Wednesday 1 October 2014

Adventures in football data 1: Fabregas & assists

Cesc Fabregas was something of a prodigy; from the age of 17 he played regularly for Arsenal & over a period of 7 seasons became a key component in their midfield. During this time he played in 212 league matches and scored 35 goals.  These facts are not in dispute; check any source you can find and they will give you these numbers.  And that makes sense: participation in a game and the scoring of goals are the two most rudimentary statistics recorded in football.  From school level upwards, throughout many an amateur team, all the way to the top tier of the professional game, these numbers are recorded.  But what about secondary numbers? For example, given his skill as an excellent passer, how many goals did Cesc Fabregas create during his Arsenal tenure?

How many league assists did Cesc Fabregas create at Arsenal?

Source: Fabregas Assists 2004-2011
A 76
B 77
C 72
D74
E 68*
F 71
*First two seasons not given, so estimated from most common total sources.

Okay, we have a difference of opinion! I have no idea who is transcribing the information for each of these online resources & it is not my intention to criticise anyone for providing that service, merely to highlight discrepancies in data collection, so for that reason i've anonymised the sources. Moreover, the differences are quite large.  So is it a matter of a specific outlier? Maybe the data becomes more unreliable as you go back in time?  Theoretically, given the increased interest in the use of football statistics you might presume the most recent numbers would be more consistent:

Fabregas Assists 2004-2005 2005-2006 2006-2007 2007-2008 2008-2009 2009-2010 2010-2011
A 4 5 12 19 11 15 10
B 4 5 13 20 11 13 11
C 2 5 6 20 10 16 13
D 4 5 13 20 8 13 11
E NA NA 11 16 8 13 11
F 3 4 11 18 8 14 13

Hmm... not the case.

Are there any two sources the same?

No.  Sources B & D concur on all years except for 2008/9, which is encouraging; or at least it is until you realise that source B is Wikipedia, so basically regurgitated information derived from elsewhere.

The most baffling season is 2006/7. One source gives a mere 6 compared to 11-13 from the other sources.  I decided to try and ascertain the truth.  Having squinted pointlessly at blurry Youtube videos, the method I resorted to was to look at the post match write ups, which are conveniently available on the BBC website.  They do a pretty decent job of describing the action during a match and the events leading up to a goal.  Here's what I came up with:

Opponent Result Scorer Fabregas Contribution Assist?
@Man Utd 0-1 Adebayor 'forced error, threaded pass' Y
v Watford 3-0 OG 'free kick flicked by Adebayor forcing OG' ?
@Reading 0-4 Henry 'precise pullback' Y
“” “” Henry 'brought down for penalty' ?
v Liverpool 3-0 Flamini 'crossed low to near post' Y
@Wigan 0-1 Adebayor 'through ball' Y
v Blackburn 6-2 Flamini 'beat Savage by corner flag, keeper parried' ?
@Blackburn 2-0 Henry 'return pass' Y
@Tottenham 2-2 Toure 'free kick chipped to far post' Y
“” “” Adebayor 'free kick chipped to far post' Y
v Fulham 3-1 Adebayor 'through ball' Y

Before you read on, I implore you to enjoy this clip from the Blackburn game:


So the upshot is, I count eight clear assists and three others that could be counted depending on how you define 'assist'.  FIFA attempted to define what they considered an assist after producing a Technical Report for the 1986 World Cup & revised one element after the 1990 tournament.  They say this:
  1. An assist was awarded to the player who had given the last pass to the goalscorer.
  2. In addition, also the last but two holder of the ball could get an assist if his action had decisive importance for the goal .
  3. After goals from rebounds those players were awarded an assist who had shot on target.
  4. Where goals resulting from penalties are concerned, the player who is fouled in the area receives an assist point (unless, that is, the player who is fouled subsequently executes the penalty himself)
Given these rules, we land on 11 for Fabregas with an addition on one each from the 2nd, 3rd and 4th categories. Two of our sources concur.  Of course, i'm equally liable as any source of making an error in data collection, particularly as i'm here 7 or 8 years after the fact.

So what is the point?

Data collection within football is not new, but the extent to which it is being collected has accelerated during the Premier League era, especially in recent years as public resources have cropped up.  What concerns me is that beyond simple measures such as goals or appearances, there are inherent flaws in the collection of any other data.  Specifically i've looked at assists here as beyond goals and appearances, they are the next level of detail in statistically describing a game; they are necessarily rarer than goals so shouldn't take much counting, yet in the example shown there is great variance within a small sample.
 
The example I have used is historic, I didn't seek it out, I just found it in the course of research & was actually surprised at the variance I found when trying to look for a definitive number. I do appreciate that both public and private data recording is ever-improving & these examples are now historic.  (Although it seems crazy to think of 2007 as a historical backwater, but comparatively, it was).  The hope is that eventually, football will realise the value of standardising some of it's metrics across the resources that are available and when an assist is declared an assist it's as fixed as any goal or appearance.

Thanks for reading.

----------------------------------------------

PS:

Not all assists are born equal; this got charged to Karim Benzema the other day & is probably not one for his highlight reel:


PPS: I was reminded to finish this article by Ted Knutson's irritation that Fabregas' wiki page had no assists credited to it, whereas Thierry Henry's did.  Well, a couple of weeks ago they were listed & for some unknown reason, now they are not. Maybe the most recent wiki editor found it as difficult as I did to create a definitive total?

PPPS: I appreciate that plenty of what i've written here will be a case of 'WE KNOW' to some people and that  greater detail could have been incorporated, but i've tried to keep things generally light and uncontroversial with the aim of making a point rather than driving a wedge.

No comments:

Post a Comment