Happy Face's Numbers

[PaddockLad 17289]

This will give HF a commotion in his underwear department

 

http://www.forbes.com/sites/zachslaton/2013/09/30/manchester-united-or-tottenham-hotspur-who-is-more-likely-to-be-the-newcastle-united-of-201314/

[Happy Face 29]

That's a thing of beauty.

 

The "NURF"!

 

[aimaad22 4156]

Wonderful. Nurf

[The Fish 10872]

[Happy Face 29]

[The Fish 10872]

Could that mean last season was a bedding in process? Or simply that this year we have a striker?

[Happy Face 29]

Looks to me like we've improved at both ends, not just in bringing in Remy.

 

I think what we had previously was a lot of championship standard players peppered with some that could cut it in the top flight. In January we added a third of a team that are capable of playing in this league...but as well as getting up to the pace of the game in England they had bank accounts, housing, kids schools and all that to cope with. 10 months later they're hopefully settling into both the league and their homes and we see the benefit, there's still mistakes happening, but the solid performances are starting to outnumber them & the teamwork seems to be approaching 11/12 standard.

[Park Life 71]

Yeah ManC main issue is in defence. Once Kompany is out they are all over the place and he also gives them much needed leadership.

[Happy Face 29]

 

Points total for every position for every season since the top flight was 20 teams....and the average.

[McFaul 35]

 

Points total for every position for every season since the top flight was 20 teams....and the average.

You've messed this table up. 2004 56 points "Souness", that had nothing to do with Souness, he came in the September of the following season. "Grrrr".

[Happy Face 29]

You've messed this table up. 2004 56 points "Souness", that had nothing to do with Souness, he came in the September of the following season. "Grrrr".

 

Got the formatting wrong for most of them to the left of '04

[Happy Face 29]

 

Fixed, I think.

[Happy Face 29]

 

 

When Simon Wilson first arrived at Southampton football club he was a consultant for a technology startup called Prozone. Prozone had developed a proprietary player-tracking software which, fed by eight cameras around the pitch, would output a two-dimensional bird's-eye animation of a football match. The machine could track each player's movement every 0.1 seconds, registering an average of 3,000 touches of the ball per game, and provide an answer to a range of statistical questions. Southampton adopted Prozone and later hired Wilson to work as a performance analyst for the first-team manager.

 

"Prozone wasn't part of the culture of the game and most managers weren't used to it," Wilson says. "I was naïve but I couldn't understand why they didn't want this kind of information." Once, just before an August 2005 football league Championship game between Luton Town and Southampton, Wilson gave a pre-match briefing to the team and the manager, at the time, Harry Redknapp. "Harry was more intuitive than analytical," says Wilson. "He was nervous about overloading the players with information." Southampton lost 3-2. On the team bus, Redknapp turned to Wilson and said, "I'll tell you what, next week, why don't we get your computer to play against their computer and see who wins?"

 

Some managers, however, did get it -- and one in particular was Clive Woodward. He had been the coach of England's World Cup-winning rugby team in 2003, and in 2005 had been offered a one-year contract to serve as Southampton's director of football. He had been the first coach to adapt Prozone to rugby, installing it at Twickenham four years before the World Cup, which allowed him to collect data on how England and its opponents played. "When I first saw it I was fascinated because I'd never seen a game where you're looking down and just see dots and data and movement," Woodward says. "It removed a lot of the preconceived notions we had about how other teams played. It made a big difference when we started to see them as data, as opposed to teams we had never beaten before." Once, after his players insisted that there was no space on the field to run into, Woodward took a printout of a Prozone freeze-frame taken 24 seconds into a match against France. It showed both teams around the ball in a small area on the pitch and acres of unoccupied space everywhere else. He stuck it on board with the message: "The space is the green stuff".

 

"Clive would challenge me at every level," says Wilson about Woodward's time at Southampton. "He would ask questions about every aspect of the game: why do we spend so much time working out how to score goals and not how to stop them? I would try to explain to him what they're doing and he'd just keep asking why." Woodward and Wilson tried things such as filming players striking the ball, to study technique from a biomechanical perspective. Those initiatives, however, never had much impact. Redknapp left before the end of the year and Woodward departed at the end of his contract. Wilson had left the club shortly before Woodward, convinced that there was a better way of running a club. "Woodward believed that evidence, be it video or statistics or any kind of data, was fundamental to how you prepare a team," Wilson says. Woodward remains his biggest influence. "He taught me that we didn't have to do things just because they had always been done in a certain way."

 

Today, 19 of the 20 Premier League teams use Prozone. Each has its own team of performance analysts and data scientists looking for the indicators that quantify player performance, the events that determine matches and trends that characterise seasons. They are scientists dissecting the world's most popular game, looking at data from Prozone and other sources to understand what dictates the difference between winning and losing. In the environment of the multimillion-pound Premier League, clubs don't just want a competitive advantage, they need it.

 

 

 

 

At 3.50pm on 19 March, 1950, a Royal Air Force accountant called Charles Reep went to watch Swindon Town versus Bristol Rovers. During the game, he took out a pencil and a notepad and started scribbling down observations using a system of symbols he had invented for annotating events on the pitch. Today, sports scientists call such a system notational analysis. Over the years, Reep annotated more than 2,200 matches. The data analysis for each game would typically take him 80 hours. The 1958 World Cup final alone took him three months to analyse.

 

Reep showed that football, a dynamic and unpredictable game, had constant and predictable patterns. He found, for instance, that teams would, on average, score once every nine shots; that 80 percent of goals were scored from movements of fewer than four passes; and that 50 percent of goals came from balls recovered within 30 metres of the goal line, the last third of the pitch. Reep concluded that teams would be more efficient if they spent less time trying to string together passes and more time lobbing the ball into their opponent's area. This strategy became known as the long-ball game.

 

There are two problems with the long-ball game. Firstly, it's not great to watch. Secondly, Reep's data analysis supporting it was too simplistic. In 2005, Ian Franks, a professor at the University of British Columbia, and Mike Hughes, a mathematician who pioneered computational-notational analysis, looked at the data from two World Cup tournaments. At first, Franks and Hughes found data compatible with Reep's analysis but, after closer scrutiny, they showed that most goals happened after fewer than four pass movements simply because most movements in football are like that, not because the odds were better. In other words, the frequency of goals is not the same as the odds of a goal being scored. What Hughes and Frank found was that teams that completed more passes had a better chance of scoring. "Of course, you need skilled players to sustain long-passing moves," says Hughes. "Up until then, everybody was ignoring the blatant fact that teams who weren't using a long-ball strategy, like Brazil, were winning the World Cup."

 

"Collecting data is always the first step and Reep was a great accountant," says Chris Anderson, a political economist from Cornell University, New York, who's been studying football statistics for three years. "But he wasn't a great analyst and had a limited understanding of what the numbers were telling." Reep, according to Anderson, had very strong preconceived notions and when he found what he was looking for -- a chance to play the game with minimum input for maximum output -- he didn't investigate other hypotheses like another analyst would do. "He was welcomed into football by people who wanted to play a long-ball game and just wanted to know how do it without considering how wrong this approach could be." In their book, The Numbers Game, Anderson and co-author David Sally write: "Reep's quest to use the numbers to inform strategy fell short because he was an absolutist, determined to use his data to prove his beliefs. He needed to abandon his idea that he was looking for the one general rule, a winning formula, and learn to seek the multiple truths and falsehoods in the numbers themselves." Reep's assertion that statistics offer us a chance to see things we'd otherwise miss was absolutely correct.

 

 

When Wilson joined Manchester City in 2006 in order to start a new department of football analytics, he hired the best analysts he knew and set himself the goal of changing how the football team used data. "After a game there wasn't any kind of analysis," Wilson says. "Emotionally, the manager and the coaching staff would just draw a line and move on. It was part of the culture. They wouldn't ask themselves if the game plan had been right or even well executed. My team of analysts had to fight that habit and create a continuous loop between what happened in games, why it happened and what we are going to do next time."

 

At the time, City were a mid-table club that struggled to win away games. In September 2008, when the club was acquired by the Abu Dhabi United Group for Development and Investment, a private-equity outfit owned by a member of the Abu Dhabi royal family, the team suddenly found itself with the resources necessary to mount a challenge for the Premier League. Today, Wilson is Manchester City's manager of strategic performance analysis. He co-ordinates five departments, including the team of performance analysis, which is now led by a sports scientist named Ed Sulley. After each match, they compile exhaustive reports about the team's performance data, focusing on statistics that they think most relevant. The list is extensive. They analyse, for instance, the number of line breaks, a term borrowed from rugby which means a forward pass that goes through the opposition's midfielders or, more crucially, its line of defenders. They look at what happens in the 20 seconds after the team wins or loses the ball. They pay attention to City's ball possession in the last third of the pitch, a measure that they found to be strongly correlated with winning matches. "When we studied the profile of the top teams against average teams, the thing we saw was that the best teams dominate the possession of the ball in that part of the pitch," says Wilson. "The success rate of the passes was very high, particularly forward passing. So now, when we recruit players, we pay special attention to individuals with high pass-completion rates."

 

Statistics such as line breaks and possession in the last third are important for City but would probably be irrelevant to a team with a different style: football analytics is a discipline in which the way a team plays dictates which statistics are significant. The challenge is to find out which. "Instead of looking at a list of 50 variables we want to find five, say, that really matter for our style of play," says Pedro Marques, a match analyst at Manchester City. Marques and his colleagues are currently using data-analysis techniques, such as principal-component analysis, to home in on the match-related variables about winning. "With the right data-feeds, the algorithms will output the statistics that have a strong relationship with winning and losing." Wilson recalls one particular period when Manchester City hadn't scored from corners in over 22 games, so his team decided to analyse over 400 goals that were scored from corners. They noticed that about 75 percent resulted from so-called in-swinging corners, the type where the ball curves towards the goal. "In the next 12 games of the next season we scored nine goals from corners," Wilson says. "You usually get six coaches and they'll have different experiences and they'll throw their opinions in, whereas we had objective evidence to suggest that this was a pattern."

 

When Wilson was working at Prozone as a consultant for Southampton, he would capture the information from the Prozone machine on a removable hard disk drive, commute back to Leeds at 2am, process the data overnight and head back to Southampton to deliver the analysis. Sometimes he would work 20-hour days. He lived in Leeds with ten other Prozone consultants. The office was essentially a warehouse filled with computers. The CEO was Ram Mylvaganam, an engineer who had been a marketing director for Mars. Mylvaganam didn't know much about football. On the wall of the Prozone office he hung a picture of a chalk drawing by the pavement artist Julian Beever, an artwork that, viewed from the right angle, creates an illusion of 3D. To Mylvaganam, data was like a Beever drawing -- everything they needed to make sense of the data was probably right in front of them. "But if you stand in the wrong place, the data looks like shit," Mylvaganam says.

 

Mylvaganam first had the idea for Prozone in 1996 when he was working for a management consultancy and had a contract with Derby County, a contact he got via Neil Ramsay, a former football agent. The first version of Prozone was a Portakabin filled with 22 massage chairs, developed by Mylvaganam with a local manufacturer, that emitted electrical pulses and, supposedly, relaxed the players' muscles and increased their flexibility. Prozone was a contraction of "Professional Zone". At 10.30 every morning, Derby's players reported to Prozone and sat on the chairs for 15 minutes while the assistant manager, a young coach called Steve McClaren, gave presentations on a video screen about their game plan. Feedback, McClaren used to say, was the breakfast of champions.

 

"McClaren worked in this glorified garden shed, where he would stay after all the players had gone home, with two video recorders and a screen. Just editing video," Mylvaganam recalls. "I asked him, 'Why don't you get one of your monkeys to do it?' and he replied 'How do they now what is a good move and what's bad? I want to show them how you win matches.'" Mylvaganam thought he could do better. He knew a small company in France called Video Sports that had developed pixel-tracking software. He bought 25 percent of the company and installed eight cameras around Pride Park stadium. "The camera technology was bad. Sometimes we'd get the analysis back and there'd be players missing, so we had to redesign the software in Leeds." Mylvaganam says. "Still, it was revolutionary. We were defining statistically what was a game of football."

 

In 1999, Steve McClaren was headhunted by Manchester United's manager, Alex Ferguson. McClaren requested Prozone. The firm had been working gratis for Derby and had no paying costumers, so Mylvaganam insisted on a financial deal. United agreed to pay Prozone £50,000 if the club won a trophy that year. That season, United won the treble -- the Champions League, the Premier League and the FA Cup -- and Prozone earned its first cheque. In May 1999, Prozone had two clients and no revenue. By August 2000, six Premier League clubs were paying customers. Mylvaganam and Ramsay sent their sports scientists to football clubs to act as Prozone consultants -- what some of them found wasn't what they were expecting from multimillion-pound businesses. "At Aston Villa, there was an old-school bucket-and-sponge physio who didn't really speak to the manager and a manager who didn't really speak to the players," says the current managing director of Prozone, Barry McNeill. "There were few coaching meetings, no preparation meetings and I was just a 22-year old guy in a suit who had to explain how the software I had in my PC could add value."

 

In 2000, Mylvaganam got a call from Sam Allardyce, the Bolton manager. Allardyce had played in the North American Soccer League with the Tampa Bay Rowdies, who shared a training ground with the Tampa Bay Buccaneers, an American football team. He'd been impressed with the Buccaneers' use of technology. Mylvaganam didn't think a lower-division team such as Bolton could afford Prozone, but also knew that if they performed well, they could prove even better publicity than United's treble. Bolton became the first lower-division team to use the system. They beat Preston in the Championship playoffs 3-0 and were promoted to the Premier League.

 

At Bolton, Allardyce conceptualised a rigid game plan around data. His backroom staff included David Fallows, a former Prozone analyst, Gavin Fleig, who had studied under Mike Hughes, and Ed Sulley. Allardyce and his performance analysts had a model he called "the Fantastic Four": four areas that dictated success. Out of 38 games, they knew that a team had to prevent the opposition scoring in at least 16 games to avoid relegation. They knew that if they scored first they would have a 70 percent chance of winning the game. They knew that set pieces, free kicks and corners accounted for nearly a third of the goals scored, and in-swinging crosses were more successful than out-swinging, so they practised not only those types of crosses but also defending against them. They also discovered that they would have an 80 percent chance of not losing if the players outworked their opposition by covering more distance at speeds above 5.5m/s. Allardyce insisted on players using long throw-ins, deep into the opponent's area -- if a player failed to follow that simple command he'd go crazy because he knew the odds of scoring had been reduced. Bolton's performance analysts studied a huge number of throw-ins and Allardyce would organise players in the places on the pitch where the ball had the highest probability of landing, the so-called positions of maximum opportunity, or "pomos", to increase the odds of scoring. "Pomos weren't just relevant to throw-ins. In training, he would shout to the players to attack their pomos when trying to score," says Sulley. Between 2003 and 2007, Bolton recorded consecutive top-eight finishes in the Premiership, a record of consistency bettered only by the top four. They qualified for the UEFA cup for the first time in 2005 and again in 2006. When Allardyce left in 2007, they had an impressive 39 points after 21 games.

 

Sally and Anderson argue in The Numbers Game that football is as much a matter of skill as it is a question of luck. Goals are rare and studies show that 44 percent are fortuitous. In any match, the favourite team wins only 55 percent of the time. Football, they conclude, is a game dominated by randomness. That, however, doesn't mean that nothing can be done to influence its outcome. Football's inherent randomness makes analysis even more impactful. "What makes a difference isn't data," Anderson says, "but the brain cells that can translate data into a theory of how you win football matches."

 

 

Analysts used to believe, for instance, that the distance run by a player was a good indicator of individual performance and that a team's ball possession had a positive correlation with winning. Those numbers, however, turned out to be meaningless. Analysts now know that it is the distance run by a player when sprinting that indicates good performance, and that it is ball possession within the last third of the pitch that correlates with success. Better metrics imply a more refined understanding of the game. "Sometimes we look only at the individuals and forget the context," says Blake Wooster, a former director at Prozone, who now runs a sports startup called 21st Club. "For instance, Barcelona's [Lionel] Messi is one of the best players ever, but what would happen if you took him out of that context and put him in another team? You can't assess talent in a vacuum." An example of that type of contextual statistics is a model recently developed by Prozone called "goal expectation".

 

This assigns each shot a probability related to its position, and thus determines how well a goalscorer is performing. The statistic filters out the quality of the opposition and the quality of the player's team. Last year, for instance, Tottenham's Gareth Bale had 161 shots and 21 goals, when, according to the goal-expectation model, he was due to score only 11. "Bale would regularly shoot from situations with a low probability of success, such as from a distance of 30 yards, and score," says Paul Boanas, Prozone's senior account manager and a former performance analyst. "This type of contextual information helps to explain why he's worth so much."

 

 

Some of the most important elements of football remain very hard to quantify and it's difficult to understand what we can't measure. Consider defence. Using data from the last ten seasons of the Premier League, Anderson and Sally compared the value of a goal scored and the value of a goal conceded. They found that scoring a goal, on average, is worth slightly more than one point, whereas not conceding produces, on average, 2.5 points per match. "Goals that don't happen are more valuable than goals that do happen," Anderson says. "It's counterintuitive. The question is: how do we measure something that doesn't happen? The challenge is to see the unseen."

 

Evaluating an attack consists of measuring what happens with the ball: shots, passes, crosses, sprints. Although actions such as tackles, clearances and saves give you a measure of defensive performance, the essence of defence lies in collective behaviour that happens off the ball -- marking, cutting off passing channels, the positioning of defenders. This is a thorny problem that the analysts at Manchester City are beginning to study. Using Prozone's tracking data, they are quantifying variables such as the area occupied by a team and the dispersion of the players. "We are trying to understand how the individual players co-operate and develop synergy as a team," Marques says. "Most analysis still focuses on discrete variables and actions, but most important for us is to understand the interactions."

 

Every week during the 2011-2012 season, Manchester's City captain, Vincent Kompany, sat down with the other defenders and a performance analyst, and examined their performance. "They would look at videos and statistics and ask questions," Wilson says. "Was the pressure effective? How many forced errors did they commit? What would happen in the ten seconds after losing the ball? On the basis of that analysis they would design their own defensive tactics for the game. You can have a fantastic analytics team but you can never win a game with data if you're not influencing the behaviour of the players." At the end of the season, Manchester City had conceded the fewest number of goals in the Premier League. "We beat a lot of records that season," Wilson says. "Most of the credit is due to the talent on the pitch. But I believe that about 30 percent of that success is down to how well we prepared and maximised that talent."

 

Wilson missed the last game of that season, when City played Queens Park Rangers. City were level on points with Manchester United, but had a superior goal difference. "I had a flight but it was delayed, so I ended up only watching the first half on TV," recalls Wilson. "By then when we were winning 1-0, so I was confident." In the second half, QPR scored twice. Two minutes after stoppage time, City's striker Edin DŽeko equalised. By then, United were winning their match and, if nothing changed, would be the champions.

 

Two minutes later, City's attacker Sergio Aguero received the ball on the edge of the box, in a position to shoot. According to Prozone's goal-expectation model, he had a 12 percent chance of scoring. Instead of shooting, he went around a defender to a corner of the penalty area and, from a spot where he had a 19 percent chance of scoring, slotted the ball past the keeper. By the time Wilson landed at Gatwick, the news ticker running across the TV screens was saying that Manchester City were the new champions.

 

http://www.wired.co.uk/magazine/archive/2014/01/features/the-winning-formula

[Park Life 71]

Prozone is only as good as the intelligence of the people using it.

[Happy Face 29]

Prozone is only as good as the intelligence of the people using it.

Think it's interesting Allardyce saw a lot of success as an early adopter, but as everyone else took it on his ability to get any edge from it seems to have dwindled.

[Happy Face 29]

 

 

Comparison of how we picked up points with 2 years back when we finished 5th.

 

Need to beat both Norwich and Sunderland to keep up.

 

If we do, then we'll have surpassed last years point total just as the transfer window closes.

[beatty 24]

That wired article is a bloody good read.

[Park Life 71]

 

 

Comparison of how we picked up points with 2 years back when we finished 5th.

 

Need to beat both Norwich and Sunderland to keep up.

 

If we do, then we'll have surpassed last years point total just as the transfer window closes.

Think we'll get 6 pts from those two. Both crap sides.

[The Fish 10872]

Think we'll get 6 pts from those two. Both crap sides.

 

Can't see either of them doing much business before we meet, either. Even if they do, they'd need wholesale investment to match our first eleven. Our strengths seem to nestle quite neatly amongst their weaknesses as well.

 

 

[ewerk 30656]

(Team has no significant strengths)

 

And is losing from winning positions a major weakness of ours? We've lost two games all season from winning positions, hardly a disaster.

[The Fish 10872]

(Team has no significant strengths)

 

And is losing from winning positions a major weakness of ours? We've lost two games all season from winning positions, hardly a disaster.

 

It doesn't say that, it says "Protecting the lead".

 

I'm not 100% sure if that means points lost when winning (7pts) or if it means simply surrendering the lead. So say we go 1-0 up, they equalise, but we go on to win, WhoScored.com may include that as failing to protect the lead?

 

As I say, I'm not sure, but that may be it?

[Happy Face 29]

do they only base it on a current season?

 

We've dropped 11 points and 18 points in the last couple of seasons having scored first and look like we're on the way to hitting the average of that again.

[The Fish 10872]

A quick tally shows that the opposition have equalised (regardless of the result) 6 times from 26 games. Now, is surrendering the lead 1/4 of the time enough to be considered a weakness?

 

do they only base it on a current season?

 

We've dropped 11 points and 18 points in the last couple of seasons having scored first and look like we're on the way to hitting the average of that again.

 

I had assumed they would only consider this season, but this could absolutely be true.

 

You'd think we'd still be considered a long ball side though, we were certainly near the top of that depressing stat last season.

Edited January 21, 2014 by The Fish

[Park Life 71]

Definitely not a long ball side now.

[The Fish 10872]

Definitely not a long ball side now.

Exactly, which is why I thought HF's suggestion that whoscored.com took data from a longer period might be a little off? Unless of course it's taking 2 or 3 seasons into account?