At the heart of the model used for BT’s 2019/20 Script of the Season is a machine-learning algorithm analysing and rating thousands of teams and players.
The likelihood of correctly estimating the final Premier League table is comparable to your odds of winning the lottery (45,000,000/1). So getting each week’s scorelines right, even disregarding goalscorers, is like taking home the jackpot on a weekly basis for roughly nine months.
Data helps, but that’s still quite a challenge.
The model was created by third-party data scientists highly reputable within the industry. Their algorithm rates the attack and defence of thousands of teams across Europe based on previous results and in-game scoring sequences. The purpose of these ratings is to calculate the number of goals any given team will most likely score or concede against any given opponent, and vice versa.
The process of producing predicted scorelines roughly follows a Poisson distribution. For example, if Everton are expected to score an average of 1.55 goals vs. Leicester City, they will fail to score 21% of the time, score once 33% of the time, twice 25% and so on. These probability weightings form the basis of the simulation. The more likely a scoreline is, the more likely it will be selected by a random number generator. Repeat this another 379 times with the appropriate teams and you have the final standings for an entire Premier League season. More information on the methodology can be found at the bottom of this article.
Upon receiving the results, Squawka set about mining the simulated findings to establish plot points and their relationship with real events from the 2018/19 season and before. For instance, should they retain the league crown, Manchester City will have equalled the rarest title-winning achievement in Premier League history. Mohamed Salah is highly likely to be competing for his third consecutive Golden Boot, something only Alan Shearer and Thierry Henry have done in the competition’s history. Sergio Aguero could score his 12th Premier League hat-trick, surpassing Shearer to claim the record outright.
The algorithm processes past data. So the table in The Script is an expression of the present state of play for all 20 teams as much as it is a projection of a possible future. It is an innovative 2019/20 season preview, one probably unlike any you have read before, intended to entertain and create debate.
What follows are some of the top-line predictions at the heart of said debate accompanied by a case as to why they could actually occur.
1. MAN CITY TO EQUAL MAN UTD’S UNIQUE PREMIER LEAGUE ACHIEVEMENT
There are some elements to City’s 2018/19 title win more impressive than the record-breaking year before, in which they became the first team in Premier League history to amass 100 points within a single campaign.
Their closest rivals, Liverpool, lost just one game and won the third-highest points total in competition history (97). But less widely known is the fact City trailed for just 132 minutes. According to Opta, this is the lowest single-season total in Premier League history.
They set this record while missing the Player of the Season from the campaign previous for half of their games. And Kevin De Bruyne’s quality remains clear. He created 2.59 Premier League chances from open play in 2018/19, the most of any player to have started at least three games.
As for the manager, Pep Guardiola has won eight of 10 possible league titles in his managerial career. The ‘three-peat’ is unique to Manchester United (2006–07, 2007–08, 2008–09) in the Premier League but not Pep, who has won three consecutive league titles with both Barcelona and Bayern Munich.
2. FORTRESS ANFIELD WILL ENDURE
Sam Allardyce is the most recent manager to win a Premier League game at Anfield. And that victory did not even come with Everton, his most recent role.
It was while managing Crystal Palace he claimed three points away to Liverpool. April 23rd, 2017, saw Allardyce win his first Premier League game at Anfield. This was the 14th attempt of his career, across five different clubs. Palace had 28% possession that day. The player most responsible for inflicting the defeat was Christian Benteke, who scored twice in a 2-1 win.
Currently, Liverpool are on a 40-game unbeaten run at Anfield and have lost just two home games during the past three Premier League seasons. They have conceded 38 goals and faced 122 shots on target at home in this period. On both accounts, those are the lowest totals of any top-flight club to feature in all three campaigns.
The home solidity extends to European competition, too. Liverpool are unbeaten in 19 European matches at Anfield, stretching back to a Real Madrid defeat in October 2014. The only longer run of this kind in the club’s history lasted 40 games between September 1974 and December 1991.
3. A DIVIDE AMONG THE ‘BIG SIX’ CREATES A TIGHT RACE FOR THE FINAL TOP-FOUR SPOT
In one order or another, the so-called ‘Big Six’ have occupied the Premier League’s top six positions for three seasons in a row. During the last 10 Premier League seasons, they have taken 54 of the possible 60 top-six slots. Or, expressed another way, 90% of them.
The reigning champions begin the upcoming campaign deemed most probable title-winners, followed closely by Liverpool. Spurs are, outside of City, English top-flight football’s most consistent team in recent years. They have made the top four in five consecutive seasons now, averaging around 74 points.
And so most of the uncertainty surrounds the remaining three of last season’s top six.
Frank Lampard is a relatively unknown quantity at Chelsea. The Blues sneaked into third-place last season despite the discontent Maurizio Sarri’s style sparked among fans. Arsenal have this week announced the arrival of Nicolas Pepe. He may not solve their defensive problems – they conceded more league goals than Newcastle United and Leicester City last season (51) – but it seems due to the scarcity of affordable centre-backs of sufficient quality on the market, Unai Emery is banking on out-gunning enough opponents in the upcoming campaign.
Which leaves Manchester United, for whom the number six is becoming a figure of dread. They lost six of their last 10 games of last season across all competitions – while winning only twice – to bow out of the Champions League and finish the Premier League season in sixth place. On 66 points. During the last three seasons, United have created the fewest chances among the ‘Big Six’ (1,212), scored the fewest goals and conceded the second-most (111), behind only Arsenal.
Sixth is the only position in which Man Utd have ended a season more than once since Sir Alex Ferguson stepped down. They average 69.3 points per season over those six campaigns – their best finish being second, on 81 points, their lowest 7th, on 64 – and roughly nine league losses per 38 games.
There remains one category, however, in which they still outstrip their rivals: United finished highest of all Premier League clubs in Deloitte’s Football Money League for 2019. Their listed revenue? €666m.
But United are the nation’s richest football club for a reason. And despite all of the above, they have of course jumped from sixth to as high as runners-up in a subsequent season before. Realistic success this time out need not call for such a high placing. A return to the Champions League appears a reasonable objective. Keeping Paul Pogba while signing Harry Maguire and Paulo Dybala would increase their chances of doing this. Cause for optimism currently comes in the fact Solskjaer will have had a whole pre-season to work with his team.
4. EVERTON THE BIGGEST THREAT ON THE TOP SIX
Everton are traditionally the biggest threat on the Premier League’s ‘Hexopoly’. The Toffees have been ‘best of the rest’ (i.e. seventh) or broken into the top six in nine of the past 15 top-flight seasons. It is a record reflected, and presumably influenced, by their status as the Premier League’s seventh-richest club (17th in Europe, just ahead of AC Milan).
They maintained the highest average share of possession and were the second-highest scorers outside the ‘Big Six’ last season; only the top four conceded fewer goals than they did (46). In fact, only the top three faced more shots last season, a clear sign of defensive organisation.
Considering their end-of-season form in 2018/19, they have much to build upon. They lost only two of their last 11 competitive games, the fewest of any team outside the top two, while winning six; again, only the top two won more. During this period, they conceded only seven times. That low tally matches Liverpool and is higher only than City’s (3). To prove their credentials in terms of matching the ‘Big Six’ head-to-head, only Man City managed to beat them in the second half of the campaign. In fact, only Spurs managed to even score against them.
Everton 0-0 Liverpool
Everton 2-0 Chelsea
Everton 1-0 Arsenal
Everton 4-0 Man Utd
Spurs 2-2 Everton
Wolves and Leicester City represent their closest rivals. And it is tight. Three points separated seventh from ninth in The Script. Everton’s summer spending is set to surpass the £100m mark, according to the Times. But Wolves took seventh place last season while Leicester City, themselves highly active in the transfer market, look forward to a first full season with Brendan Rodgers as manager.
5. NOT ALL THE THREE PREMIER LEAGUE NEW BOYS WILL SURVIVE
History suggests the top flight will have waved goodbye to one or more of Aston Villa, Norwich City or Sheffield United by next May. In all but three of the 27 seasons of the Premier League era, at least one newly promoted club has gone straight back down. This is reflected in the fact Sheffield United (4/6) are the bookmakers’ favourites to be demoted this season, followed by Norwich City (10/11).
Norwich have spent comparatively little in the summer’s transfer window and, with Daniel Farke at the helm, are likely to eschew the conservative playing style many associate with top-flight survival. Farke’s more imaginative playing style has defied critics before. But he must do so once again if Norwich City are to avoid setting an unwanted Premier League record: avoid the drop, or become the most frequently relegated team in the competition’s history (5).
Newcastle United are considered the division’s fifth-favourite relegation candidate, ahead of Burnley and Brighton, but they are the most popular bet (22%), according to Oddschecker. Steve Bruce finds himself immediately behind the eight ball as he hopes to defy the ‘wisdom of the crowd’. “This is without a doubt the toughest and most toxic situation Steve has ever walked into,” said Alan Shearer, the club’s all-time top goalscorer (206) and also someone who personally advised Bruce to turn the job down.
The man who will wear Shearer’s old No.9 shirt last season, summer signing Joelinton, is under pressure to replace some 56% of his new club’s league goals from last season, which Newcastle let go in the form of Ayoze Perez and Salomon Rondon, whose loan deal was not made permanent. Their next-highest scorer in the league was centre-back Fabian Schar (4).
Key to Newcastle’s survival hopes will be how well Bruce maintains the defensive organisation instilled by Rafa Benitez. His team conceded 48 goals last season, fewer than Manchester United and Arsenal. Scratching deeper beneath the surface, they allowed Premier League opponents only 147 shots on target, the seventh-lowest total in the division. A better record than Spurs and Man United managed, too.
The 2019/20 Script of the Season is, of course, a simulation based on a predictive model. It is no Grays Sports Almanac and bets are not encouraged.
But the more probable a result and scoreline is, the more likely the random selection will fall on that scoreline. Thus, it is most probable over 38 games, City will have scored the most goals and accrued the most points.
From said scorelines, goals totals are established for all 20 teams. And from those goals totals, player contributions are determined using expected minutes, XG (expected goals) and XA (expected assists) weighted over the previous two seasons.
Champions League and Europa League routes are based on an algorithm conforming to Uefa tournament parameters. A draw was simulated and team ratings were used to generate results and scorelines in accordance with the aforementioned process. City are currently the model’s highest-rated team in Europe. However, the higher level of competition and added chaos entailed in knock-out tournaments means they will only win around 20% of Champions League simulations. In the Premier League, their chances of winning the trophy jumps up to 60%.
They say the league table never lies. But this is not totally true. Uncontrollable factors create the high level of variance that makes football special. Because football is a low-scoring game, a single goal has more bearing on the match results than ways of scoring in many other sports, such as basketball or rugby. And a single result, or point, can define an entire season. Such fine margins mean the better team does not always win. Around every corner, there is an upset waiting to rip up the script.