Performance Analysis Blogs of The Week (17/07/12 - 24/07/12)

Every week, via Twitter I see a vast number of Performance Analysis blogs which deserve every chance they can get to be read by as wide an audience as possible.

Since starting my blog I had the idea to include links to the best of these in a post each week so that people had every chance to view them. I've since seen a couple of other blogs which do something similar by I'm going to do this anyway, at least then it gives me something to refer back to if I ever need to!

James Grayson's excellent blog post about the correlation of Total Shot Ratio to success (tweeted via @OptaPro)

The always excellent Power Of Goals blog by Mark Taylor looks at an example of Arsenal's shot comparison when taking stronger and weaker feet into account (follow Mark at @MarkTaylor0)

From the Performance Analysis in Sport Twitter feed, a link to Chris Anderson's Soccer by The Numbers blogpost looking at Height discrimination when referees award fouls (tweeted by @Perf_Sport)

Some developments in American Football about the use of the iPad in the modern game, will probably be the next step in football, a lot of clubs already using this (Thanks to New York Red Bulls Performance Analyst Dave Lee for the link to this one, follow him @davelee_NY)

Richard Whittall's State of Analytics blog looks at the recent buzz that crosses aren't an effective way to score goals (follow Richard @RWhittall)

And following on from that, a post on the EPL Index about the statistics behind how many crosses it takes to score a goal (Thanks to Simon Kuper for this link - follow Simon @KuperSimon)

New research into penalty shoot outs for International teams produced by Alex Bryson (Thanks again to Simon Kuper for the link)

Not so much a blog but very interesting news about the new Adidias MiCoach tech and how it will be pioneered in the MLS (Thanks to Simon Banouby for the link - follow Simon @Banouby)

More on MiCoach from We Ain't Got No History and how it might not be the analysts dream that it is made out to be (tweeted by Carefree Chronicles - @CareFreeChronic)

Ford Bohrmann's Soccer Statistically blog looks at how Twitter measures Transfer Rumours (follow Ford @SoccerStatistic)

Some excellent mini posts on Paul Riley's differentgame blogsite, the most recent of which looks at the decline of the Arsenal back 4 (follow Paul @footballfactman)

Excellent work on Ben Pugsley's Bitter & Blue website looking at how possession ultimately played a factor in Manchester City winning the Premier League title (follow Ben at @benjaminpugsley)

And finally, more from Mark Taylor's Power of Goals about how venue affects shooting accuracy


That's all for this week, I hope this has been useful

Also follow me @donceno and if you haven't already check out my statistical post on Euro 2012

Euro 2012 – A Statistical Overview of the European Championships in Ukraine & Poland

Euro 2012 – a Statistical Summary

Like all keen football fans I watched the 3 week spectacle that was the 2012 European Championship finals with enthusiasm, enjoying the attacking football on display and the change in temperament from the previous World Cup in South Africa when many teams seemed to come to defend and try to not lose rather than win.

Given the increasing thirst for statistics and analysis in modern professional football, I decided to follow up my previous blog with a look at the tournament as a whole. What patterns emerged, was anything surprising, did anything stand out and can the statistics tell us how the teams played.

This will be split down into 3 parts as follows: -

Part 1) An overview of the whole tournament. I have compiled some information into graph format, to show the differences between the 16 teams, which should help explain why Spain won

Part 2) 10 Impressive players. This isn’t the usual run of the mill list including Xavi, Iniesta and Ronaldo, everybody in the world knows they are fantastic players. I have tried to look beyond this to players that maybe weren’t as well know, especially in England before the tournament began. I have backed this up using statistical performance data.

Part 3) 10 Disappointing players. Again, this is a statistical look at 10 players who didn’t hit the heights expected of them; maybe they didn’t score enough, pass enough or even play enough. I’m sure a couple will raise some eyebrows but hopefully you’ll see how I have used statistics to define why these players were chosen.

Finally I’d like to thank Opta for their outstanding Stats Zone app for iPhone, from which I pulled the majority of this data.

Euro 2012 – an Overview

Football statistics are more plentiful now than ever before. Many people believe we are in a golden age of information with reams of data available at our fingertips. The problem with having all this information is that sometimes you can’t hear what you want for the rest of the noise. By using data collected by Opta I analysed over 25 different key match statistics and their varying degrees of success to see if any patterns emerged which could help deduce why the outcomes were as they were.

Couple of early provisos on this information: -

1) As half the teams only played 3 games and the maximum played was 6 by Italy and Spain this information has a large margin for error. For example, you will see that I analysed Attempts on Goal, after 3 games the Czech Republic had a quite respectable average of 12.3 attempts on goal per game. However in their Quarter Final against Portugal they either didn’t try to attack/were unable to (depending on your viewpoint!) and only had 2 attempts bringing their average to 9.75 per game. This large movement (almost a quarter in one game) shows that in this kind of tournaments it is more about form than reading what will happen in the long term. There is an excellent blog post on this recently by Ed Feng at The Power Rank (http://thepowerrank.com/blog/)

2) Whilst statistics are a valuable tool they should always be taken at face value for what they are. Further digging into the information can sometimes reveal more clues as to why events happened. For example, when Italy played England they had a tournament high of 36 Attempts on Goal. Sounds impressive, but further investigation reveals that 25 of these Attempts were from outside the penalty area. While shooting from outside the penalty area wouldn’t be frowned upon, it would appear that speculative shooting was the order of the day, especially against a world-class goalkeeper like Joe Hart. After 120 minutes they hadn’t scored.

FIFA Ranking & Pre-Tournaments Favourites

The first thing I’m going to look at is the pre tournament odds. I took the odds to win the tournament from William Hill on 07/06/2012 and compared them with their official FIFA Co-efficient Ranking on that day.

*Click the graphic to enlarge

The first thing that stands out is how far from the rest of the teams the 2 host nations were ranked. Despite both putting up a relatively good show, neither qualified from the group stage and it seems to be weaker teams in recent years that have hosted tournaments (South Africa, Austria, Switzerland).

The rank outsiders were Denmark at 100-1, despite being ranked in FIFA’s top 10. This was largely due to the strength of the group they were in, with Group B containing 4 teams from FIFAs top 10.

The bookies don’t often get it wrong and they didn’t in this case with the 4 semi finalists being amongst the 7 top teams, including Spain being favourites, number 1 ranked team by FIFA and eventual winners.

So, what does the graph tell us? Not much that we didn’t already know. Favourites usually win. FIFA Rankings, despite not being everybody’s cup of tea and having flaws, are pretty accurate barometer of direct success.

The Netherlands were the biggest surprise disappointment but it’s unlikely the graph above was needed to show that!

Shot Conversion Rates

The obvious correlation in football is that goals win games. Football is such a fluid, dynamic sport that you could survive 90 minutes of pressure, have 1 shot a win the match. In all likelihood though, this doesn’t happen that often. More often than not the teams who take the most shots, get the most on target and convert the better % of their chances will win the match.

*Click the graphic to enlarge

The graph above looks to compare the different rates by which teams managed to firstly get shots on target and by which they converted these efforts.

The first 2 to notice are Greece and Russia’s on Target Conversion percentages. Greece scored an incredible 63% of the shots they managed to get on target (roughly 2 in every 3 shots) while Russia scored 55% of on Target shots (just over 1 in 2 on target). As mentioned earlier the problem with this is the low amount of shots that each team got on target, Greece with just 8 shots on Target in 4 games, Russia going slightly better with 9 in 3 games. Despite this it shows that the finishing by both teams was clinical when they managed to hit the target. It’s unlikely that this would have continued if there was a greater statistical sample to pull from but it shows that in Greece’s case from only 32 shots in the whole tournament they scored 5 goals, a very healthy conversion rate.

In terms of actually hitting the target Spain, Germany and somewhat surprisingly Sweden had the highest % of shots that hit the target.

Clinical strikers could be given as being the case for all 3 as especially with Spain’s fluid 6 man midfield/attack formation they would often look to play short passes and get into better goal scoring positions rather than taking pot shots from outside the area (59% of attempts were inside the area).

Germany and Sweden tended to rely on the finishing of a top class striker, with Mario Gomez and Zlatan Ibrahimovic respectively having the most attempts for each team. Ibrahimovic’s transfer to Paris St Germain taking him to a combined transfer fee total of almost £150m in his career speaks for itself whilst Gomez, despite being seemingly underrated in England, has an outstanding goal scoring record.

Passes per Game/per Goal

Spain’s ‘tika-taka’ style of football has become prevalent over the last 6 years resulting in winning 2 European Championships and a World Cup. However this style didn’t happen overnight and has evolved from Luis Aragones in 2006 and 2008 to Vicente del Bosque in 2010 and 2012. The building blocks were put in place to play this style man years ago. Even so, it seems many other teams have tried to emulate the Spanish way of playing, with varying degrees of success. So how do the other teams fare?

*Click the graphic to enlarge

Spain are far and away the team who attempted and completed the most passes with 720 on average per game and 637 completed. France, Italy, Netherlands, Germany and Russia all played a similarly passing game, in fact Italy’s may have been even higher had they not had to play Spain twice in 6 games.

What the graph does show is that despite playing the ball around a lot, Spain make it count by scoring on average for every 319 passes they completed. Contrast this with the Netherlands and France who passed the ball a lot but didn’t use it effectively.

The most effective teams in terms of goals per passes completed were Greece, Sweden & Croatia. 1 team made the Quarter Finals, somewhat surprisingly even in a poor group and 2 teams didn’t qualify.

There is no single way to play the beautiful game but it does seem that even by scoring despite not passing the ball much does not equal success. Passing can be used not only to score (such as Barcelona) but also to prevent the other team from scoring (Swansea in the early part of the 2011/12 season)

What was interesting was the next 3 teams in terms of success England, Germany and Portugal. (240-244 passes per goal)

All 3 teams could quite easily have made the final (in fact all 3 would have made the semis if England could get over their phobia of penalty kicks!) and this mid range would seem to be the most effective. Portugal and England both played a direct style, looking for the wingers early and relying on individuals to create chances and goals (Ronaldo/Nani and Rooney/Gerrard).

So although passing isn’t the defining statistic to win games, it can certainly help if used in the right way.

Attacking 1/3 Passes

I mentioned above the difference in passing and keeping possession with teams like Barcelona and Swansea. The media in England have often referred to Swansea as the Barcelona of the Premier League due to their fluid passing style. This is not strictly speaking the case as Swansea’s style of football, especially in the early part of the season, was intended to keep the ball away from their opponents and prevent them from scoring as much as it was designed to initiate attacks.

That’s where the attacking 1/3 passes come in. This statistic shows the teams who not only pass the ball but who do more passes further forward, thereby showing more attacking intent.

*Click the graphic to enlarge

This graph shows The number of Attacking 1/3 passes as a percentage of Total Passes Attempted during a game, compares this with the number of these that were complete and contrasts this with the number of Attempts on Goal per game.

Surprisingly Croatia are the team with the highest % of their passes in the attacking 1/3. Unfortunately this doesn’t result in them having many more attempts on goal than England, Denmark or the Republic of Ireland. Perhaps this is one reason they crashed out at the group stage.

Germany, Russia and Portugal also had a high percentage of their passes in the attacking 1/3, but in contrast to Croatia had a higher than average number of attempts on goal per game.

The reverse of this sees the Netherlands have the second least number of their passes in the opponents 1/3, but the joint most attempts on goal per game. In fact the Netherlands and Russia had the most attempts and both exited the tournament at the Group Stage so perhaps having the ball in the attacking 1/3 doesn’t correlate with success.

Crossing & Take Ons vs Passing

This tournament saw the average number of passes attempted rise to a higher level than at any time before in European Championship history. Due to that you would think a lot of goals would be scored from short passing moves and through balls. However, this tournament saw more headed goals scored than any previous competition.

I thought it would be useful to look at the number of crosses each team did, linking this with Take Ons (simply for the fact that you would generally have to beat a defender to get a cross in although this is not always the case) and compare this with the number of passes.

*Click the graphic to enlarge

The flags on this graph represent the number of passes attempted per game (clearly showing Spain way out in front) but Germany and the Netherlands had both done a high number of crosses despite attempting well over the average number of passes (they were actually 2^nd and 3^rd). Germany averaged 25 crosses per game and notably scored from 4 of these occasions, proving a very successful tactic.

I was a little surprised to see the Netherlands average 23 crosses per game. With van Persie the central figure in a 4-3-3 (not renowned for his heading ability) and Robben and Afellay both prone to cutting inside and shooting the number was dramatically higher than I expected, especially given that they scored no goals from crossing situations. They did however have the highest number of Take Ons.

Despite doing the 5^th least passes per game on average, England equaled 6^th on number of crosses attempted (equal with France) and scored from 4 crossing situations.

Again all the graph above goes to show is that football is a game that can be played in many different styles and there is no ONE BEST WAY to play.

Defensive Ability

England were pilloried in the press back home for being so defensive. We never had the ball and played akin to Chelsea in the Champions League final all the time according to the tabloids, but do the statistics back this up?

I looked at the number of Attempts allowed on average per game, along with Tackles, Interceptions and Blocks to see if there was a correlation.

*Click the graphic to enlarge

As you can see Republic of Ireland, Greece and England conceded more attempts on goal than any other. As mentioned above however, this does not take into account where the shots were from (using Italy’s example of 25 of their 36 attempts against England from outside the box). 2 of these teams that conceded so many attempts made the Quarter finals, England could conceivably made the Semis had they done better in the penalty shoot out.

They also have a significantly higher number of Blocks per game on average than most of the other teams, England and Ireland averaging 7.5 and 8 blocks per game against an average of 3.3 per team per game. Desperate defending? Perhaps. The throwing yourself in front of the ball style encapsulated by John Terry and Scott Parker to prevent the opposition from getting a good shot on goal is clearly shown on the graph.

Tactically as England dropped so deep and effectively gave up possession to the opposition they had 2 banks of 4 on the edge of their own penalty area. This allowed plenty of opportunities to get blocks in as the opposition were forced to shoot from distance rather than try to play short balls to create space in behind.

The team with the most tackles and 2^nd most interceptions was rather surprisingly Poland; they also allowed the fourth fewest attempts on goal behind Spain, France and Portugal. That they crashed out at the Group Stage despite this seems somewhat surprising, generally if you don’t let the opposition shoot and win the ball back from them you’d have an excellent chance of winning, as demonstrated by the results for Spain.

Conclusions

These graphs are designed to highlight the statistics from the European Championships but can any conclusions really be drawn from them?

Let’s look at Spain.

1) They were ranked 1^st by FIFA and were favourites to win with William Hill

2) They got the 2^nd highest percentage of Shots on Target (42%), 7^th in converting their overall attempts into goals (12%) and joint 9^th with the Ukraine on converting their Shots on Target into Goals (29%)

3) They were clearly 1^st in Average Passes Attempted (720) and Completed (637) per Game, but were 10^th in the average number of passes per goal (319). However when you do so many passes this average still works out at 2 goals per game.

4) They were 7^th in average percentage of passes being in the attacking 1/3 (30%) and 5^th in percentage of passes completed in the attacking 1/3 (27%) but were 4^th for average attempts on goal per game (17)

5) They did the 13^th most crosses (16 per game) and 2^nd most Take Ons (18)

6) They allowed the least attempts (just 8 on average per game), Did the 5^th most Tackles (20), 9^th most Interceptions per game (16) and the least blocks (just 1 per game)

What all these stats do help us understand that you don’t have to be that clinical all the time, you don’t have to cross and you don’t have to block shots because if you control the ball with passing and possession more often than not you will score eventually and win more than you lose.

Thanks for reading, feel free to follow me on Twitter (@donceno) and check back for the next part on impressive players soon.

Finding The Right Statistic

As a budding Performance Analyst, I have always been intrigued by the use of statistics in sport, particularly in football. I have kept a close eye on the standard indicators within football such as how many goals and appearances a player has made, possession statistics, number of corners….the type of information which is easily available on the BBC Sport Football website.

While pursuing my path as a professional Performance Analyst, it has become clear just how many statistics there are within football. Companies like Opta and Prozone capture virtually every action on a pitch (roughly around 2,000 per game) all mapped out by individual players. The digital revolution has certainly aided the Sports Science side of the game. While capturing information and statistics as a way to aid the manager's decisions through training, performance, recruitment and fitness was barely thought about 10-15 years ago, we are now at a stage where there are so many numbers flying around, clubs are needing to employ a specific member of staff to sift through these and decipher them – putting them into football language!

Some clubs have as many as eight analysts (including interns) and particularly in the Premier League and Championship it is practically unheard of to not have at least one analyst. As part of the Elite Player Performance Plan (EPPP), to gain the higher levels of accreditation it is mandatory to employ a full-time analyst. This shows how much emphasis is put on the numbers game if it is something the Premier League and FA are wholeheartedly embracing and many League One & Two clubs are gradually getting on board with the opportunities this data can provide.

This got me thinking about how different players would excel in different indicators and what managers would look for when looking to buy a certain player. It should come as little surprise that according to www.whoscored.com, Robin van Persie and Wayne Rooney top the list of players who average the most ‘Shots per game’ (4.6) and Juan Mata plays the most key passes (3 per game on average), but the area I want to focus on is flair players.

What brought this to my attention was during my time working at Rotherham United, they had a talented young player called Ben Pringle. He has a great deal of potential and towards the end of the season found his feet in the team and helped push them on a late surge towards the play offs playing in a relatively free role in an advanced central midfield position. While his passing is good, he has a tendency to shoot when he should pass and pass when he should shoot, a trait that a lot of young players have and often better decision making skills come with experience. One thing he had in abundance was a combination of determination and skill and he would regularly take 2 or 3 players on before being tackled or losing the ball.

While statistically this would go down as the ball turning over to the opposition or an unsuccessful pass, it regularly excited the crowd who were right behind him and the next time he got the ball he would have players coming out of position to attempt to tackle him, leaving space for other players to exploit. He regularly won man of the match awards towards the end of the season and I expect him to be a big part of Rotherham's promotion push next season.

The closest players I can think of in the Premier League in terms of style and who compare statistically would be Victor Moses and Junior Hoilett. Both players played for teams battling at the wrong end of the table and are exciting young talents.

Hoilett and Moses top the charts for the most successful dribbles per game (Hoilett 2.6, Moses 2.5). They are two players who regularly get the ball, run at opponents and have a reasonably good product at the end of it. Despite this, they both have reasonably low number of key passes (Hoilett 1.3, Moses 1.1)

This could partly be put down to the teams they play for. If Hoilett breaks out of defence and runs at the opposition, will Blackburn’s defensive style mean fewer players are up in support, therefore fewer options? That is surely a factor to a degree. Another likely cause is that the wrong ball is chosen. This might be a cross over hit, pass misplaced (his passing accuracy was less than 80%) or an off target shot. But one thing he did manage was to get noticed by the fans, get noticed by the opposition and if rumours are to be believed, be noticed by some of the top teams in Europe.

By running at players, Hoilett excited the Blackburn fans, and was one of the few bright moments of an otherwise difficult season. This in turn encouraged other Blackburn players and had a positive effect on the team overall. Even if Hoilett lost the ball, it was likely that next time he was in possession, an opposition player came out of position to track him and potentially double up the marking on him, leaving space for someone else to exploit.

You only have to look across Europe at players like Lionel Messi (4.8 successful dribbles per game) and Franck Ribery (4.0 per game) to know the benchmark and see what can be achieved. If Hoilett and Moses can tie their considerable skill and dribbling ability in with an end product they could become stars, and at such a young age, time is on their side. With better team-mates, the potential is there for them to flourish.

Utilising analytics for recruitment has only just begun to be exploited but the value is there to see. This can lead to more astute purchases in the transfer market and depending on the clubs’ ambitions either success or financial viability. The days of a formula as in the analyst bible “Moneyball” are unlikely to be close due to the fluid nature of football as a game with many mitigating factors. It could however be the difference between that one player scoring the goal that seals promotion, the defender that helps prevent relegation or the keeper that is sold for ten times the fee paid for him, which keeps the club running for another year.

Both Hoilett and Moses are potential transfer targets for teams this summer. How many managers will look at their key pass statistics and consider that the data is telling them that their end product does not merit a £5m transfer fee? How many managers will consider other factors we have mentioned which hampered their final ball options, but may have potential to develop playing alongside better players?

The use of analysis in sport is lauded by some and criticised by others but in the modern game it is a brave team who shuns the modern way of thinking, possibly in danger of being left behind.

The post was original blogged at Onside Analysis definitely worth following on Twitter! @OnsideAnalysis

If you want to follow me on twitter just go to @donceno