Data Rant: box-to-box midfielders

July 12, 2014 Tags: Data 15 comments
featured image

The soon-to-depart midfielder Anderson was converted from a typical advanced playmaker to an ‘eight’ by Sir Alex Ferguson – the rationale being that the Brazilian has all the attributes of a great box-to-box midfielder. Anderson has not lived up to a promising first season at Old Trafford after joining in 2007, leaving Manchester United short of attacking thrust from central midfield for the past seven years.

Ferguson could afford to get by on a functional midfield partly due to Michael Carrick and Paul Scholes’ excellent passing, but mostly due to a reliance on high-flying wingers. Cristiano Ronaldo’s sheer explosiveness helped mask a pedestrian midfield, while Antonio Valencia, for a while at least, consistently provided dangerous crosses from the right.

That Valencia has suffered two poor seasons in a row, while Ashley Young and Nani face an uncertain future at Old Trafford, says something about the degradation in that department. Luke Shaw and Rafael are exciting young fullbacks, but they are limited by their position on the pitch. Yet, having thoroughly figured out 5-3-2 at the World Cup, Louis Van Gaal might even do away with wingers entirely if he is to get the best out of his United squad.

Central midfielders, therefore, must break the lines to provide attacking impetus from midfield; Ander Herrera has been recruited for that task. The question of whether United actually needs such a player remains though. After all, direct football, such as that employed by Netherlands in Brazil this summer, tends to bypass the middle entirely.

In this Data Rant we look at the best dribbler from each Premier League side’s central midfield then see if their skills have any correlation with points the player’s team earned last season. In figure 1, below, there seems to be little relationship between the two. That is, having a good dribbler in the engine room is not strongly related to winning games.

Figure-1: Dribbles vs Points

Notice, however, that six points are clearly distinguishable from the rest in upper-middle areas – those belong to the top six sides. With these clubs removed, in figure 2, the correlation is stronger. In addition, United is now an outlier – the Reds are the best of the bad bunch and the worst of the good.

Figure-2: Dribbles vs Points

The fact that United’s best midfield dribbler is Tom Cleverley, and that the Reds had the worst such player in the Premier League is, in a sense, interesting. From a purely box-to-box point of view, the Reds should have ranked lower in the table. Put simply, there are other factors that make up for the lack of an all action midfielder.

In figure 3, below, the top four lie above the blue line. It is that line van Gaal’s United must aim for in the coming season. Notice the horizontal spread between these clubs. There are significant gaps between City, Chelsea, Liverpool and Arsenal’s best midfield runners, despite little differences in points. This is another piece of evidence that on-the-ball running is not an important a factor in winning football matches.

Figure-3:  Dribbles vs Points

This conclusion runs against intuition, of course. When a skilful midfielder is pinned down, he can work his way out of trouble, enabling his team to create an overload or attacking opportunity. Yaya Touré, for example, would not have scored twenty goals had he not been an exceptional dribbler. Meanwhile, stronger sides have to regularly break down defiant defenses. While holding on to that ball could also earn the team that last-minute set piece, it is relevant to note that any player with the ball can accomplish this feat.

Finally, in figure 4, there is a significant relationship between points and number of take-ons attempted by each Premier League club. United is about 200 take-ons short of the top four.

Figure-4:  Take Ons vs Points

Juan Mata chalked up 52 in his last full season at Chelsea, while Herrera supplied 73 for Athletic in 2013/14. Provided the pair performs to that level in the coming season United should dribble its way into a stronger position in the Premier League table.

This bodes well for Ander’s impact at Old Trafford. United was deficient in many things last season, but at the very least, David Moyes was short an Ander Herrera.

All data from Squawka
A brief note on methodology:
1) All categories are weighted equally
2) Each figure has been adjusted relative to the ‘best’ in each category
3) Assumptions dictating linear regression have not been held strict

Data Rant: van Gaal’s Netherlands model

June 28, 2014 Tags: Data 5 comments
featured image

Spain was humiliated in Brazil, to an ‘anti-tiki-taka’ Netherlands side to boot. It begs the question of just what kind of system new Manchester United manager Louis van Gaal will use at Old Trafford? Say it quietly – David Moyes has not been gone long enough for United fans to consider direct football without flinching – but the direct approach has been in vogue for some time. van Gaal might persist with the Dutch model at Old Trafford.

There is a clear relationship between possession gained and points won in the Premier League, below. The correlation is solid, but not so much so that this can be taken as fact. In addition, this analysis lacks a crucial element in distance covered by the ball.


A short passing game can increase a side’s possession statistic, but so can a long ball game if the direct passes connect accurately. In addition, chances can be made by more agricultural methods as well as the advanced; a key pass made through a clever one-two is worth the same as one created by a punt from inside a team’s own half.

However, the general inefficiency of long ball football has long been known – United under Moyes epitomised this point. Theoretically, though, a well-drilled side full of capable passers can create even more chances from deeper positions by utilising the extra time and space granted.

Data Rant computes ‘Chances In Meters’ (CIM) by multiplying pass accuracy, average pass length and number of chances created, to measure the amount of ground covered by key passes, while adjusting for prolificacy of attempts. CIM has a very strong relationship with points won and supports the theoretical argument that more accurate long balls can be just as effective as a series of neat passing triangles near the box.



For example, a team capable of one well-executed Hollywood ball every game can win just as many points as a side making two incisive through balls each match. Direct football can work in the right hands.

Consider, once again, the Spanish and Dutch World Cup sides. The Netherlands is well behind in pass accuracy to Spain, but has attempted on average longer balls. Despite this, the Dutch national team has produced one more chance than Spain and has a higher CIM count versus Spain. It is worth noting that Spain was beaten 5-1 by the Netherlands.


The direct approach espoused by Moyes left United seventh in the Premier League last season. Still, United currently lacks the midfield – Ander Herrera notwithstanding – to attempt a more cultural approach. The pragmatic van Gaal might continue with the Dutch philosophy, albeit much more adroitly.

Should van Gaal indeed decide to mimic the current Dutch side, Adnan Januzaj and Shinji Kagawa will probably perform the Arjen Robben role of providing direct running in the final third. Juan Mata, on the other hand, will likely be supporting the attack rather than having the play funneled through him. A central midfielder must then connect defence with attack.

Tom Cleverley can provide energy, while Maroune Fellaini may bloom into a top central midfielder under new management. United needs a surefire option though and a midfielder who can perform in the box-to-box role has been purchased in Herrera. Curiously though, Cesc Fabregas was passed over by van Gaal and the more callow player recruited.

Bryan Robson is perhaps the most memorable box-to-box midfielder at Old Trafford. There are qualities that are inherent in such player; an ability to carry the ball and beat opponents and certain feistiness, especially in the air. In addition, the box-to-box midfielder must protect the defence while putting up the numbers offensively.

Taking Robson’s two goals in every 10 games as a benchmark we analyse a number of Premier League central midfielders and aggregate key metrics, below.

Data Rant uses interceptions, tackles and blocks as a proxy for defence; take-ons, headers, and header success for box-to-box midfielders; and shots, goals and assists for attack.

Herrera boasts the second highest score in the box-to-box category, while contributing almost equally in defence and attack. Fabregas, on the other hand, contributes relatively little defensively and concentrates on making an impact in advanced areas.

Jonathan de Guzman is in charge of transition in van Gaal’s Dutch side and his aggregated statistics bear a great resemblance to Herrera’s. Both are remarkably well-rounded midfielders and a direct option, at least as a plan B, is clearly in the new United manager’s thinking if the story about him turning down Fabregas is true.


The new contract for Antonio Valencia together with Herrera’s arrival provide strong evidence that United may very well start the 2014/15 season with an approach that ruined last season. Moyes brought in Fellaini after failing to secure Herrera – the £27.5 million spent didn’t amount to much. Will United get some bang for the buck with £28.5 million spent on Herrera?


All data from Squawka

A brief note on methodology:
1) All categories are weighted equally
2) Each figure has been adjusted relative to the ‘best’ in each category
3) Assumptions dictating linear regression have not been held strict

Data Rant: Antonio Valencia

June 24, 2014 Tags: , Data 7 comments
featured image

To much surprise Antonio Valencia signed a new, three-year, contract with Manchester United this week. Once bearer of the famed number seven shirt, Valencia’s star has fallen as much in reputation as output in recent years. The appointment of high-flying Louis Van Gaal is seen a sign of United returning to its attacking history, but retaining the Ecuadorian winger has taken much wind out of the United faithful’s sails.

Valencia’s form is such that many fans feel that the Ecuadorian is entirely undeserving of a new contract. There had been reports of Liverpool’s interest – and so poor has the player’s form been that very few United fans would object to a move – even to a major rival. Perhaps the most rational explanation is that United’s management has sought to preserve Valencia’s value in the market.

Contrary to many reports, the new United manager is not wedded to a 4-3-3 formation. In fact Van Gaal has often deployed a 4-2-3-1, while the Netherlands national team under his management is playing a 5-3-2 at the World Cup in Brazil – a point we’ll return to later.

Either way, a backup winger is required with United’s cupboard especially bare in wide areas. The suspicion, however, is that Valencia provide will cover Rafael da Silva at full-back instead, with the Reds lacking cover for the Brazilian, and Valencia having been deployed in that role already. Phil Jones and Chris Smalling will be required in more central areas, after all.

table-1 defensive actions

Valencia is now primarily known for his defensive dependability than his attacking numbers; the new Ji-Sung Park, if you will. The Korean was, with some irony, a protégé of another Dutch coach in Guus Hiddink.

Yet, the data shows that the 28-year-old Valencia is rather different from other defensive wingers. The blocks that Valencia made last season suggest that he dropped deeper than Chelsea’s Willian, for example, while the Ecuadorian was far more studious in tackling than United’s Ashley Young or Danny Welbeck.

table-2 - defensive actions

The Ecuador captain has exhibited statistics more typical of an attacking full-back, although notice how many of Valencia’s defensive actions were interceptions last season. Valencia made less defensive actions than Rafael – a factor of Valencia’s higher position the pitch relative to the Brazilian – and he clears the ball less often than the full-back.

This observation is obvious, although the number of chances created by Rafael and Valencia is also interesting.

The images below – Rafael left, Valencia right – show that the Ecuadorian winger created chances in much deeper areas than Rafael. This suggests that despite Valencia’s more forward deployment, it was Rafael who attacked the byline. That is, Valencia was stationary and functional, Rafael more dynamic.

Rafael Chances Valencia Chances

To pin down Valencia’s default position we must establish that the proportion of interceptions in each player’s defensive actions represents the player’s positioning. To do this we look at a randomly chosen defender and midfielder from each of the Premier League’s top 10 clubs, and then compare the number of interceptions to the number of shots per game. Each figure has been adjusted by the club’s point tally, ensuring that the experiment is unaffected by team quality.

The idea is that more advanced players will make more attempts at goal than deeper players. If there is correlation then we can deduce where on the pitch, and with what role, Valencia was deployed last season.

Data towards the right of the chart, below, represents among others, Frank Lampard, who aggressively ventures forward, and some central defenders. They are outliers for our purpose and we remove them and see if there is any underlying trend.

figure-1 - shots per game-interceptions

There is a distinguishable relationship between shots-per-game and number of interceptions. In this Valencia’s positioning is shown to be very close to that of Michael Carrick and Newcastle’s Vurnon Anita – both defensive midfielders. In other words, Valencia played in line with Carrick. Valencia played as wing-back rather than a true winger.

figure-2 - shots per game-interceptions

Very few top clubs play five across the back. Juventus, however, plays such system with Kwadwo Asamoah as left wing back. The similarities between the Ghanaian and Valencia are remarkable. Valencia left, Asamoah right in the images below.

table-3 - Valencia vs Asamoah

Asamoah Chances Valencia Chances

The data is illuminating. In essence David Moyes played a back five last season, which helps explain the excruciating rigidity of United’s approach – and Rafael’s exile from the first team. This also sheds light on Valencia’s new contract – van Gaal could very well continue with the 5-3-2 system deployed at the World Cup. After all, United’s ace in Robin van Persie appears to be very happy with the approach.

All data: Squawka

A brief note on methodology
1) All categories are weighted equally
2) Each figure has been adjusted relative to the best in each category
3) Assumptions dictating linear regression have not been held strict

Data Rant: is attack the best form of defence?

June 17, 2014 Tags: , Data 9 comments
featured image

David Moyes’ defensive outlook at Manchester United failed to bear fruit despite the adage that ‘defence wins championships.’ Even with the Scot’s obvious emphasis on conservatism, Moyes’ Reds failed to better his old club, Everton, in terms of goals conceded last season. United’s back-four is set to be reconstructed this summer after Nemanja Vidić and Rio Ferdinand left the club. Data Rant examines United’s rear guard… and looks to a future under Louis van Gaal.

First, it comes as no surprise that fewer goals conceded leads to more points gained in the Premier League.


Curiously though, it is hard to distinguish one team from another in terms of defensive statistics, below. Apart from Chelsea, where José Mourinho’s side made far fewer tackles, it is hard to point to a trend purely from the data.


If defence does indeed wins championships, then attack is said to win matches. But could it be the other way around? A team needs to scor to win, of course, and one is automatically defending when in possession and occupying territory in the opposition half – the old adage that ‘attack is the best form of defence.’

Key passes is one proxy for attack – and are by definition threatening to the opposition. There is a fairly strong trend between assists and goals conceded, below. That is, more assists – as a proxy for attacking – means a more secure backline. Fulham is a clear outlier in this analysis.

Figure-2 Goals conceded vs assists

Along with Liverpool and Manchester City, Fulham is one of three clubs that scored more than 10 goals from corners. The London side’s goal figure is significantly distorted by this focus on dead ball situations and Fulham’s removal from the dataset makes clear the correlation, below.

Figure-3 Goals conceded vs assists ex Fulham

While intuitive, it is far too precarious to assume that attack is the best form of defence just yet. Too much commitment in opposition half can backfire – accuracy therefore must also be accounted for.

Shooting is also a good measure of time spent in advanced areas. We use not shooting accuracy, but the number of shots on target to make sure that the time each team spends in the opposition half is accounted.

The number of shots on target is the a good barometer of a team’s collective attacking ability – a good playmaker will consistently line up better passes than a mediocre one, while a great striker will simply finish better. If attacking is positively related to defending then accuracy will have a strong relationship with goals conceded by each team, below.

A solid, yet inconclusive, trend appears when the number of shots on target by each Premier League club is compared to goals conceded.  Fulham again stands out, but so do four other clubs. The cluster to the bottom left of the chart below – Stoke City, Crystal Palace, Hull City and West Ham United – points to four teams sacrificing attack for a solid defence.

Each managed to stay in the Premier League. In fact, apart from West Ham, each improved from the season before. Liverpool, on the other hand, could have let in far fewer goals judging by performances of other clubs in this data. Brendon Rodgers’ attacking approach, however, has put the Merseyside club back in the Champions League.

Figure-4 Goals Conceded vs Shots on Target

Figure-5 Shots on Target vs Goals Conceded

This brief study has demonstrated a potential causal link between attack and defence. Something to note when considering the significant headache United faces with Ferdinand and Vidić no longer in the squad.

Palace may have finished 11th with a solid defense, but Liverpool’s approach took the club into Europe. It is, therefore, clear which approach United ought to replicate. van Gaal’s Netherlands defeated Spain 5-1 with intensity and not solidity, for example.


All data from Squawka

A brief note on methodology:
1) All categories are weighted equally
2) Each figure has been adjusted relative to the ‘best’ in each category
3) Assumptions dictating linear regression have not been held strict

Data Rant: Rooney in midfield

June 3, 2014 Tags: , Data 18 comments
featured image

Paul Scholes’ recent claim that Wayne Rooney has “all the ability to take over my old position at Manchester United” has reopened the door for a long-thought-buried conversation. Those supporting Scholes do so reasoning that Rooney can hit sweeping passes, just like the retired United midfielder, and that the 28-year-old’s well-roundedness will shine in deeper areas. The opposition has claimed that Rooney simply does not have the technique required to star in a modern midfield.

Data Rant compares Rooney to top Premier League strikers and midfielders to see where the England international might feel more at home. As usual, normal assumptions dictating linear regression have not been held strict. Players have been judged on following:

  1. Consistency in goal scoring
  2. Ability to get into the box
  3. Dribbling past markers
  4. Aerial presence
  5. Passing
  6. Incisiveness

The number of assists made by most creative teammate has been included to reflect the strength of a player’s side. The figure has been adjusted so that players playing for relatively weaker sides can be compared fairly to players at leading clubs, such as Manchester City’s Sergio Aguero.

The top 10 forwards in the league have been considered. That is: Aguero, Luis Suarez, Daniel Sturridge, Olivier Giroud, Edin Dzeko, Wilfried Bony, Jay Rodriguez, Romelu Lukaku, Loic Remy, Rickie Lambert and, of course, Rooney.

In Figure 1, below, it is awkward to summarise from the trend line, but a pattern can be quickly recognised – data points are diverging with Suarez the outlier. The vertical axis marks the number of goals scored, with more rounded players towards the right of the graph (the horizontal axis measuring a sum of attacking attributes: shots, take ons, successful headers, pass accuracy, chances created, assists.

The Uruguayan striker scored 31 goals and assisted 12 times last season and his position in the graph reflects his excellent all-round campaign.

Figure-1 Top Premier League Strikers

In Figure 2, below, Rooney is added to the mix. The new trend line is not a perfect fit, but Rooney is very close to it. The model suggests that Rooney should have scored 19 Premier League goals – the United forward scored 17. Notice that Rooney, along with a few others, lies away from the cluster of players to his left. We’ll get this this momentarily.

Figure-2 Top Premier League Strikers

In Figure 3, four players stand out from the model; Sturridge, Aguero, Remy and Rooney. If it wasn’t for these four, the trend line would perfectly connect the group.

Figure-3 The Four

Given his all round performance, Sturridge significantly outperformed the model’s expectation in his scoring feats. Rooney, Remy and Aguero all scored a decent amount of goals in the 2013/14 season, but their creative abilities push them towards the right of the chart. The data therefore confirms Rooney’s completeness as a player.

We now compare the 11 forwards to attacking midfielders: Yaya Toure, Steven Gerrard, Aaron Ramsey, Raheem Sterling and Oscar. They are top five central players by goals scored and have been judged on the same criteria as the strikers.

Figure-4 Forwards + Attacking Midfielders

As in the first graph the model doesn’t produce an easy trend line. There are two ways of interpreting data here.

Two almost parallel lines have been drawn, below, to reference two groups in our model. The blue line passes through our group of midfield players and Aguero, who is comfortable playing deeper. The black line hits many of forwards and Toure. Toure scored 20 Premier League goals and it would shock few if the Ivorian is categorized a ‘striker’ in this model. The ‘parallel lines’ model implies that attacking midfielders are simply strikers with more tools to utilise. However, Rooney lies awkwardly in the middle – close to both lines but not quite enough to belonging to either group.

Figure-5 Forwards + Attacking Midfielders

Suarez, below, is plotted far away from the rest, but the Liverpool striker had such an excellent season that it would have been more surprising had there been a player close to him.

The black line connects strikers and Toure closely, while the blue line does a very good job of attaching attacking midfielders. The ‘scissors model’ suggests that attacking midfielders can just as well labeled as a forward – not that far-fetched considering that Scholes himself had been a deep lying forward before moving deeper as he matured. It is hard to argue that Rooney will make such transition, though – even in this model Rooney sits in the middle of two clusters.

Figure-6 Forwards + Attacking Midfielders

It is perhaps too early to predict Rooney’s positional future as Louis van Gaal might dictate that Rooney’s future lies elsewhere in the country or even the continent. People for and against deploying Rooney as central midfielder all make very good points. Unfortunately, data supports both and neither at the same time.

Addendum – Robin van Persie

David de Gea is the only player who has ended United’s 2013/14 campaign with his reputation intact. The performance of Robin van Persie was particularly disappointing, whose output in front of goal went from 26 to 12 in the league. The Dutch striker suffered significant injury problems under Moyes, but his relationship with van Gaal should cure most of these ailments. Will the new United manager get van Persie firing on the pitch as well?

Robin van Persie

Compared to 12/13, van Persie created less chances and dribbled more last season, suggesting that the Dutchman was rather isolated. The fact that the former Arsenal captain did less defensively in his second season with United solidifies this argument.

The graphics, below, detail the areas in which van Persie made key passes over the past two season. The 12/13 map can easily be confused to that of a playmaker, while the more recent view shows someone who has been crowded out of attacking midfield zones. Indeed, during the season, van Persie complained of a player who “sometimes [occupied] the spaces I want to play in.”

Robin van Persie

Robin van Persie
(Source: Squawka)

United’s lack of form could have adversely influenced van Persie, but it is Rooney who likely occupied van Persie’s spaces. Rooney attempted almost twice as many take ons as the season before. By having the ball longer, the English striker would have forced van Persie to stay further forward.

We revisit the recent Data Rant model above. Rooney of 12/13 and 13/14 and van Persie or 12/13 and 13/14 are considered in addition to top ten Premier League strikers and top five central midfielders.

Wayne Rooney, Robin van Persie

van Persie has gone from being a complete forward to a solid striker. Rooney, in fact, has regressed slightly in his all-round performance, although his scoring record has improved – Rooney’s best role remains unclear. In Sir Alex Ferguson’s last season, Rooney clearly played as a conventional attacking midfielder and van Persie had an excellent season. Maybe having Rooney in the engine room could work.

Data Rant: Rooney’s sale makes sense

May 31, 2014 Tags: , , Data 34 comments
featured image

Paul Scholes caused much ruckus this week by claiming that “Wayne [Rooney’s peak] could have been … when he was 26.” Sir Alex Ferguson’s vehement efforts to dispatch Rooney also hints at something more than simply a personality clash. After all, on the pitch, the 28-year-old has been mediocre for three seasons now. The new four-year contract gifted to Rooney may prove to be the worst legacy of David Moyes’ era at Manchester United.

Louis van Gaal’s opinion on Rooney will certainly be interesting. The English forward’s versatility might appeal to van Gaal, but the Dutchman has surely spent too much time at Ajax and Barcelona to look past Rooney’s, sometimes horrifying, first touch. The incoming United manager could very well have instructed Ed Woodward to find a new home for the Scouser already.

For United, transferring Rooney to a continental club would be ideal, but not only does his huge wage offer a stumbling block, many top European clubs do not need the forward or have a better version already. Thomas Muller and Angel di Maria are just two examples.

England beckons, then, and Chelsea is the only realistic destination for the forward. Only Eden Hazard succeeded in scoring more than 10 goals in 2013/14 for Chelsea so there is a clear need for a number nine at Stamford Bridge. José Mourinho has long been an admirer of Rooney and the English striker would offer dependable firepower to Chelsea’s frontline. It is a marriage that suits all parties.

United’s most iconic player leaving for a rival could be a public relations nightmare, of course, and the Mourinho system is tailor-made for Rooney, which may allow the Englishman to flourish. Yet, if Rooney’s physical decline continues, the Reds will have the last laugh.

The London side was just as defensive as United last season, but managed to score seven more goals and concede 16 less across the campaign – this was achieved with the same 53 per cent average possession. The mobility and technical approach offered by Hazard and Willian, however, allowed the Blues to dribble past opponents where United did not. A classy striker will make the system tick.

table-1 - attacking attributes/goals scored

The data shows that Willian and Samuel Eto’o bear resemblance to Rooney and the Englishman would be a great replacement for the departing Cameroonian. Rooney’s scoring record last season was better last season than any Chelsea strikers.

Tactically Rooney should fit with Mourinho’s system. Juan Mata was ostracized at Stanford Bridge for doing little of the dirty work and Hazard was publicly chastised by Mourinho for the same reason. Considering that Rooney is diligent to the point of indiscipline there is every reason for the Portuguese to chase the wayward English striker.

Figure-1 - attacking attributes / goals scored


Deducing from the statistics of Chelsea’s forwards, Rooney could be expected to score 10 league goals if he moved to London. Apart from penalties, Rooney has largely monopolised set pieces at United – a luxury he will not enjoy at Chelsea – and his figures might drop even further as a result. Oscar and Hazard run the midfield so Rooney would be on the pitch to finish alone.

There are plenty of other striking options for Chelsea, with the London club closing in on Diego Costa, who is better finisher and might even cost less. Indeed, Rooney was probably earmarked for a defensive forward role last summer, such as that taken by Ji-Sung Park, which would have diminished Rooney’s output further.

Of course, statistics are only a guide, but the analysis suggests that Rooney might fall further from his peak in the coming years. If United took a Machiavellian view, the damage inflicted on Chelsea by Rooney declining rapidly at Stamford Bridge, far outweighs any chance of the English forward recovering the form of yesteryear.

The really frightening thing for United fans is that this analysis assumes Rooney will stay in rude health next season. He rarely has in the past.

Data Rant: creative performance vs. league position

May 28, 2014 Tags: Data 3 comments
featured image

Even the very best teams are occasionally beaten by a less able side – shock results are part of football’s fabric – but consecutive draws and losses become alarming for any club. Under David Moyes Manchester United endured four such ‘poor streaks’, averaging 2.75 losses and/or draws across the quartet of disappointing runs.

Yet, the data shows that it is far too simplistic to blame ephemeral ‘form’ alone – the cause is more nuanced and rooted in the relationship between individual and team performance. Figure 1, below, compares each Premier League club’s total number of poor streaks to league position. The correlation is strong yet predictable – the team that loses and draws the least will, of course, win the title.

Figure-1 - Poor streaks vs. league position

The streaks themselves tell an interesting tale. Stoke City, for example, suffered a dreadful spell, drawing four and losing four, during the campaign. Aside from the 3-1 loss to Arsenal though, Stoke came very close to securing a win or draw in each game. The details count.

United also had a similar sequence of two 1-0 losses and two 2-2 draws during the season. Robin van Persie was injured for the first three games and returned against Newcastle United, although the Dutch striker was clearly unfit. Perhaps the lack of finishing edge helped turn draws into losses and wins into draws?

There is little trend to speak of and the fact that van Persie and Wayne Rooney are among the most consistent finishers in England suggests that the answer lies elsewhere. In fact Data Rant previously noted that a reduction in the quality of chances created was more at fault for United’s performance last season.

Figure-2 Goals-per-game ratio of leading scorer vs. league position

This scatter plot, above, charts the relationship between key passes-per-game of each club’s leading creator and final position in the table. There may seem only a tenuous link, although the clear outlier in the upper right corner and bunched data points below the trend line are clearly distinguishable.

Figure-3 Key passes per game vs. league position

However, with outliers removed, the correlation between the number of key passes per game of a team’s leading creator and league position is striking.

Figure-4 Adjusted key passes per game vs. league position


Rooney, Kevin Mirallas, Adam Lallana, Marko Arnautovic, Jason Puncheon and Jonathan de Guzman are all relative under-achievers, while Robert Snodgrass performed significantly better than the rest of Norwich City’s players last season.

Key passes per game ranking

Rooney’s peer group have been heralded to a varying degrees, but the data suggests that they should have done a far better given their clubs’ standing.

Meanwhile, Juan Mata performed to the expectation of the seventh best side in England. Yet, the gap between Rooney and Mata is staggering. Yet, at the same time, Louis van Gaal’s more fluid style might bring out the best in the Spaniard, enabling the former Chelsea Player of Year  to drag United back into the top four next season.

Figure-5 Key passes per game: Rooney vs. Mata

Rooney’s decline has been noticeable and it will likely only accelerate as the England forward ages. The 28-year-old has produced the numbers, especially assists, but the dearth of key passes hints at a measure of luck rather than intention. After all, if a picture is worth a thousand words, Rooney’s output is more at home at Craven Cottage than Old Trafford. In fact the data supports the notion that many hold: it is becoming increasingly hard to justify deploying Rooney in the hole, especially when the chief creator has such an impact on league position.


Research Methodology
1) All categories are weighted equally
2) Only players who made more than twenty domestic appearances have been considered, unless otherwise noted
3) Assumptions dictating linear regression have not been held strict
4) All data from squawka

Data Rant: how much will the Glazers release this summer?

May 24, 2014 Tags: , Data 16 comments
featured image

Helped by few suspect refereeing decisions Guus Hiddink took South Korea to the semi-finals of the 2002 World Cup. In the process the Dutch manager confused many Korean journalists by referring to his 3-5-2 system as “1-3-3-2-2” – he often preached that a player’s position is always relative to others and that the goalkeeper is crucial in building up play. In a similar fashion incoming Manchester United manager Louis van Gaal also prefers to use the term “1-2-3-2-3” instead of 4-3-3.

Even though former Chelsea manager Hiddink rated the more innately gifted Chun-Soo Lee higher than Ji-Sung Park – it was Park that Hiddink brought to Europe after Lee turned down the Dutchman’s offer. The former Red winger was preferred throughout the 2002 World Cup campaign because versatility is more important than virtuosity in any system that starts with “1.”

At United van Gaal will find one too many specialists for these Dutch sensibilities, but the Dutch manager could appreciate the versatility of Ashley Young and Danny Welbeck. Even Tom Cleverley may have a shot at redemption with the new boss. And while many players at Old Trafford fall short of world class, van Gaal will find an array of ingredients at his disposal for a system that emphasises the collective. After all, United’s disastrous 2013/14 campaign was carried out with more or less same players that won the 2012/13 Premier League.

van Gaal will strengthen this summer, but United lagged only Manchester City in net transfer spending last season. Considering that the Reds won the Premier League in the previous campaign with an 11 point gap, United could very well boast the best “collective” in England.

Table 1

In a previous column Data Rant looked at the relationship between playing career and managerial success. Each point in Figure 1, below, represents a manager at one of the top 25 European clubs as measured by UEFA coefficient.

1) All categories are weighted equally
2) Each figure has been adjusted relative to the “best” in each category
3) Assumptions dictating linear regression have not been held strict
4) In case of a caretaker manager/vacancy, the previous permanent coach’s figures have been used

Close examination of the data shows two distinct clusters of data: points to the far right behave as if they are separated from those to their left. These are managers, such as Sami Hyypia and Diego Simeone, who enjoyed a highly distinguished playing career, including numerous international appearances. We’ll limit our analysis to managers who had a more pedestrian playing career.

Figure-1 - managerial experience

Using this analysis – judging by his ‘peer group’ alone – van Gaal will boast a win ratio of around 66 per cent, which puts United in a title fight next season. This is not guaranteed, of course, but does suggest that the Dutchman’s experience will count for much more than David Moyes’. Using this model the Scot is predicted to produce a meager 55 per cent win ratio – he actually won 53 per cent in his time with the Reds.

Managerial history does not exist in a vacuum though; squad depth and strengthening counts. Although every new manager likes to bring in few players, the mooted £150 million summer budget seems out of scale given the Glazers’ history. It may not even be needed to compete next season. Manchester City’s hands are tied by the Financial Fair Play, Chelsea is walking a fine line with UEFA, while neither Arsenal nor Liverpool are likely to be the largest players in the market this summer.

van Gaal has to shore up his defence, but it is a task that is unlikely to cost £150 million. After all, Liverpool has spent two seasons out of the Champions League with little repercussions in terms of sponsorship revenue, so the Glazers could conceivably ask van Gaal to complete a season with a seemingly decent squad at hand before breaking the bank if disaster strikes again.

Yet, it seems unlikely that United is briefing the press merely to hype the Dutchman’s impending arrival. Indeed, the club has spent heavily in the market of late, which begs the question: what motivates the Glazers to spend now, rather than take the money for themselves?

Figure-2 - United's net spend

It makes logical sense if clubs all stopped spending, of course, but each wants to improve its position. In addition, no club knows whether its rivals will invest, or how much. This has created an ‘arms race’ for players, much like the one during the Cold War where the US and Russia stockpiled nuclear weapons.

But there is also a profit trade-off in heavily investing, with little bottom line difference between finishing fourth and winning the title. Yet, it is impossible for the Glazers to know how much they need to spend because there are other clubs aiming to reach the top four as well. After all, United could get trumped by Arsenal pulling off a big last minute signing. Rational owners of leading teams, therefore, must aim for the top simply because the possibility exists that rivals will beat them to the real prize, Champions League football.

In this scenario each transfer window becomes an auction for a single, collective, “top four.” The Glazers are bidding blind since they have no accurate idea of how other clubs will behave. This auction is particularly cruel in that they have to pay up even if they lose.

Glazers Pay Off Matrix

However, it is irrational to bid the true valuation of a top four finish in this ‘auction.’ It doesn’t make sense for the Glazers to pay more than the value they believe Champions League football is worth. Should the Glazers bid to the limit of their valuation, the Tampa Bay Buccaneers owners will not make any profit even if they “win.” By bidding lower than true value of Champions League football, the Glazers will minimise their loss. Not only do they minimise the loss, but they really profit if, and only if, they pay less than the value available to them.

Notice that the Glazers’ expected payoff increases as they lower their bid to reach the top four. Indeed, the Glazers siphoned much from the club’s coffers prior to 2011, during a period in which they supported Sir Alex Ferguson with words rather than the wallet.

However, losing out on £40 million by refusing to spend £20 million is frustrating to any investor. The Glazers stumbled onto this conclusion in 2011 when City became a genuine contender for the title. Paradoxically, investing as close to true valuation of Champions League qualification as possible minimises risk at this point.

United remains a force in the Premier League so heavy recruiting will only reinforce an already strong squad. As long as the Glazers spend less than their valuation of a top four finish, they will profit – and this has proven to be their primary motivator. If the Reds bring in even one player less than they should, however, the Glazers will then ‘lose’ the entire summer’s worth of investment. The family should have learned that lesson after the failed Moyes experiment.

Tottenham Hotspur and Arsenal will also challenge next season, with Everton lurking on the periphery. Only the Glazers know how much the Champions League is truly worth to them – in prize money, matchday income and sponsorship – but they now have to spend closer than ever before to the expected outcome. In this scenario £150 million might very well be available this summer.

Data Rant: United’s summer game of Moneyball

May 13, 2014 Tags: , Data 19 comments
featured image

Manchester United will start the 2014/15 season with only domestic honors at stake. Being out of Europe, however, could turn out to be a blessing in disguise, enabling an overhaul of a squad in desperate need of renovation. Louis van Gaal certainly has the track record to revitalise a slumbering giant, but the Dutchman faces a great challenge at Old Trafford.

First, van Gaal must attend to United’s attack – there less goals at Old Trafford compared to the season before and the decline in the chances created is noticeable.  Wayne Rooney has maintained his shooting accuracy while Danny Welbeck has improved significantly. Robin van Persie has not matched the form of last season, but the addition of Juan Mata should have more than made up for the Dutchman’s decline.

United goals and chances

Shot Accuracy

Meanwhile, the charts below show the relationship between chances created/assists and goals at all 20 Premier League teams this season. Assists and goals scored understandably have a particularly strong correlation. In the past season the Reds have seen the number of assists drop dramatically – and while United has not only created less chances, they’ve created poorer chances too.

Figure-1 - Chances vs Goals

Figure-2 - Assists vs Goals

Michael Carrick created four goals from deep last season, while he failed to record any assists in the campaign just finished. Similarly, three assists from Ashley Young last season has dwindled to just one in this. The summary is that United’s midfield has been mediocre offensively and improvements, from players already at Old Trafford or otherwise, must be made.

However, the 2012/13 Premier League winning side has been augmented by Adnan Januzaj and Mata. Given the offensive talent at hand reclaiming the title next season remains a realistic goal, but to do that United must match Premier League winners Manchester City.

Manchester City goals assists chances

Manchester United goals assists chances

As a unit United’s midfield has scored 0.71 goals, created 3.7 chances and recorded 0.5 assists less than the City engine room per game over the past season. The need for ‘a David Silva’ is obvious, but curiously United’s midfielders tackled more than rivals at City. The Blues bettered United in maintaining possession so one interpretation is that the Reds simply had to regain the ball more often.

Successful tackles United

Successful tackles City

The chart above demonstrates that Fernandinho played in a holding role for City. The fact that there are four United players – Tom Cleverley, Maroune Fellaini, Antonio Valencia and Carrick – who did a lot of defensive work paints a negative picture of David Moyes’ tenure at United. Cleverley and Fellaini, combined, scored and created less than Fernandinho. The data also points to the need for a defensive midfielder who is also at ease in the attacking third – if only because Mata and Januzaj do less defensive work than Silva or Samir Nasri.

However, United does not have to recruit a single player. Modern football is a squad game and the objective is to match City’s midfield. Sir Alex Ferguson ushered in the era of rotation by switching between Teddy Sheringham, Dwight Yorke, Andy Cole and Ole Gunnar Solskjaer  in 1998. With the three worst performers in Fellaini, Cleverley and Young to be replaced, the goal is simply to buy: 0.82 goals, 5.3 chances and 0.61 assists per game. One player must be capable of playing as a holding midfielder.

Yet, the new manager may be bound not only by the Glazers enforced austerity, but also by circumstances. United’s defence requires attention given that Nemanja Vidic, Rio Ferdinand and possibly Patrice Evra will all leave this summer. It means that a top signing in the faltering engine room may prove to be a transfer too far.

Yet, just like the Oakland Athletics led by Billy Beane – of Moneyball fame –  United could resort to the bargain bin for a hidden gem this summer. For the exercise our search criteria is as follows:

  • Up to three players
  • Under 25 to fit the Glazers’ typical policy
  • Not farfetched e.g. does not play for a top, established side
  • Played at least thirty games in the top-tier in England, Germany, Spain, Italy or France last season
  • At one of Europe’s more ‘pliable’ clubs.

To complete the analysis we need the figures – and a recognition that one pricey signing’s impact can be matched by two cheap players putting in twenty games each.



The data shows that there is no two-player combination that solves United’s midfield problem in one go, but purchasing Marco Reus and Mainz’ Johannes Geis will leave the Reds only 0.16 goals and 0.2 chances per game short of City. Having declared that he will stay at Borussia Dortmund, Reus will need more wooing than previously thought. There is hope though. While the €35m buy out clause might be cheap for a player of Reus’ caliber, United get the same number by purchasing:

HSV’s Hakan Calhanoglu OR Dortmund’s Henrikh Mkhitaryan plus Hoffenheim’s Roberto Firmino OR Sociedad’s Antoinne Griezmann OR Montpellier’s Remy Cabella. Plus a pick of holding midfielders – Geis is the most creative, but Southampton’s Morgan Schneiderlin, Udinese’s Allan and Lille’s Idrissa Gueye are more dependable.

At a pinch United can add a 0.30 -0.40 goal-a-game player and then hope that Januzaj develops or Shinji Kagawa regains the form he showed at Dortmund.

Young, Cleverley and Fellaini are each on high wages so cannot be moved on easily, which means United might settle for a holding midfielder to anchor a 4-3-3 formation under van Gaal. Perhaps it is worth noting that only Malaga’s Ignacio Camacho created less than Fellaini this season – the Spaniard scored, tackled and intercepted more though.

It must, of course, be said that an individual’s statistics are simply manifestations of their team-mates and a team’s tactics. Ferguson infamously sold Jaap Stam on the basis of decreased number of tackles and was proven wrong. Still, statistical analysis is a valuable tool in evaluating players and due diligence is generally carried out in multi-million-pound deals. After all, United has learned that lesson the hard way with Moyes.

* all data from domestic league games
** all data from Squawka
*** assumptions dictating linear regression have not been held strict

Data Rant: manager Giggs could benefit from time on the road

May 7, 2014 Tags: , Data 10 comments
featured image

Manchester United’s tepid 1-0 loss to Sunderland last weekend took the wind out of the sails of those advocating for Ryan Giggs’ appointment as permanent manager at Old Trafford, even if the Welshman’s team followed up with a fine win over Hull City. After all, David Moyes’ disastrous tenure has dampened United fans’ appetite for experiment, despite the lure of the “class of ’92”.

Still, as things stand, prospective appointment Louis van Gaal is simply a better choice for the job than Giggs. The argument for the Welshman revolves around the idea that Giggs “understands” United and its philosophy, but one can be successful without being in tune with the club in modern football.

Whatever the philosophy, players must eventually deliver it on the pitch, and a club’s youth system, no matter how successful, cannot produce the entire first team squad. In this sense, a club’s performance is bound by finance rather than the manager alone – even if it is a coach of van Gaal’s experience. The Dutchman has won 15 major trophies with culturally unique clubs such as Ajax, Barcelona and Bayern Munich.

There are exceptions to this rule – as Gary Neville has argued. Some, including Pep Guardiola, have taken to their jobs with great gusto. Jurgen Klopp spent his entire playing career at Mainz before managing the club. Manuel Pellegrini, another one club man cum manager, ended up relegating Universidad de Chile. It begs the question: was Anders Lindegaard sincere or sycophantic in branding Giggs the “next Guardiola?”

We can look at a range of data to find an answer. First, we look at the top 25 clubs by UEFA coefficients – to see if managers’ playing career has any influence on their managerial performance.

Win Ratio / Playing Career

It is clear, above, that there is little relationship between length or quality of playing career and that as manager. Andre Villas-Boas has built a successful coaching career despite having never actually played professionally. And despite the popular conception that good players are poor coaches Atlético Madrid’s Diego Simeone and Paris Saint-Germain’s Laurent Blanc rank very highly in terms of winning ratio.

Taking the tenuously relevant trend as guide, David Moyes had under-performed at United, while Guardiola is doing better than expected given his playing background. Essentially, a top 25 manager must produce at least a 59 per cent win ratio – and given this analysis Giggs’ stellar Old Trafford background is expected to produce a 60 per cent winning percentage over a coaching career. It would be enough to qualify for the Champions League.

To be more accurate in assessing Giggs, however, we cannot simply compare the United legend to established managers and must consider natural talent. It is possible that the “Welsh wizard” is an innately gifted manager who will belie his callowness should he be appointed?

Win Ratio / Managerial Career

Above is a comparison of managerial experience – number of clubs managed and number of years of training before a first managerial post – and win rate. Again, there seems to be little correlation, although most managers lie in the middle, suggesting that there is a Gladwellian amount of experience required to fulfill a manager’s potential. This perhaps explains why Clarence Seedorf, a player as decorated as any, is doing poorly at AC Milan this season, while Simeone and Blanc, who are both much more experienced managers, are considered among the best in Europe.

Given this analysis Guardiola’s relatively rapid success suggests a possibility of genius – a category in which Giggs could possibly reside. There are five managers who hold a similar career trajectory as the United caretaker boss:

  • Seedorf – spent a significant amount of time at AC Milan then became manager
  • Moyes – player/manager at Preston North End then made permanent
  • Guardiola – a long history with Barcelona – became manager after a year of apprenticeship at Barcelona B
  • Klopp – one club man at Mainz then took over the manager’s job
  • Pellegrini – 0ne club man at Universidad then took over the manager’s job

First job win ratio (%) / Career win ratio (%):

  • Seedorf – 50/50 (on his first job)
  • Moyes – 48/45
  • Guardiola – 67/73
  • Klopp – 40/49
  • Pellegrini – 33/50

Indeed, Guardiola took to managing immediately, while Moyes regressed despite moving up a division with Everton and then taking over the reigning champions in United. Pellegrini, on the other hand, first won a trophy in 1994 – six years after his first job. The Manchester City manager was a slow starter. So is Giggs a ‘Guardiola’ or a ‘Pellegrini’.

Win Ratio / One Club Men

Above, we look at number of playing appearances for the manager’s first club and see whether that knowledge had any role in the subsequent winning ratio. Guardiola is a clear outlier and there is a downward trend with Guardiola removed. This is bad news for Giggs’ supporters, although it does not answer the question: what makes Guardiola special?

Win Ratio / Number of Countries Played In

Above, the horizontal axis charts the number of countries in which each manager has played prior to taking his first job. Guardiola had played in three continents before taking over at Barça B. The Spaniard, along with Villas-Boas and Carlo Ancelotti – who have also played/managed in four countries – form the top three in terms of win ratio at their current club.

Win Ratio / Number of Countries Managed In

The number of countries each has worked in has the strongest relationship yet to managerial success. In fact this data suggests that to break the 70 per cent win ratio mark in a top European club, a manager needs to have been to at least four countries.

The advantage of being cosmopolitan is obvious – exposure to diverse football cultures can only help a manager’s knowledge. This observation strikes close to home: Sir Alex Ferguson’s ex-players have yet to distinguish themselves as managers, with Mark Hughes of Stoke City the best of mediocre bunch. Is it a coincidence that the former United striker has Barcelona and Bayern on his resumé?

This is more a plausible argument than ‘proper’ statistical research, yet the conclusion also makes sense. An English player might have his horizon broadened by moving to the Eredivisie and being taught 4-3-3, for example. An expatriate footballer will also have to cope with a new culture and language – surely players pick up pearls of wisdom as they wander the globe. With United in peril, van Gaal will surely be appointed to steady the ship. Giggs might benefit from a voyage or two himself.

A brief note on methodology:
1) All categories are weighted equally
2) Each figure has been adjusted relative to the ‘best’ in each category
3) Assumptions dictating linear regression have not been held strict – 25 managers should be enough in deriving general conclusions
4) In case of a caretaker manager the pervious permanent coach’s figures have been used