Tag Archives: MCFC

Premier League Review 2012/13: Manchester City

Cumulative pts - MCFCFirst of all, lets just say that 78 points isn’t a bad total by any means. In 1996/97 it would have been enough to win the title. And it would have taken something extraordinary to beat Manchester United this season considering that they had already reached 80pts by the conclusion of game 31.

City’s season began well enough for a title-winning year, keeping pace with United until about game 15 when they followed up a home draw against Everton with a 2-3 loss to United and then a few games later experienced a disappointing away loss to Sunderland. A few more stutters were evident along the way – in particular a 3 game run from the end of January which included a draw with QPR and Liverpool and an away defeat to Southampton. This season the margin for error at the top was too small for these slip-ups to be turned around triumphantly.

Goals For - MCFCGoals Against - MCFCThe above graphs show that City struggled a bit in front of goal for much of the season, not quite hitting the heights expected of such a expensively assembled squad. Defensively, however, they were generally brilliant and had the best goals conceded record for the season.

Home & Away - MCFCNot much wrong with their home record this season, although it is hard not to compare it negatively to their unbelievable record of the season before. I’d look to the low goals scored away from home as a problem to that needs to be addressed next term.

Vital Stats - MCFCHere’s a stat that is possibly of no use: City had the heaviest team in the league last season weighted by playing time. This is despite their team being slightly under the average height, so may be something to do with the kind of nutritional/physical conditioning City are pursuing behind the scenes. It could potentially affect player speed and reaction times although I’m sure they test these things so perhaps it’s of no significance!

Relative strength - MCFCIn terms of City’s return from each segment, more than anything it looks like they didn’t score enough against teams destined for the top half (6-10th). They’d probably expect to gain an additional 0.5pts per game from these teams (against Everton in particular they only took 1pt).

% Playing time - MCFCMPS vs League TPOEM - MCFCCity fielded a lot more players who played only between 1600-2000mins than the league average – this group of players quite crucially included Aguero, who City really did miss for a large part of the season. Joe Hart and Yaya Touré were also the only players that City fielded for more than 2800mins.

If they had been able to field Aguero, Silva and Kompany in maybe 5 more games each I have little doubt that they would have improved their points total.

General stats:

  • Highest proportion of tackles won: 79.5%
  • More total chances created than United or Chelsea (509)
  • Lowest goals per shots on target ratio out of the top 4
  • Second lowest number of goalkeeper saves (78)
  • Second lowest number of fouls won
  • Highest number of final third passes attempted (6931)
  • Lowest total long balls attempted (1612)

Players - MCFCShall I say it again? City missed Aguero and Silva. People might be inclined to think that these players just weren’t as good this season – I disagree, their stats remained impressive. Richards also made a great contribution to the team when he played but again City missed him through injury for a large part of the season. Nastasic actually outperformed Kompany according to TPOEM, and Zabaleta was brilliant at right back – but we knew that already, didn’t we?!

The overall summary of this review is that City needn’t make any rash decisions and sack Mancini. Oh, wait… Fine, in that case Pellegrini (or whoever comes in to manage City) will only need to have Aguero, Silva and Kompany at his disposal more often next season to improve on this year’s points total in the league (ceteris paribus). And of course, additional attacking options will always be welcome.


Model pitfalls and further discussion of TPOEM

Since my previous post introducing a new model for football analysis, TPOEM, I have developed and integrated some significant improvements to it.

Firstly the speed in which I can give predictions based on team starting line-up (involving less manual input, more automation) is much better, so last Saturday I was able to tweet about the model’s predictions well before the 3pm kick-offs began.

Secondly I have added a manager/leadership factor into the analysis which is dynamic and unique to each team.  This adjustment is intended to ‘smooth’ the team level aggregate scores that TPOEM calculates, where the model would not otherwise capture a persistent difference between a team’s results and their underying scores. This offsets (albeit not completely) the difference between the model’s league table compared to the actual league table. Why does that happen? Well, the basic underlying reason is the same as why a shots on goal league table does not reflect the real league table. I attribute this to a kind of quality factor that I am not picking up in the statistics I use: quality in terms of shooting can relate to the position on the pitch of a shot, whether defenders pressured the attacker and how much of a contribution the assist added to a goal scored. This quality factor will also incorporate a team’s record at home or away. For reference, the model currently seems to think that Stoke and Norwich are outperforming particularly well whilst Wigan, Southampton and QPR are all doing worse in the league than TPOEM suggests they should be doing. That might be due to luck, team playing style, management, player leadership, quality or all of the above. The model should now be slightly better at accounting for that.

Predicting part 2

So the first week of predicting using TPOEM brought me a net proft, although my biggest win was West Ham away win vs Stoke – and I’ve already explained that the model was distinctly anti-Stoke before the most recent update!

Again, as ever, I am seeking value so even if TPOEM suggests a probability of an event win/draw/loss of about 40%, if the bookmakers quote odds of 35% then I consider it an attractive bet. As it stands I haven’t been that selective about what I bet on: in fact so far I’ve been betting on every match that I ran the model for even though in many cases the model didn’t really suggest any particular value vs bookies.

The result this week, from 5 games, was another net profit, this time +26% return (it was +56% last time). But that came from 2 wins, 1 void, 2 lost bets, so in a sense the net result was neutral.  I profited overall because I weighted my bets towards the most attractive in terms of value – the biggest win being a draw-no-bet backing Everton at home to Man City. The model really liked Everton’s chances mostly because Kompany, Aguero and Yaya Touré were all missing for Man City.

I also backed draw-no-bets for Liverpool, Villa and Stoke: lost, won, void respectively. And lastly I went with a draw for Swansea-Arsenal (lost) but in retrospect I shouldn’t have bothered with that bet because the model gave no conclusive direction for the game and the odds weren’t good either.

As I reformat the model’s data and find a better way of communicating its predictions/results I will publish more information on the blog as I recognise I have kept most of the details pretty close to home so far. When I’m at my desk for the 3pm kick-offs I will also tweet about the model’s predictions so if you’re interested look out for that but if you bet then you are doing so at your own risk!!!

Premier League 2011-12: Player Impacts – discussion

In previous posts I have tested different ways of rating players using Opta data to mark out key fields for each major position which correlate positively to points.  The summary of these reviews can be read here.

What troubled me about some of the findings in this process was the underperformance of some high-profile players whose strengths were clearly not rewarded by the analysis. For example, Ashley Cole, Theo Walcott, and even Fabricio Coloccini – who actually made the PFA Team of the Year last season. Although I’m pretty keen to separate subjective opinion from raw data analysis, in particular the presence of Coloccini in the PFA Team of the Year – voted for by fellow professionals – cannot be disregarded lightly. Not to mention his superb performance at the weekend!

So in this series of posts I have published another ‘view’ of footballers – this time looking at team performance in the league with and without a particular player in the starting line-up. This can be used as a simple indicator regarding which players’ presence helps/detracts from their team. I used Tableau Public for the first time for this, and had some teething issues attaching my graphs/tables, so they are shown in separate posts below.


I calculated the average points gained, team goals scored and team goals conceded for every team and player and compared this to the team averages without that player in the starting line-up. Of course, those who started every game don’t have a ‘without’ average so I removed players who started every game. In addition, I took out players who started fewer than 4 games, and players who started more than 34 games. I did this on a whim after I saw that Robin van Persie had a negative impact to Arsenal’s points average – this happened because he started 37 games for Arsenal last season, and in the 1 game he didn’t start Arsenal won against Stoke. This annoyingly made Arsenal’s points average without RVP as 3pts per game, which is a bit ridiculous when he came off the bench and scored 2 in that game anyway! Players with 1 start had a similar problem, as the result of that game determined their impact. That example serves a purpose in explaining the limitations of a data table like the one below, even though the bias is reduced by increasing the min/max number of starts to 4 and 34. Of course if a player started in 34 games but the 4 he missed were away visits to Man City, Man Utd, Arsenal and Spurs then again his points average is more likely than not to be a little too high.

All the impacts below need to be taken with a pinch of salt but information is power, and I think this review is complementary to my previous player analyses and will help to give a better profile of players and their contribution to team performance. Incidentally, in this review Coloccini didn’t qualify because he started 35 games last season.

Hopefully, the tables/graphs are self-explanatory, but here are some highlights:

  • Adebayor for Spurs had the biggest positive effect on points for any team, followed by Arteta for Arsenal
  • Theo Walcott and Ashley Cole both had a strong positive effect for Arsenal and Chelsea respectively despite the poor stats analysis rating in previous posts
  • Notable ‘unlucky mascots’ for their teams were Berbatov for Man U and Ramsey and Arshavin for Arsenal
  • Swansea had a comparatively short range of differences between their players, which shows not only that they were able to field a remarkably consistent team for much of the season, but also perhaps indicates that no matter who was in the starting line-up, the player positions and tactics were relatively easy to substitute

Premier League 2011-12: Position Analysis ST

Last, but by no means least, is my position review for strikers in last season’s premier league. Robin van Persie, rightfully acclaimed for his performances last season (in which he appeared in all 38 of Arsenal’s games, starting 37 of them – and even scored 2 from his solitary subsitute appearance against Stoke) bagged 30 goals in total. But he still didn’t quite manage to top this rating.

Before I discuss the results, I ought to discuss some formalities about the rating I have used.  As with most of my previous posts, I reviewed the statistics from players who started in the position of striker: that is, the central player in a 3-man forward line, both players in a ‘flat’ forward 2 or the lone player up front. By looking at player starting statistics only, I am perhaps unfairly judging players who made a habit of having an impact from the bench – in addition, as you will also see later, my goals scored below for RVP is ‘only’ 28 because of this filter.

I then shortlisted the strikers who played greater than 1000mins from the start (34 in total), and added Agüero to make a 35th because according to Opta he mostly played behind the central striker last season and so would not have otherwise qualified. Notable absentees from the list include Defoe, Balotelli and Jelavic – all of whom played over 800mins in games they started but still not enough to make the cut.

I looked at Opta key statistics and reviewed the correlations between these fields and Wins, Draws, Losses – purely for strikers. For the strongest correlating fields I calculated ratios to try and remove some bias towards playing time and team biases eg. the players who played for better teams generally had more shots on target so to dilute this bias I created a shooting accuracy ratio to judge shots on target vs shots off target.

The table of statistics above is ordered by playing time from Papiss Cissé (1037mins) to RVP (3311mins). It is dominated by shooting and goalscoring statistics, with additional credit for chance creation, passing accuracy, dribble success and recoveries. I toyed with the idea of including offside frequency, because it IS quite significant in its relation to wins but I still couldn’t bring myself to add it into my rating. It does however show which players are so keen to break through the last line of defence they fall foul of being offside very often: the top 3 ‘offenders’ were Hernandez, Best and Bent. The players least likely to stray offside were Rooney, Doyle and Torres.

Papiss Demba Cissé was the standout candidate for killer instinct, leading the way in goals per shots on target (0.63), 2nd behind Bent in general shooting accuracy and 4th for shooting accuracy inside the box.

Steve Morison, Yakubu and Rooney scored well in headed goals and accuracy, whilst the best creators of chances were van Persie, Suarez and Zamora.

The best dribble success ratios were held by Klasnic, Carroll and van Persie – whilst Helguson kept it simple all season with 0 dribbles attempted! (I gave him the average success ratio so as not to unfairly disadvantage him).

Terrier-like high recovery rates were found with Welbeck, Rooney and Ngog.

I weighted all of these factors based broadly on contribution to WDL in order to calculate the following final ranking:

*Offsides, on the far right, have not been counted in the total score.

Rooney just about steals the top spot ahead of Agüero, a good 7pts ahead of RVP in a re-jigging of the top goal-scoring charts for the year. Rooney’s statistics are basically a lesson in how to be an excellent all-rounder, and he would be almost 20pts ahead of the competition if it weren’t for the inclusion of the rather dubious ‘touches inside the box’ statistic which disadvantages deeper-lying forwards.

Surprises in the top 10 include Zamora, Holt, Klasnic and Best. Best in particular was probably 3rd or 4th choice striker at Newcastle last season but was never given the faith that his statistics seem to justify.

Andy Carroll and Fernando Torres, part of an £85m transfer merry-go-round in January 2010, are 23rd and 24th respectively and underperformed their collegues Suarez and Drogba.

The only positive in Louis Saha’s stats (35th, last in the list) was his decent passing accuracy. Niklas Bendtner, now at Juventus, who would surely be higher in my rating if I included an ego statistic, only finished 29th.

Premier League 2011-12: Position Analysis CM

Reviewing Opta statistics for goalkeepers, central defenders and full backs was easier in terms of how to classify each position because as I explained in my first post on formations, almost every team lined up with a back four last season (on only 21 occasions out of 760 did a team not start with a back four).

Now that we are looking at midfielders, although four ‘flat’ midfielders was also the most common line-up, in general there was a lot more variety. A defensive midfielder should not really be compared with a winger, so I have had to be more careful when it comes to classifying the position that a midfielder plays depending on his position in the team’s formation. As we will see later, by pigeon-holing players in this way the review may be flawed for some players who qualify for the CM category but in reality are more defensive than average.

For transparency, I have listed my classifications below:

As we can see from the above list of CMs, which is fragmented by the formations, it was more of a challenge to remove the players who did not play in this position from the initial correlation review – due to the less consistent position ids. I also removed players with fewer than 1000mins in the position of CM in order to maintain consistency with my previous reviews .

I found that offensive statistics for CMs dominate defensive statistics. For example, goals, assists, through balls and big chances were all slightly more important to picking up points than the most significant defensive fields such as headed clearances and ground duels. This seems to suggest that the all-action central midfielder is more important in attack than defence. As a result my rating has 5 offensive statistics which rate the 48 midfielders with points from 1-6 (worst to best), 1 neutral field (pass success rate, middle third), and 3 key defensive statistics each with a maximum of 6pts on offer for the best 8 midfielders in each category. Errors leading to goals and red cards were so infrequent that it seemed unfair to apportion too much significance to them to I simply deducted a point per occurance for each player.

The final table is below. Note the absence of any Swansea or Arsenal players from the list: this is because neither team set up with a formation including a CM as per my definition – both Rodgers and Wenger prefer to employ the 4-2-3-1 formation.

The winner is another Manchester City player: Yaya Touré. Touré was without doubt instrumental to City’s success last season and as we can see from the rating he was excellent in all the categories that matter in my chart. I am starting to wonder if there is an inherent bias in the data MCFC have made available to the public considering how many categories City are dominating! The reality however is that the best players in the best teams should generally stand out in this type of analysis. Arguably Touré’s best statistic is middle third passing success, in which he was 2nd best overall. On average he completed 17.1 passes in the middle third before one was misplaced which compares to an overall average of 7.5 (successful passes:unsuccessful passes). However the winner in that category was the influential Paul Scholes, who on average completed a colossal 23.3 passes before one was misplaced!

Notable players in the top 10 include Stiliyan Petrov (whose season was sadly cut short due to his diagnosis with leukaemia) and Wes Hoolahan, both of whom were key midfielders who played for bottom half teams. Petrov was strong in most fields, but had a particularly low rate of being dispossessed, whilst Hoolahan was very strong in offensive categories – with similar attacking qualities to Frank Lampard.

Steven N’Zonzi, who was the player from a relegated club with the highest position, rated =14th in the list despite being considered by many to be a defensive midfielder – a significant result considering the general lack of more defensive-oriented CMs in the top half of the list.

For the second post running a Newcastle player finished last – surely not?! Rated by many to be one of Newcastle’s outstanding players, Cheick Tioté’s qualities are clearly not suited to this system. As we might expect, Tioté’s offensive statistics were poor: he didn’t score, he made one assist and played one through ball all season. What is perhaps more surprising is his below average middle third passing success ratio (6.5), the fact that he lost more ground duels than he won (169 to 122) and he was dispossessed more times than any other CM (69 in total).

And another surprise in the form of Luka Modric, who might be assumed to perform well in this model but only rated 33rd. These players deserve better, and so at a later date I will look to ascertain the qualities that set them apart from the crowd.

Premier League 2011-12: Game changers pt2

This post reviews sendings off in the 2011-12 season: which teams lost or profited most?

To breakdown the MCFC/Opta data this time I filtered out any sendings off which occurred with less than 10 minutes of the game to go. Why? We can learn more from games where the team with a numerical advantage has longer on the field, and therefore a better opportunity to capitalise on their opponent’s weakness. I am most interested to see if any particular managers/teams seem to be better or worse at coping when a game becomes 11 vs 10 or even 11 vs 9 as happened in the QPR v Chelsea game last season at Loftus Road.

Last season a player was sent off with more than 10mins remaining a total of 51 times. QPR were the worst culprits, with their players sent off a total of 7 times before the 81st minute.  Interestingly, all of these red cards occurred in 2012, Mark Hughes being in charge for 6 out of the 7 games. If Mark Hughes specifically promoted the tactic of being more aggressive in the tackle, it was certainly a costly tactic as QPR lost all but 1 game in which they had a player sent off* (*with more than 10mins to go). In fact, in the only game they won from this situation they were already in front against Spurs and the red card for Adel Taarabt occurred in the 78th minute – only just meeting the constraints of my rule. Delving further into the stats on points gained/lost/unchanged, based on the score at the time of the sending off to the final score, QPR lost 2 games in which they were in a winning position until a player was sent off (at home to Norwich and Wolves) and 2 further games lost from a position of drawing a game (Man Utd and Man City – the famous conclusion to the season). Hence we can say that QPR had a -8pts swing, no doubt impacted by the red cards in those games. As it was, QPR only just survived relegation, yet with 8 more points to their total they would have been equal with Sunderland and Stoke in 13th/14th place.

Compare QPR’s stats to Spurs, who were much more disciplined last season with only 1 red card before the 81st minute: Danny Rose vs Aston Villa. On this occasion Spurs were actually able to turn their losing position to a draw, thereby gaining a point. The only other team to seemingly benefit from having a player sent off was Blackburn (vs Fulham, home). Here, Steve Kean’s team actually won having been level with Fulham when Yakubu was sent off.

55% of the time the red card did not change the result, in 41% of the matches the team with the player sent off lost points and 4% of the time the team actually gained (the 2 aforementioned).

Perhaps Bolton can count themselves unlucky – they had players sent off* in 5 games but no team had a player sent off* against them. On the other hand, Fulham’s players did not get sent off once with more than 10mins to play, but they benefited on 4 occasions, taking 9pts but arguably dropping 1pt against Steve Kean’s Blackburn.

At the top of the table, Arsenal were involved in 4 games in which they had a player sent off and 4 games in which the opposition had a man sent off*. In the games they had a player sent off they took only 1pt, but with the opposition down to 10 men they managed 3 wins and a draw. Manchester United’s opposition had players sent off 6 times; in these games although the result was often in their favour already, Man Utd took 16pts (5 wins, 1 draw). Man City and Chelsea both took maximum points in games where opposition players were sent off (6 and 9pts respectively), and City had an excellent record in games they had players sent off: 6pts from 3 games.

Any attempt to make a serious claim on which team or manager is best or worst at dealing with situations of 10 vs 11 players is flawed due to the small sample size.  But why not throw some names into the hat anyway?! Adding together the points swing after a player was sent off for either side shows that Chelsea come out on top (+6pts) and QPR bottom (-8pts). These teams both had manager changes during the season but for Chelsea Andre Vilas-Boas was the dominant force in their positive statistic, whilst Mark Hughes was in charge for most of the games that contributed to QPR’s poor points swing result. A special mention is reserved for Martin Jol’s disciplined Fulham side, as discussed earlier, who acheived a points swing of +4 from 4 games.

Team # players sent off* (a) Points (a) Points swing (a) # opposition players sent off* (b) Points (b) Points swing (b) Points swing (a) + (b)
Chelsea 3** 3 0 3 9 6 6
Fulham 0 0 0 4 9 4 4
Liverpool 4 3 -1 3 6 4 3
West Bromwich Albion 1 0 0 1 3 3 3
Manchester United 1 0 0 6 16 2 2
Newcastle United 2 1 0 3 5 2 2
Swansea City 2 3 0 2 3 2 2
Everton 1 0 -1 1 3 2 1
Manchester City 3 6 -1 2 6 2 1
Sunderland 1 1 -1 2 6 2 1
Tottenham Hotspur 1 1 1 5 12 0 1
Wolverhampton Wanderers 4 1 -2 2 3 3 1
Blackburn Rovers 4 4 0 2 3 0 0
Norwich City 2 1 -3 3 9 3 0
Stoke City 2 0 -2 2 2 2 0
Wigan Athletic 3 1 -2 2 1 1 -1
Arsenal 4 1 -4 4 10 2 -2
Bolton Wanderers 5 0 -2 0 0 0 -2
Aston Villa 1 0 -3 2 4 0 -3
Queens Park Rangers 7 3 -8 2** 3 0 -8
*with more than 10mins of normal time remaining
**Chelsea had 2 players sent off against QPR at Loftus Rd

*Sent off refers to my definition of sendings off with more than 10mins of normal time remaining.

Premier League 2011-12: Game changers pt1

Having used the MCFC Analytics / Opta data for my previous post on team formations, I am continuing to collate and group it into team data sets to try and see if I can find any interesting trends or analyses that may help to highlight manager/team strengths and weaknesses.

This post reviews substitutions: the frequency they are used, average playing time and goals scored.

Team # substitutions Subs per game Sub mins Ave mins per sub Sub goals Goals per sub Winning goals by sub 0 subs?
Arsenal 109 2.87 1959 17.97 6 0.06 3
Aston Villa 94 2.47 1815 19.31 2 0.02 1
Blackburn Rovers 87 2.29 2151 24.72 3 0.03 0 1 vs MU (H)
Bolton Wanderers 104 2.74 2056 19.77 2 0.02 2
Chelsea 105 2.76 2299 21.90 7 0.07 2
Everton 111 2.92 2081 18.75 10 0.09 3
Fulham 85 2.24 1521 17.89 1 0.01 0 2 vs WBA, Bolton (H)
Liverpool 91 2.39 1720 18.90 4 0.04 1
Manchester City 110 2.89 1952 17.75 15 0.14 2
Manchester United 100 2.63 2194 21.94 7 0.07 0 1 vs Liverpool (H)
Newcastle United 112 2.95 2158 19.27 6 0.05 2
Norwich City 111 2.92 2350 21.17 9 0.08 2
Queens Park Rangers 92 2.42 1867 20.29 7 0.08 2 1 vs Swans (A)
Stoke City 107 2.82 2372 22.17 4 0.04 1
Sunderland 92 2.42 1824 19.83 4 0.04 1
Swansea City 94 2.47 1859 19.78 3 0.03 1
Tottenham Hotspur 94 2.47 1719 18.29 5 0.05 1
West Bromwich Albion 101 2.66 1733 17.16 6 0.06 0
Wigan Athletic 102 2.68 2080 20.39 5 0.05 1
Wolverhampton Wanderers 108 2.84 2278 21.09 3 0.03 1
Total 2009 2.64 39988 19.90 109 0.05 26 5

Of course, in a premier league game a manager can make 3 substitutions. Last season, the average subs used per game by each team was 2.64 i.e. on most occasions a manager uses his full allocation of substitiutes. The managers most likely to use all 3 subs were Alan Pardew (Newcastle, 2.95 subs per game), Paul Lambert and David Moyes (Norwich and Everton, both 2.92). Substitutions are clearly of partcular importance to these managers, whether for tactical reasons, to help combat fatigue, squad rotate or simply run down the clock towards the end of a game. In Alan Pardew’s case I suspect the latter tactic was possibly used more than most, since the average number of minutes a Newcastle substitute played was 19.3mins, below the league average of 19.9mins.

The managers least likely to use all 3 subs in a game were Martin Jol, Steve Kean and Kenny Dalglish (2.24 subs per game, 2.29 and 2.39 respectively). Martin Jol used his bench warmers less than average, and unfortunately for them they were given a paltry 17.9mins playing time per game. Although Steve Kean didn’t use his subs as often as most managers, when he did he made sure they got a decent runaround: his substitutes had easily the longest average playing time in the league of 24.7mins.

Only on 5 occasions last season did a team not make any substitutions (Fulham: 2, Blackburn, Man Utd, QPR: 1).

Those statistics on substitutes will probably only interest the most serious football nerds (like me), so how about some information on managers who brought on super-subs? Well, this paragraph belongs to Roberto Mancini. An amazing 15 substitutes for Man City scored after being brought on, well above the average of 5.45 substitute goals. Dzeko’s contribution to the mayhem on the last day of the season was probably his most important goal from a total of 4 he scored coming off the bench. Perhaps you would think that since Man City won the league the statistic of 15 goals from substitutes isn’t that impressive, however for Ferguson and Wenger who both made an almost identical number of substitutions, their subs only scored 7 and 6 goals respectively. Certainly the attacking firepower Man City had on the bench last season was a significant weapon, perhaps if Jol had Aguero/Balotelli/Dzeko/Tevez on his bench throughout the season he would have used his substitutes more! Nevertheless the 8 goal differential between City and Utd from substitutes also tallies with City’s superior end of season goal difference of +8 that handed the title to the sky blues.

When it comes to goals per substitution, Mancini is easily top with 0.14 goals per sub, whilst David Moyes follows with 0.09. Martin Jol comes out worst with 0.01 goals per sub (no wonder he didn’t want to use his subs too often).

Winning goals scored by substitutes are harder to come by (and therefore more difficult to apportion any meaningful insight). However, Mancini doesn’t fare quite so well here, with only 2 out of the 15 subs goals scored classified by Opta as a winning goal (Balotelli’s efforts at home against Everton and Spurs). Wenger and Moyes topped the charts with 3 winning goals per sub apiece – very memorably for Arsenal fans this included Thierry Henry’s late winner at Sunderland in February. Four managers failed to bring on a sub who scored the winner: Ferguson, Hodgson, Jol (surprise, surprise) and Kean.

Premier League Formations 2011-12: Two banks of four

First of all many thanks to Gavin Fleig and the teams at MCFC Analytics and Opta for making so much quality data available to the public.  I received the ‘basic’ form of the data at the start of the week and intend to write a few posts on my findings over the coming weeks, covering a range of areas.

My initial analysis using the data began by looking at team formations, where I found that the overwhelming favourite amongst EPL managers remains the 4-4-2 (accounting for 33% of team line-ups) followed by 4-4-1-1 (21.7%) and the 4-2-3-1 favoured by Arsene Wenger and Brendan Rodgers (16.4%). It certainly seems as though most EPL managers believe that two banks of four provide a superior foundation for success in the premier league, considering that 54.7% of the time last season a premier league team lined up with two banks of four. Only two teams did not use either 4-4-2 or 4-4-1-1 throughout the year: Arsenal and Chelsea.  Special mentions go to Sir Alex Ferguson, Martin Jol and Tony Pulis for fielding two banks of four in over 85% of their games. Indeed, Tony Pulis’ Stoke City used the pure 4-4-2 in 36 out of 38 games, experimenting wildly(!) with a 4-5-1 formation away at Arsenal and Liverpool (Lost 1-3, Drew 0-0 respectively).

Nevertheless, in terms of consistency in formation, Tony Pulis takes second place. Arsene Wenger appears to have such an unshaking belief in the 4-2-3-1 formation that he used it in every game, and undoubtedly this style of formation is important to Arsenal’s famed passing game and control of possession. Brendan Rodgers’s Swansea City, who also dominated possession statistics last season, lined up with the same formation 87% of the time.

Only 5 teams experimented with 3 or 5 players in defence: Blackburn Rovers, Liverpool, Norwich City, Swansea City and Wigan Athletic. Of these teams, only Wigan used any system with 3 or 5 at the back more than once, for the rest it was an experiment to be forgotten as none managed to win the games in which they used this new formation. Roberto Martinez’s decision to try 5-4-1 a total of 7 times from February 2011, despite using it earlier in the season at Old Trafford and losing 5-0, helped Wigan to gain revenge against Manchester United and also beat Arsenal and Newcastle. In a remarkable turnaround to their season, Martinez’s Wigan used 7 different formations in their last 14 games earning 27 points in the process – form which over the course of a season would yield somewhere in the region of 72pts, enough to finish 3rd in the league table.

It may be interesting as well to look at the average points per game for each formation, I have copied this below.  Unsurprisingly, the formations preferred by certain teams distort the analysis here. For example, the 4-2-3-1 preferred by Wenger, or the 4-4-1-1 preferred by Roberto Mancini lifts the average points per game for that formation. It would be folly to say without further analysis that had Wolves used the 4-2-3-1 throughout the season they would have had more success. It is also difficult to conclude anything from the the results from those formations used less than say 30 times.

Total times used Ave. Pts per game
4-1-2-1-2 6 1.67
4-2-3-1 125 1.50
4-4-1-1 165 1.47
5-4-1 9 1.44
4-4-2 251 1.40
3-4-2-1 3 1.33
4-3-3 72 1.22
4-5-1 94 1.19
4-1-4-1 26 1.15
3-4-3 3 1.00
5-3-2 6 0.83

My last note is an ode to Terry Venables. I am afraid El Tel that your beloved Christmas Tree formation, 4-3-2-1, was not used even once last season.