted another insightful blog late last night, which offers his viewpoint on big data in performance analysis alongside posing relevant questions
Following a tweet by @CPAUWIC (now @DMCPAP), there were a couple of responders who both confirmed the importance of asking performance-related questions to establish performance-related answers. Thanks to @Uppy01 and @SimonGleave.
I regularly get inspiration from Bernard Marr, a leading global authority and best-selling author on organisational performance and business success. This is coupled with this recent publication “Counting what counts: collect the right numbers to make the right moves.”
Within this article, Marr “compares the data-usage approaches of most organisations today to the crude tactics of California gold rush prospectors, who sifted through tons of silt and dirt to find a few precious shiny nuggets”. Jim Kimpton reinforced this view with “we can collect lots of information, but we don’t necessarily always know why we’re collecting it. If we narrow it down and connect it to our strategy, we’d probably be much further ahead.”
Within sport, have we matured to this point yet? I offer this storyline as a starting ground for discussion…
Sporting Performance can be observed and annotated/recorded to provide valuable information. The variables within sport can be combined into KPI, with a degree of selectivity by coaches/analysts. These will tend to have a numerical outcome, which can provide indication of performance.
When linked to video, the data is immediately functional to provide indications of single performances for learning and reflection. A small sample of matches (such as a tournament) can also provide key platform for benchmarking and direction. Beyond this point, it can become unwieldly and misleading as sport is so volatile.
Sporting performance is highly volatile due to many compounding elements such as opposition quality, venue, weather, stage of season, injury, etc. along with periodical rule changes. Therefore, big data may only be relevant for particular time windows.
However, longitudinally, sport doesn’t change and most people look at the same things. The fundamental principles and KPIs are relatively stable. Some authentic coaching teams put their own interpretation of the rules and the game can be impacted as a result of novel tactical interventions (which can transform sport — e.g. more organised defensive systems in rugby requiring a shift in offensive strategies). Some coaching teams are capable of exploring and extracting the key components that influence on performance outcome (that actually be influenced).
The key is to database the data in a flexible manner to create an interactive resource to deeply explore the data visually and dynamically. Beyond this, a prime purpose of big data should be to use it to provide a reflection of current performances against norm values in previous matches.
Through observation, arguably, some people are either trying to be too clever with the variables, some just try to make it look pretty… Some people don’t get the math, some the performance element, some the visualisation element… The critical balance is drawing all these components together powerfully. This is not necessarily rocket science; in fact simplicity could the very thing that is overlooked? The goal is ease of use and palatable applicability to deliver meaningful information, that can be interpreted in the coaching process to provide evidenced reinforcement of decision-making and ultimately impact on performance. At this point, coaching teams will be more receptive to the information available.
I have some ideas to the solution… Do you? TBC…
