In 2015, sports data consultancy 21st Club ranked Liverpool and Tottenham as the 23rd and 24th best teams in the world, lagging behind the likes of Southampton, Shakhtar Donetsk and Borussia Monchengladbach.
Their rise since then has been remarkable. In four years, the Premier League pair have jumped up to second and sixth place respectively in those rankings, and they will go head-to-head in Saturday night’s Champions League final, the most prestigious game in world club football.
Both sides are routinely outspent in the transfer market by moneyed rivals – Spurs haven’t signed a single player for 18 months – so they’ve had to rely on other methods to improve performances. Data science has played a huge role.
“People have been talking about data in football for quite a long time now but most clubs are still nowhere near utilising the power of what you can get from data,” says Omar Chaudhuri, head of football intelligence at 21st Club. “Football is by and large a conservative industry – they either see data as a threat or don’t see the value in it.”
Liverpool and Spurs are different. Both have used stats to an increasing degree to identify and do due diligence on potential transfer targets, and even to feed into player instructions and tactical information. At Melwood, Liverpool’s training base, physiotherapists and psychologists work alongside data analysts schooled in polymer physics and astrophysics.
Liverpool’s use of data has grown more sophisticated in recent years. In 2010, the club were bought by Fenway Sports Group – who had a track record of success in baseball with the Boston Red Sox, building on the approach outlined in the Michael Lewis book Moneyball, of exploiting inefficiencies in the transfer market.
The club embarked on a data-driven spending spree, splashing eyebrow-raising amounts on relatively young English players with good attributes in a particular aspect of the game. They spent big on winger Stewart Downing for his crossing ability, midfielder Charlie Adam for his long passing, and £35 million on striker Andy Carroll for his heading. All three players flopped.
Football is more complex than baseball, and there’s a bit more to scoring a goal than just getting someone good at crossing to lump the ball in towards someone good at heading. “More research shows that crossing is a relatively inefficient strategy – you have to adjust for the quality of those chances created,” says Chaudhuri.
Now, as detailed in a recent feature in the New York Times Magazine, the club’s data scientists use complex mathematical models and more sophisticated metrics. The purchase of midfielder Naby Keita last summer came about because he ranked highly in a model created by Liverpool’s head of research Ian Graham that calculated how each individual action on the pitch changed the chances of scoring a goal. Keita came near the top, as did Mohamed Salah – a surprising signing who had struggled in a previous spell in England, but has become one of the best players in the world since joining Liverpool.
Spurs, meanwhile, have used data to adopt a tech investor’s approach to the transfer market. “As an investor you either spend a lot on sure-fire things, or spend less on start-ups but spend across a bigger portfolio,” says Chaudhuri. It’s worked – although for every Son-heung Min (67 goals in 187 games), there’s a Vincent Janssen (6 goals in 42 games). In the last few years, Tottenham’s transfer activity has been seriously curtailed by the cost of rebuilding their stadium, but that’s given their young squad – the greenest in the league in 2017/18 – the chance to reach their peak together. Analysis by 21st Club shows that while 25 per cent of Mauricio Pochettino’s squad were at their peak age in 2017/18, this season that proportion had risen to 38 per cent, with the same group of players. Tottenham may not have signed anyone, but they’ve brought through players from the academy and unearthed enough gems to make it to the final.
So, who will win?
Number crunching helped both clubs get to Madrid, but the data makes Jurgen Klopp’s Liverpool side the overwhelming favourites. 21st Club’s model, which looks at the historical performances of the teams in all competitions as well as the contributions of individual players in terms of goals and assists, puts the chances of a Liverpool win at 63 per cent. The prediction website FiveThirtyEight (run by political statistician Nate Silver) puts the odds even higher – at 72 per cent. Punters on Smarkets, a betting exchange, put Liverpool’s chances at 65 per cent, with a 1-1 draw the most likely result after 90 minutes.
For Spurs, striker and talisman Harry Kane is an injury doubt – he hasn’t played since injuring his left ankle in the first leg of the quarter final against Manchester City. But 21st Club’s model suggests that even if Kane does play, it will only improve Tottenham’s chances very slightly, from 34 per cent to 37 per cent, thanks to the quality of back-ups Heung-min Son and Lucas Moura.
An injury to Liverpool’s star man Mo Salah, would be much more damaging – cutting their chances of winning from 63 per cent down to 55 per cent. Fortunately for Klopp, Spurs don’t have a player quite as ruthless as Sergio Ramos, the Real Madrid centre back who dislocated Salah’s shoulder after 25 minutes of last year’s final, and then gave goalkeeper Loris Karius a concussion that may have contributed to a crucial mistake.
Football is a particularly volatile game – its low-scoring nature means there is a high-degree of luck involved in who runs out winners. That is perhaps best demonstrated by both Tottenham and Liverpool’s turbulent ride to the final. The Smarkets exchange put Tottenham’s odds of winning as low as 0.8 per cent in mid-November, after they failed to win all three of their opening group games, while Liverpool’s odds fell to 3.5 per cent after their 3-0 defeat to Barcelona in the first-leg of the semi-final.
This season's Champions League might have been decided by data, but its also demonstrated that statistics can’t predict everything – and whoever lifts the trophy, it will be something of a miracle.
This article was originally published by WIRED UK
