How the English Football Association is building a player analysis tool with Google Cloud
- 03 June, 2019 06:20
The Football Association (FA) has entered a long-term technology partnership with Google Cloud to centralise the data it holds on footballers across all levels of the game and to build a set of tools which will allow coaches quicker access to player insights.
The multi-year deal will see G Suite installed as the collaboration software of choice across the organisation and data being migrated to Google Compute Engine and Big Query to power analytics.
The FA says it currently has multiple terabytes of data stored within Google Cloud that is accessed by its analysis teams. It wants to expand this analytics-driven approach now by building a proprietary tool on Google Cloud called the Player Profile System (PPS) to measure performance, fitness, training, and the form of players at all levels.
Dave Reddin, the FA's head of team strategy and performance told Computerworld UK that as wearable technology has proliferated the whole industry is living with "an explosion in the availability of data", but admits that football is some way from fully taking advantage of that yet.
"As with most markets what has been the most exciting development has been insight and not just collecting more and more data," he said.
"I definitely think we are a long way from the really meaningful stuff we look to get from PPS. We are in the midst of a data wave but we are not yet working across those streams and getting the more non-linear or non-numerical insight."
Reddin formerly held roles with England's successful Rugby World Cup winning side in 2003 and was involved with Team GB during the London Olympics.
Now at the FA, he is responsible for strategic and performance planning for Gareth Southgate's England men's team, including oversight of performance analysis and insight and player insights. In 2018 he was accused of bullying and implementing a "culture of fear" by an anonymous whistle-blower, before being cleared after an internal inquiry.
The FA now wants to enhance the PPS to gain more real-time data analysis, allowing coaches and analysts to better compare and analyse team and player performance. It will look to leverage Google Cloud smart analytics, data management and machine learning predictive maintenance tools to achieve those goals.
Expanding on this, Reddin talked about being able to not just assess historical performance, but starting to project forward for better game planning.
"We want to see how effective our interventions are on other styles of play," he said. "How to break down a low block or how best to attack at set pieces. We also want to project forward and use data to apply machine learning and model these things forward and not just analyse retrospectively."
Speed to insight is clearly a driving metric when it comes to a technology decision like this, but Reddin is realistic about how the FA may be able to leverage advanced techniques like machine learning in the near term.
"I hear a lot of talk about machine learning and the insights it might provide but we are still in the hard yards period of this journey," he added. "We need to get the data captured and structured correctly to make machine learning effective.
"As a national team we don't have the volumes of data clubs might have. Although we do consume Premier League data, it's not quite the same thing. With Google we see fantastic opportunities with a company with great expertise in that area."
So far The FA has been able to consolidate most of its player data into a single GCP system, from which it is able to launch specific products for end users to get insight from that data.
One key early project with Google that is nearing completion is the creation of a new app for coaches.
"We want to visualise the parameters we collect on a player in a way that is easy to interrogate for coaches," he explained. "That resides in multiple repositories right now, some legacy and some quite crude systems, so consolidating that for that audience is the primary piece."
The result will be a means to quickly compare key characteristics of players to help inform selection decisions. So when England play the Netherlands next month in the Nations League, do they pick a ball-playing midfielder like Jordan Henderson, or a more hardened performer like Declan Rice?
"We want to create the ability for a coach to supplement their expert view with a more data-informed view," he said. "This isn't an algorithm for player selection and never will be."
This tool is currently in the final stages of product testing and should be ready to launch in "the next few months" according to Reddin.
Reddin talked about the advantages of having Gareth Southgate as head coach due to his history with the FA, having coached the Under 21's before getting the men's senior head coach role in 2016. He said that Southgate has been around for the entirety of a broad organisational shift towards being more analytical.
"Gareth has been part of the organisation for a significant amount of time now, so he hasn't directed [this approach] but he has been a big supporter," he said.
When Reddin joined the FA six years ago he said he was somewhat shocked to find an organisation where all information was stored on on-premise drives and across numerous Excel spreadsheets residing on people's desktops.
"E-mail and version control culture was rife," he said. "The planning process for everything from a World Cup to a standalone training camp all now happens using G Suite."