Analytics in football

[danielmak]

I looked through relevant threads to see if this had been posted yet and couldn't find anything. Here is an interesting article in the New York Times Magazine about Liverpool's use of analytics. Most of the comments avoid some of the idiocy that is found in many discussion boards but the commentators do seem to miss the larger point: football has yet to figure out how to apply analytics to game action because the game is so dynamic. But analytics can do a lot to identify players who are undervalued.

https://www.nytimes.com/2019/05/22/m...liverpool.html

[Antepli Ejderha]

An interesting read. A bit too US in language at times but I think it shows how far we're behind in using data.

[Vicarious Thrillseeker]

I'm not sure. I have a few questions: Didn't Liverpool embrace analytics and data previously and end up with Andy Carroll and Stewart Downing? The Origi/Alexander-Arnold goal against Barcelona had nothing to do with data, did it? And Sam Allardyce - mentioned in this article as a data cynic - is very keen on using data in the game, as far as I recall.

I think.

[ursus arctos]

You are treating "data" and "analytics" as static concepts, when in reality they are extremely dynamic, particularly w/r/t "flow" sports such as football.

[jwdd27]

Sam Allardyce was indeed a pioneer of data in English football, especially Prozone. During his spell as a player at Tampa Bay Rowdies he was inspired by Tampa Bay Buccaneers' use of data, and brought it in when he moved to England.

I think he's always played up to the Big Sam, unsophisticated caricature, whereas in truth his players were drilled to know exactly where they had to stand, and indeed had been scouted and analysed as the best available players to do what he required.

[seand]

I'm not sure. I have a few questions: Didn't Liverpool embrace analytics and data previously and end up with Andy Carroll and Stewart Downing? The Origi/Alexander-Arnold goal against Barcelona had nothing to do with data, did it? And Sam Allardyce - mentioned in this article as a data cynic - is very keen on using data in the game, as far as I recall.

I think.

That was my first thought. The last lengthy piece I read on the superb quality of Liverpool's analytics proudly boasted how analytics had helped identify Carroll and Downing as targets. That was a few years ago, things have moved on but it's not a new area in football and it's been shown to have been really bad for Liverpool quit recently. And the TAA-Origi goal is the antithesis of analytics.

Analytics is certainly under-used in football at the moment, but it is an incredibly difficult sport to model.

[3 Colours Red]

I think he's always played up to the Big Sam, unsophisticated caricature

My take is that he's had to do that to stay part of the Old Boys' Club. If he was blathering on about heat maps and xG and what have you, he'd have been shunned and his pension kitty would be noticeably smaller.

[Bored Of Education]

I was thinking this the other day and wondered how much analytics confuses correlation with causation but it has advanced so much that that must all be taken into account. However, I think that the reporting of it often falls into that trap. As an odd aside, Rhys Long, the head of performance analysis for the English FA (and formerly the WRU and Lions) and Michael Hughes who has done Performance analysis for the Lions, England Rugby, British Cycling and England Squash both come from my home town. Maybe Porthcawl is Analytics Valley, erm, Beach..Seaside...or something.

[Antepli Ejderha]

What's the difference between data and analytics? I'm confused and have confused the two.

[ursus arctos]

"Analytics" is the current (largely North American) term for the collection and analysis of all kinds of data relating to sport.

On the one hand, "analytics" has existed ever since people began keeping score (data relating to sport). On the other hand, the volume and type of football data available for analysis has increased exponentially over the last few years, with many (but not all) of the techniques having roots in basketball and ice hockey.

[Antepli Ejderha]

"Analytics" is the current (largely North American) term for the collection and analysis of all kinds of data relating to sport.

On the one hand, "analytics" has existed ever since people began keeping score (data relating to sport). On the other hand, the volume and type of football data available for analysis has increased exponentially over the last few years, with many (but not all) of the techniques having roots in basketball and ice hockey.

Cheers. Makes sense. So the NFL testing that's broadcast live is a massive exercise in gathering data?

[ursus arctos]

It is a subset of the data that clubs are already collecting for their private use (this is also true of "analytics data" broadcast in the other North American sports).

The fact that clubs and analytics companies regularly take data (and data collection methods) out of the public domain is one of the real challenges facing its development as an academic discipline.

[Antepli Ejderha]

Is that to protect their methods or for a more sinister reason?

Cycling and the idea of marginal gains is when I first understood how much was being invested in this. I only saw the US sports as being the collecting of stats for geeks but it's obvious it's so much more than that.

A really interesting area of sport that doesn't get too much coverage.

[ursus arctos]

It is primarily a reflection of the fact that both the data and analysis have been demonstrated to have a very significant monetary value in North American sports.

Every club in the four major North American leagues (and MLS) has an in-house analytics staff that can range from a one or two people to dozens and they all subscribe to and/or own interests in proprietary entities that both collect data and crunch numbers. It is easily a three digit millions industry in terms of annual spend now.

North American sports have led in this space for two basic reasons: 1) salary caps and other restrictions on "traditional" ways to gain a competitive advantage have encouraged clubs to spend money on concepts, data, equipment and personnel that aren't subject to such restrictions and 2) many of the current generation of North American club owners made their fortunes using roughly similar techniques in other fields (John Henry (of Fenway Sports Group, the Boston Red Sox and Liverpool) being a prime example, but by no means unique).

[danielmak]

I assume there was some level of assumption happening in the article itself that North American readers who care about sports know that analytics have become an important feature of all sports but the article downplays a variety of tools used by even the most conservative teams. For example, the monitoring of player health, the distance covered, when to press and when not to press. All of these data points affect how teams strategize. But as I posted up front, football presents problems for implementing analytics as opposed to baseball, for example. Football has too much unpredictable movement all happening at once. There is no equivalent of a baseball shift or heat map for pitch location.

I know this was discussed in one of the sports threads, but one thing I hope will happen with these sports science departments is a shift away from different tactics for home games and road games. I get that when an English side plays in Kazakhstan in the Europa League or an Argentine side travels to Bolivia in the LIbertadores that the travel (or altitude) can be such that different tactics are needed. But when Saint Etienne travels to Lyon, the tactics should be the same as playing at home. I'm like a broken record with this one but road fans don't play the game.

Anyway, I'm sure there are market inefficiencies that can be exploited via smarter and more complex analytics. And if analytics led to the purchase of Andy Carrol then that's a problem with the specific data that was used. We see similar mistakes in baseball where some teams think a player's value over a shorter period of time is worth the investment. Carrol was poor with Liverpool but also hurt a lot. Some baseball teams would be fine with a player being hurt since they like what they get when he's healthy. The LA Dodgers seem to operate with this model. I think it's a mistake. Liverpool's signings since Klopp arrived have mostly hit so something's working right for now.