Moneyball: Could sabermetrics offer Sunderland a brighter future?

Trends, be they in clothing, music, film or football, tend to operate in cycles. Certain styles come into style, becoming the ‘in’ thing, before, eventually, something else comes along to replace it. But if you wait long enough what was once ‘out’ soon comes back ‘in’ repackaged as something new and innovative. The reality is that there is very little – if indeed anything – that is new under the sun.

If you doubt me on this, take a deeper look into footballing history. While every era comes with a new theory that is claimed to transform the sport, if you look hard enough you’ll find examples of the ‘system’ having been used previously.

Take for instance Guardiola’s ‘total football’ which is heralded as transforming the game. Yes, it’s good to watch but it is eerily similar to descriptions of Puskas’ Mighty Magyars and the football that teams under Johan Cruyff played in the 80’s and 90s. And it’s just one example. Is Klopp’s ‘heavy metal’ style really any different to the Liverpool teams of the 70s and 80s that pressed high and hard? It’s also readily forgotten that Alf Ramsay first pioneered the narrow style of play that many teams adopt today with his ‘wingless wonders’.

As these countless examples prove, many ‘innovations’ are actually illusions. At best they are reinventions, largely because the sport hasn’t fundamentally changed in over a century.

Photo by R. Powell/Daily Express/Getty Images

---

Data: the bottomless pit

As such, it becomes tempting to view the latest football trend – a reliance on data analysis to make key decisions – as another ‘here today, gone tomorrow’ theory. After all, the use of data in of itself, has been in existence since the beginning of time.

At its most basic level, it is hard to imagine that back in the late 19th century that then Sunderland manager Tom Watson did not take into account, when selecting his Team of all the Talents, that Johnny Campbell, having scored a substantial number of goals before, was therefore the most likely player to score again. Ned Doig, who kept a remarkable 87 clean sheets in 290 top flight appearances, was presumably the first name on the team sheet because he had a track record of not letting in many goals!

Of course, since then the amount of data gathered and utilised in the sport has grown, right up until present day where the amount of data available is seemingly endless.

Roker Report readers will be more familiar with some of what follows than almost any other football fans, because of Roker Report’s admirable dedication to publishing data analysis. So, some of you will have to bear with me as I list out just some of the data points that are now commonly available for analysis, some outside clubs and some inside clubs.

Expected Goals (xg); challenges won; key passes; distance travelled; individual actions; distance travelled in sprints; ball retention; action areas and heat maps; regular patterns of play; heart rate; Vo2. Some of these stats relate to the team as a whole and some to individuals. And whereas historically there was broad consensus that data points such as goals scored for and against, and shots (on and off target) were relevant, there is no such consensus around which modern data points are or aren’t important, or how to utilise them.

As a result, this rich depth of available data makes the temptation to dismiss analytics as ‘just another trend’, demonstrably and, in football terms, potentially fatally (for those making that judgement) wrong.

Photo by Mark Fletcher/MI News/NurPhoto via Getty Images

---

Moneyball – a lesson in utilising data effectively

By accepting this, we shift the crux of the matter away from whether data collated from previous appearances is used, onto something altogether more pertinent. Instead, the relevant questions become:

1. How deep a level of data does a club utilise?
2. How central is the analysis to decision-making, as compared to the use of “trained instinct”?

A blueprint for how these addressed can be seen by taking a look ‘across the pond’ and how American sports teams have approached mining that data to improve results. Of course, the most famous example here is Billy Beane and the Oakland A’s baseball team as brought to life in the Hollywood blockbuster ‘Moneyball’.

Having narrowly missed out on making the World Series, the Oakland A’s lost a trio of their best players to free agency, unable to offer them the high salaries that other teams were willing to offer. With a limited budget, Beane, the teams GM, hired the unknown Paul DePodesta, who was developing a new method for assessing players abilities: sabermetrics.

DePodesta’s method used statistical analysis to identify players that were undervalued based on traditional scouting assessments. By utilising data, he was able to find players that, when you combined their ‘undervalued’ skillset with other players with an ‘undervalued skillset’ would form a competitive team, capable of challenging, and outplaying, their big budget rivals. The approach would lead to the As recording a record setting 20 game winning streak and make it to the play-offs, although once again they fell narrowly short of their ultimate goal.

Following the success Boston Red Sox owner John Henry tried, and failed, to hire Beane to take over as a GM. Henry goes on to order his management team to copy Beane’s methods, which ultimately leads to the Red Sox breaking the ‘curse of the Bambino’ as they claim their first World Series title in 86 years. It’s a feat they have repeated three times since, making them one of the most successful baseball teams of the 21st century.

Photo by MediaNews Group/Bay Area News via Getty Images

---

Replicating moneyball

Not surprisingly, then, when John Henry bought Liverpool FC in 2010 he insisted that the club implement the techniques learned from baseball. Gradually, LFC started to build up a data analytics team, but in itself that was not enough. Unlike with baseball where there was an existing model to work from (Beane’s Oakland As), in football Liverpool had to pioneer how to use data analytics in an effective way, applicable to the sport. Furthermore, they had to persuade the ‘trained instincts’ people not just to listen (with Henry as owner, they had to listen), but actually to base their decision-making on data.

If you think that that should be straightforward, think again. In a recent New York Times interview, Liverpool’s data chief Ian Graham – who has a Cambridge doctorate in theoretical physics – said that he only really persuaded Jurgen Klopp in 2015 that he knew what he was talking about when he demonstrated, by the use of stats, that Klopp’s Dortmund team had actually been very unfortunate in his final season there.

He started to talk about individual matches in such detail that Klopp asked if he had been in the stands. Graham said he had not seen a single one of the matches, live or on TV, and Klopp became a convert there and then. He said last year: “The department in the back of the building are the reason I am still here.”

The Liverpool example is now a very famous one. But just over the North Sea is an even starker case, this time involving Beane himself. In the Dutch Eredevisie, Ajax’s closest challengers are not Feyenoord or PSV, but AS Alkmaar, who are just three points behind. Beane is Alkmaar’s football performance and recruitment adviser – backed by statisticians reporting to him in the US – and he has achieved this feat while making a net transfer market profit of more than 100 million euros.

An example better-known to English football fans will be Brentford FC, who are competing against Leeds and West Brom at the top of the Championship despite crowds of only 12,000. Brentford consistently pick up good players before the market has recognised their potential. Then they sell them once they have reached their potential and re-invest the proceeds in the next batch of potential stars and in paying Championship wages that their gates wouldn’t usually sustain.

The interesting thing is that Brentford do not achieve these successes by hiring scouts and hoping for the best. Instead, a team of data scientists analyse players from all over the world and isolate the small number whose numbers indicate they are better than current market value. Only then is a scout sent out to watch them, to check for ‘soft’ attributes such as attitude.

For instance, we knew that Oxford’s Baptiste and Fosu caused us big problems every time we played them. But it was Brentford who swooped in and bought the two youngsters for a combined £3 million on Jan 31st (it is of course only a coincidence that £3 million is the same figure we spent on Will Grigg). The bet is that both will become EPL players or, at worse, major Championship players, and be worth treble that within 18 months or so.

Photo by Alex Davidson/Getty Images

---

Changing attitudes

However, and reverting to our original questions, it is not so much a question of finding this all out (it’s all on Google if you type in ‘use of data in football’). Nor is it a matter of obtaining the data (easy) or even analysing it effectively (harder, but achievable). The really difficult bit is getting the people who have spent decades in the sport basing their careers on instinct to actually view data analysis as the basis of decision-making, rather than a small part of it.

If you recall this was one of the biggest hurdles Beane had to overcome in Oakland in the early noughties. In one scene in Moneyball, Beane walks into a room full of ageing scouts with the newly hired Peter Brand (DePodesta’s character name). The head scout says something like “Billy, you can’t listen to this guy – you’ve got decades of experience in this room.” Later in the film the head scout again challenges Beane: “Billy if you are going to listen to that kid instead of me then f**k you”. Beane ignores the views (or hurt feelings) of the ‘trained instincts’ veterans, sticking to his guns, responding: “OK, now you’re fired”.

---

Making the case for sabermetrics at Sunderland

So how should clubs like Sunderland view this gradual revolution?

In my view, we should view it with total excitement, as a tool to propel us back up the leagues and, perhaps, to enable us to compete with the very biggest clubs. Unlike paying huge transfer fees, or having global scouting networks, the collation and analysis of data is cheap and available to most clubs of any significant scale (see Brentford). A team of data analysts headed by a former professor costs less a year than, say, a Bryan Oviedo.

There are significant challenges, though. The club’s owner has to be brave enough to really ‘go for it’ and only employ football management who are fully bought in to these new methods. In all honesty, given his bread and butter business is in insurance, an industry dominated by statistic-based risk analysis modelling, I’m shocked that Stewart Donald hasn’t already implemented this approach. It’s a model, in the broadest sense, that is familiar to him and offers a way of reducing risk and maximising value from a business perspective.

But it isn’t just the owner and football staff that have to change their attitudes to make it a success. We, the fans, also have to change our outlook and lose our obsession with star names and how much money is spent on individual transfers. Instead we have to become a bit more patient as players develop (hang your heads in shame those, who like me, wrote off Luke O’Nien after one appearance against Sheffield Wednesday!).

After all the evidence is in front of our eyes and screaming out at us for attention. In the last year, Grigg, bought based on ‘perception’ cost £3 million and is now worth a tiny fraction of that fee. In contrast Jordan Willis arrived on a free and would probably command a fee in the ballpark of £3million.

I was listening to a podcast the other day when the generally astute Kieron Brady said that although the players brought in to SAFC in January looked as if they may be an improvement, his big concern was we hadn’t spent enough. As he said this, it struck me as being eerily similar to the ‘traditional’ scouts of the Oakland As that opposed Billy Beane and Peter Brand as they changed the sporting world.

Ultimately, if, on a reliable basis, we want more Willis’ and fewer Griggs then surely we need to be part of the future, rather than clinging to the past. We want to be Liverpool, not Derby County. So instead of shouting and screaming at the owner to get his chequebook out, as Brady suggests, we should be demanding that Sunderland be the smartest kid in the room, rather than the mug punter we have perhaps been in the past.

Or put another way that throws back to our more recent successful (relatively) times: Jermain Defoe or Jeremain Lens anybody?