Data Rant: how many goals will United need this season?
Manchester United wrapped up business for the summer on Tuesday with the £36 million acquisition of Anthony Martial. Ed Woodward failed to significantly improve Louis van Gaal’s defence, but added two of the best young forwards in Europe in Memphis Depay and Martial. Meanwhile, midfield has been significantly improved by Bastian Schweinsteiger’s experience and Morgan Schneiderlin’s steel, while David De Gea’s service has been retained – for now. So where will United finish next May?
This column has previously noted the weaknesses in popular in predictive models, such as the expected goals approach (xG), so does not attempt to build a fresh model that spits out general predictions. Instead, Data Rant makes a few broad assumptions and then we discuss the possibilities that arise.
United is assumed to be just as solid defensively in the current season as last year. After all, there has been no personnel loss in defence, aside from the departures of reserves Rafael da Silva and Jonny Evans. Luke Shaw has started the season well, working towards justifying his price tag, while Matteo Darmian has been commanding down the right flank. Louis van Gaal has also persisted with a possession-oriented philosophy this season.
Elsewhere in defence, David De Gea must be affected by the ludicrously botched deal between United and Real Madrid. Still, the Spaniard now has a strong incentive to perform to the best of his considerable abilities this season, with the Spain national team’s number one jersey very much up for grabs. Good performances will also allow De Gea to leverage a better deal with Real next summer. Given these assumptions, it is not unreasonable to conclude that United will concede no more than the 37 goals that De Gea let in last season.
The question of goals scored is more tricky, given United’s inability to secure a new striker over the summer to go with the two youthful attacking acquisitions. However, the relationship between goals scored and league position is not all that strong, with a weak correlation over the past 10 Premier League campaigns. This is not all that surprising since it does not take into account each team’s defensive strength. Another factor is the inherent volatility in goalscoring.
The average number of goals scored by teams in each season fluctuated during the past 10 seasons, while standard deviation, which measures how widespread the teams’ goals scored are in a given season, varied even more wildly.
Goal difference by definition takes into account a team’s ability to defend as well. Indeed, there is a significant correlation between goal difference and league position. This is solid enough ground to make a prediction: specifically the goal difference required for a given position in the league table.
With these figures we can also compute the number of goals United need to score to reach a particular ranking since we assume that the Reds will concede, at most, 37.
Note: for this table, a model from logarithmic regression has been used to account for the ‘kick’ in the higher league positions.
United scored 62 goals and came fourth last season so the model looks passable at the very least. Notice that reaching third or fourth is significantly easier than challenging for the title. The £100 million question, therefore, is the number of goals United will score this season.
As noted, there is no model – grounded in sound statistical theory and methodology – currently used that can accurately predict a player’s goal tally. There are simply too many factors in play. Take, for example, this hypothetical: suppose there are two Lionel Messis who perform exactly the same in any given game – the Messi at Old Trafford will probably score less than the Messi at Nou Camp because Barcelona has better players than United.
Instead, Data Rant observes that the current United squad will score 130 goals this season if Wayne Rooney and company manage to score exactly the same number as their career season high.
This is, of course, totally unreasonable on many grounds. Ashley Young, for example, will probably not play enough games to score eight goals. Elsewhere, players might do better though. Juan Mata’s career best of 12 goals in the Premier League seems feasible. However, Rooney’s form seems too dismal for the Englishman to score anywhere near the 27 goals he managed in 2011/12. But, then again, there is also a chance that Martial doubles his goal tally from last season.
Taking United’s goalscoring potential as a whole, the current squad can ‘win the league’ if the players perform to 78 per cent of their peak goalscoring potential. Since most players in the squad are too seasoned to realistically demand improvement, much of the responsibility probably falls on Depay and Martial. If United does well, the teenage striker’s fee may very well be justified at the end of 2015/16 Premier League season.