Using Big Data to discover the next Lionel Messi

According to TV pundits, there are many factors that go into winning a football match: flair, guts, tactics, set-pieces, luck. There’s another one that probably isn’t on Robbie Savage’s list of clichés, but most definitely should be: data.

Using Big Data to discover the next Lionel Messi

In today’s football industry, decisions about which player to spend £35 million on aren’t based purely on the manager’s gut reaction or the notes jotted down in the chief scout’s little black book – they’re made after months of detailed analysis tracking thousands of players from across the world. Birmingham-based Scout7 has one of the biggest scouting databases in the business, allowing top clubs such as Chelsea, Tottenham Hotspur and Liverpool to marry their own, highly confidential scouting data with the video analysis and data captured by the company.

Combining all this information and serving it up to clubs around the world, 24/7, requires powerful, custom-built data centres, tight data security and impeccable UI design. We caught up with Scout7’s operations director, Bradford Griffiths, to find out how they manage it.

bradford-griffiths-scout7

Photo: Bradford Griffiths – Scout7’s operations director

Scouting with SaaS

“Scout7 was one of the pioneers of using computer software to identify transfer targets.”

Scout7 was one of the pioneers of using computer software to identify transfer targets, something that was largely confined to players of football sim Championship Manager when the company was founded in 2001. It was also one of the earliest proponents of cloud computing.

Scout7 has a two-tier approach to scouting: it provides its own database of player data and also allows partner clubs to create a scouting database of their own. In the first instance, clubs can search the company’s database of 135,000 players worldwide to create a list of potential targets. This details the player’s full history: appearances, goals, positions, injury history, transfer fees and more. Clubs often have specific requirements – they might, for instance, want a right-back, no older than 26, with an EU passport, who’s played at least 75% of the games for his club over the past three seasons – and Scout7 allows clubs to filter against such detailed criteria.

scout7_video_player1

When the company first started offering this database to clubs in 2001 it took a gamble, because not only was it asking clubs to adopt entirely new methods of scouting, it was asking them to access this data over the internet.

“This was a time when the internet was a fledgling as a business tool,” says Griffiths, “but the decision was made to start collecting global football data and presenting it through an online application. It was ambitious to go for Software as a Service back in 2001, because these scouts weren’t even working at the stadium. They were working the training ground, and training grounds weren’t connected to the internet. By luck or judgement, it was absolutely the right decision.”

“Previously, clubs would have spent tens of thousands flying scouts around the world: now they can watch talent from the PC in the chief scout’s office.”

Since 2009, Scout7 has offered curated video clips alongside the player data. Now, once clubs have a shortlist of players matching their criteria, they can watch the player’s recent matches to help them make an initial assessment of whether he’s the type of player they’re after. Previously, clubs would have spent tens of thousands of pounds flying scouts around the world to watch a player in action: now they can do it from the PC in the chief scout’s office.

Having used the video to whittle down the shortlist, the clubs can then send their scouts to do what they’ve always done: visit grounds and make a first-hand assessment of a player. The top clubs employ dozens of scouts, and potential transfer targets will be watched several times by multiple experts before the manager is asked to watch the player himself. The reports from the club’s own scouts are fed into the Scout7 database, so the chief scout can collate and add it to the information that Scout7 provides to all of its member clubs from its own database. 

scout7-player-profile-harry-kane-crop

This poses two challenges for Scout7. First, it has to make the scout’s report forms as simple and accessible as possible, because even today your average football scout is more comfortable with a pen and notepad than a laptop. “We’re not talking about guys from technical IT backgrounds; we’re talking about guys who might struggle to find things in an internet browser,” says Griffiths.

Second, it has to ensure the tightest possible security, both internally and externally, because at the highest levels of football you’re dealing with intellectual property worth tens of millions of pounds. Each club’s data is encrypted, and only authorised personnel within the club can access the key scouting information. This is achieved through tight management of access controls by Scout7 itself, because “clubs either didn’t trust themselves to do it properly or didn’t want the burden of it”.

Consequently, individual scouts entering data into the system aren’t granted access to the shortlists of the club’s main transfer targets, because these (often poorly paid) freelancers could pass that hugely valuable information on to other clubs. What’s more, if the chief scout receives a better offer to join a rival club, he can’t take all that valuable information with him, because it’s all stored on Scout7’s servers – something that makes Scout7’s cloudbased service even more attractive to the CEOs of the leading football clubs.

scout7_video_player2

Powering the scouting

“The 800TB of video footage in Scout7’s database grows at the rate of 3,000 full-length HD match videos every month.”

Scout7’s scouting database isn’t only used to identify new players, it’s also a source of intelligence on rivals. The video database can be searched to, say, review the past ten goals scored by a forthcoming opponent, or watch a team’s set-piece routines. These curated video clips can be shared with players in team meetings or downloaded onto their iPads, so they can do their homework on the guy they’ll be marking on Wednesday night.

Delivering all this data and HD video footage on demand, to clubs that pay hundreds of thousands of pounds per year for access, is no small feat of IT management. Scout7 runs a mixed environment, using some third-party cloud services and some of its own high-performance Intel Xeon-based servers. “The bare-metal machines are used where we have specific performance requirements that just can’t be delivered through an Infrastructure as a Service environment,” says Griffiths. “That’s really the database side of things, because we do a lot of pre-aggregation of data, so that from a user experience you receive instant results. You need bare metal to get that level of performance.”

scout7_teamsheet_page

The company also uses its own server equipment to store and deliver the 800TB of video footage in its database, which grows at the rate of 3,000 full-length HD match videos every month. “In terms of video, we’ve built up that infrastructure ourselves,” says Griffiths. “To go to a third party and say ‘we need 800TB of storage, it has to grow by this amount every month and we need certain bandwidth’, the costs would be huge. So we manage that within a dedicated hosting environment.”

With customers that include many of Europe’s leading football clubs, which have scouts located right across the globe, the Scout7 database must be constantly available, and the company puts a lot of effort into ensuring hardware failures don’t disrupt its service. “Redundancy is inherent all the way through the environment,” says Griffiths. “For example, the way we set up our hard disk arrays: if any number of disks go pop, everything will keep working. We have hot standby on the databases, so if something goes down we’ll always have good service with the other one. We have off-site backup of the video archive, so that the entire archive is reproduced in a geographically separate location.”

Griffiths says the next step for Scout7 is working with partners to bolster its databases with even more information. Clubs don’t want to log in to different systems to get players’ medical records, for example, they want it all in the one place. Griffiths says the company is looking to partner with other software providers so it can collate that information: “We accept we can’t do everything ourselves.” Football is a team game, after all.

Do you work for a British technology company that could be profiled in Alphr and PC Pro magazine? If so, get in touch: [email protected]

Lionel Messi photograph by Christopher Johnson 

Leave a Reply

Your email address will not be published. Required fields are marked *

Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.