Trying to figure out how data can work on the pitch

Putting figures and young statistical turks together is a good thing

Three years ago Simon Farrant, a man from the sports data company Opta, brought football bloggers and football clubs together. It was the first Optapro Analytics Forum.

The clubs might see some good ideas; the amateur analysts get the chance to work with large football data sets. The sort of place where physicists apply themselves to football; where you find out if a goalkeeper’s height is relevant when asking how good they might be at dealing with crosses.

This idea of putting data and thrusting young statistical turks together is a good thing, and the third edition of the forum took place last month.

To get your hands on the data and present at the forum it’s a matter of proposing an area of the game that you’d like to examine, along with the dataset you want to request, and keep your fingers crossed that the judges smile upon it.


Those five judges included four from football clubs – an important step, guiding the analytical efforts towards things clubs might find useful. Things that might lead to winning, not science projects.

Football clubs from across Europe were there in various guises; most listed as being directly with a club, other attendees via what one might call a shared analytical concern. Smartodds, covering Brentford and FC Midtjylland, owned by Matthew Benham. Arsenal, who own StatDNA.

Career curve

Topics presented included the concept of surprise in attacking play, measuring how players control areas of the pitch, as well as whether playing time at a young age at the top level affects a player’s career curve.

Sam Jackson, a student from the University of Sussex, found no statistically-significant correlation between a goalkeeper's height and their performance dealing with crosses, at least when it came to every cross in the Premier League, Bundesliga, Serie A and La Liga over the 2014/15 season.

Judging the quality of a goalkeeper is not just about saving shots. It’s also about controlling that penalty area and, for some of the new breed of sweeper-keepers, a good way beyond it.

Jackson’s metric rated goalkeepers on how aggressive they are at coming to collect crosses and how successful they are at doing it. According to the metric Thibaut Courtois is at the more aggressive end of the scale while still claiming 10 per cent more than average.

It’s more subtle than it might seem on the surface. Generating a stat on how often a goalkeeper claims a cross? Straightforward but not always useful, and potentially perverse if a goalkeeper was incentivised to stick to the easy crosses.

Generating a stat on how often that goalkeeper claims a cross into a particular area of the penalty box compared to the average goalkeeper from Europe’s top leagues? That’s more insightful, and quickly understood. It recalled work in the US measuring fielding effectiveness in Major League Baseball. The ball might be a different size but the question, at its heart, was very similar.

Then a question from the crowd. “How would you go about feeding that back to a goalkeeper in a club environment?” asked a head analyst from a Premier League club. It’s a key question. He continued: “You mentioned incentivising the goalkeeper, so how do you get that message across to the goalkeeper? What you’ve mentioned is a very statistical and very detailed explanation, which is good, but how do you simplify that? ”

Variants of this question arise repeatedly in this field: nice work, but how do you impact the playing pitch?

It would be addressed by guest speaker and NBA analyst Dean Oliver. Bringing a basketball man across the ocean to a room full of football people might seem strange but there is value in hearing about specific issues encountered in different sports; lessons learned with wider meaning.


Being able to speak to the player or coach was simply about knowing the sport and knowing its language, said Oliver, while leaving the numbers behind.

By all means use numbers and analytics to inform ones opinion, but the language that players and coaches understand is the one of their own sport. And it must be related to winning. Not science projects.

And if you want to really drum an analysis message home to a player?

Show them something that affects how they get paid, said Oliver. They’ll make that connection pretty quickly.