You Can Prove Anything with Proof

I think my burgeoning fascination with statistics is getting me a reputation. Ah well. I like numbers; they offer me succour in a world I barely understand. But I’m not daft enough to think that numbers tell you everything. Data is data. It’s only when human beings interpret them that they become ‘facts’. Now there’s a loaded term.

You can of course prove anything with facts. You can also wildly misinterpret statistics. One of my favourite stories about statistics (yes, welcome to the world of Jay) is the one where a couple of academics espoused the theory, based on a statistical analysis of deaths in Southern California. As this BBC piece explains, they then contacted each person’s family to enquire if the individual had been left or right-handed. Their resultant data showed them that the mean age of deaths among right-handed people was higher (around 9 years) than that of left-handers, and from that they constructed the theory that left-handers (such as your author) are more likely to die younger, presumably in some hilarious slapstick accident using machinery designed for right-handers.

The trouble was, they’d misinterpreted the data. For most of human history, left-handedness has been considered unnatural, a literal sign of the devil. The word ‘guache’ in French also means awkward. It in turn replaced the original French word for left, ‘senestre’. I believe most cultures and religions around the world have displayed varying degrees of intolerance towards left-handedness. This, combined with adapting to a world where virtually everything is geared (again, pretty much literally – righty tighty, anyone?) towards right-handedness would have meant that in the early part of the 20th century, many lefties would have had little choice but to assimilate as righties. Conversion as a result of physical abuse has been reported, and even as late as the 1980s, educational establishments in the first world weren’t willing or able to support left-handers. Thus a scientific survey became skewed.

I was reminded of this story this morning, when reading about the ongoing fallout from the West Ham United versus Manchester United match on Sunday. After the game the West Ham manager accused the opposition of being ‘long-ball merchants’, a phrase considered within the footballing world to be one of the most grievous insults. Like describing a film as sub-Michael Bay. His counterpart, Louis van Gaal, an unorthodox Dutchman took umbrage to this, and appeared at a press conference the following day with a pamphlet of diagrams and statistics that proved that it was West Ham, and not Manchester of the Uniteds that were the long-ball merchants.

I’ve been following this with interest, because the sort of statistics van Gaal has produced are what I’d like to tackle next. The trouble is that the major football statistic companies just aren’t interested in the game north of the border, so there isn’t much data to get hold of. Shot of compiling my own numbers, I’m limited to using what the BBC provide, such as possession, fouls, shots on target. As the BBC piece above, and this Guardian article from this morning (hat-tip to Doug) illustrate, one man’s aimless punt is another man’s tactically astute long pass.

I’ve heard many Rangers fans criticise the club’s play over the last few years as consisting of little more than long balls. It’s not my personal experience of most of our play, and while I suspect this is just another example of subjective interpolation, it’s difficult to prove either way. I don’t have any pass completion statistics. The only empirical evidence I have to go on is the possession figures. Jonathan Wilson notes that Manchester United’s possession statistics are the second highest in the English Premier League. Both the Guardian and the Independent have highlighted Crystal Palace and Burnley as being the two teams that play long passes most often. As evidenced in this Squawka chart, possession, pass completion and average pass appear to be interlinked. This makes sense; the more often you kick the ball aimlessly down the pitch, the less likely you are to retain possession of the ball.

Rangers’ average possession statistics (and chance creation stats) in league games for the last season and a half* are as follows;

Season Games Played Mean Possession Total Attempts Attempts Per Game
2013-14 29 58.21 456 15.72
2014-15 20 53.40 273 13.65

*Unfortunately, the BBC only introduced possession stats a few games into the 2013-14 season.

Using the Squawka data as a benchmark, this doesn’t suggest that Rangers (current awful form notwithstanding) have been quite the negative, defensively set-up outfit that they’ve been portrayed as the last few seasons. They do generally experience long spells of possession (only in 7 games out of 49 have they had less than 51%.) 14.87 attempts on goal per game over the last 49 doesn’t fit the profile of an overly-defensive team either.

Do we play long balls though? Yes, we do. We have for the last good few seasons, as part of a range of passing types. Looking through my old posts, we appeared to use it fairly regularly in the first part of the 2012-13 season. I suspect that we do go long when we do to attempt to bypass a team that are defending deep and closing out space in our attacking third of the pitch. To quote Jonathan Wilson again, we also use it as a plan B. The archive of our goals from the last few seasons on the Rangers TV website illustrate that when we do score, it tends to be as a result of quite attractive, short passing football. But this sport is a subjective one, and this blog is my interpretation of the data, and even if I did have full passing stats, I’m not sure I’d change anyone’s mind.

Advertisements

Nostalgia ain’t what it used to be

When I first started studying photography, around 18 years ago, I came across a reader’s letter in a magazine regarding the ‘right’ way to shoot moving water. Long exposures, where the various flows are recorded over a second or longer, were derided as they apparently didn’t represent water the way humans see it. I immediately disagreed with this for two reasons; firstly, humans simply don’t see the world the way a camera does, regardless of what exposure we use. We can focus on the spray, or we can let it flow. Have you ever looked out of the window of a train travelling a high speed? If you stare at a fixed point, the scenery close to the train zips by in a blur lines. Allowing your eyes to pan on one particular detail, and you’ll be able to resolve more details.

Secondly, I was already becoming interested in using photography to see things that the naked eye can’t. Like erosion, tides, decay, growth. Things that happen so slowly the human brain can’t quite process, but film and moving pictures can. Nearly 20 years on, I find myself applying the same logic to my new hobby of statistics. Analysing data helps to allow us to identify patterns and trends we couldn’t conceive by ourselves. We should also remember that human beings are emotional souls, and at times, that clouds our judgement. Our memories are not as good as we think they are. We find it much easier to remember good times and bad times than we do when things were just okay.

That’s mainly why I was drawn to the use of statistics in football. For the last five years or so, I’ve been part of a burgeoning online community of Rangers fans. Having never had too many supporters of the light blues in my social circle, I was somewhat surprised at just how much we moan about things. About everything. You name it, I can pretty much guarantee I’ve seen a tweet or a Facebook comment or a forum post criticising the club about it. It’s like rule 32. Google it. No wait, don’t google it.

Some of these criticisms I thought were incredibly harsh. Others I’ve found myself wondering how much basis they have in fact. The amount of game time we give to young players is one particular recurring theme. At what point during a game did Ally McCoist make substitutions. All stuff that is not particularly important in the grand scheme of things, but which Rangers fans will argue the toss over online.

The main reason for these statistic-oriented blogs is because I’m a Rangers fan, and it pains me to see the club under what I consider to be unfair criticism from its own fans because of a perception that barely turns out to be true. I don’t think such non-constructive criticism is helpful either, as it forms a standard the club can’t possibly live up to. And when the fail to meet these standards, there’s more criticism. It’s a vicious circle, a feedback loop of recrimination.

In this blog, I’m going to look at the respective youth systems of Rangers and Dundee United. Since Charlie Telfer left the former to join the latter in the summer, I’ve developed the impression that United are considered to have a far more effective youth set up than Rangers, that they brought through more players, of a better quality. 1 2

http://www.bbc.co.uk/sport/0/football/28863676

http://www.dailymail.co.uk/sport/football/article-2518538/Ryan-Gauld-Dundee-Uniteds-Baby-Messi-wowed-Brechin-Under-Nines.html

Such a reputation probably has its roots in the Jim McLean era; up until the 80s, it was more common for football clubs to draw all their talent from youth teams in the local area. Slowly though, as football continued its metamorphosis into the global industry it is now, clubs started to buy in more and more players. The finished article, as it were. The Manchester United European Cup winning teams of 1968, 1999 and 2008 featured 75%, 33%, and 28% home grown players respectively. A similar trend can be observed in the Rangers teams that won the 1973 European Cup Winners Cup, reached the group stages of the 1993 Champions League, and lost in the final of the 2008 Uefa Cup: 56%, 25%, and 6%.

At the same time, most Rangers fans would tell, I suspect, that their club has stifled their youth development programme. They’d also add that Walter Smith and Ally McCoist were the two managers least likely to give youth a chance. So, as a starting point, I decided to look at the youngsters that have come through the ranks at Rangers, and Dundee United, over a period of 23 seasons, starting with Smith’s first full season in charge in 1991-1992. Some basic data is included in the following table.

  Seasons Players League Games Games per player Av Age on Debut Int. Caps (total) Young debuts per season
Rangers

23

66

1836

28

18.73

277

2.87

Dundee Utd

23

51

2617

51

18.04

32

2.22

From the above, I would infer that although more young players go on to become established first team players for United, Rangers debut more young players, many of whom go on to play at an arguably higher level. It’s up to the beholder to decide whose system then is the more effective.

I’ve recently started exploring the usefulness of standard deviation. Rangers have debuted around 3 new young players a season, sometimes more, sometimes less. But I was curious to see during which seasons we played more, and fewer. The results are in the chart below.

To explain the chart, the blue bars represent how many more or fewer players than the average of 2.87 were debuted each season. The orange line displays how many combined games those debutants would go on to play for the club. At a glance, we can see that 96/97, 99/00, 01/02, 03/04, and virtually the entire McCoist reign of 2011 to 2014 were particularly fruitful. This didn’t surprise me; I’d argued in a previous blog that the production of young players is not an exact science; it goes through periods of boom and bust. Sometimes you may have four or five great young talents emerge in 1 season. Sometimes you may go 2 without. It’s probably worth looking at a similar chart for United.

United too have experienced peaks and troughs, even during Ian Cathro’s much-lauded spell as the head of their youth academy. Recently, they have debuted slightly more players, but each player has played fewer games, although it should be noted for both sides that players from more recent seasons won’t have had the chance to accrue as many appearances.

Finally, I wanted to break down Rangers’ youth debuts by manager. As I’ve mentioned before, Walter Smith and Ally McCoist have a reputation of being distrustful of youth, even if the data above doesn’t quite support that. I looked at the number of young players given their first start by each manager, and that’s in the data below. For a bit of context, I’ve also included each manager’s transfer spend, adjusted for inflation.

    Games Managed Youth Debuts Youth Debuts Per Game Games Played by Youth (Total) Spend per Game (inflation)

1

McCoist

127

19

6.68

331

£ 39,448.82

2

McLeish

172

15

11.47

537

£ 132,178.44

3

Smith (1)

276

17

16.24

495

£ 376,491.20

4

Advocaat

128

7

18.29

245

£ 959,844.99

5

Le Guen

23

1

23.00

3

£ 220,950.10

6

Smith (2)

167

7

23.86

228

£ 238,788.39

Somewhat surprisingly, McCoist, McLeish and Smith (during his first spell) were more likely to turn to youth players than Advocaat or Le Guen. Advocaat had far more money to spend, an eye-watering £960,000 for each game he managed the club, and McCoist had noticeably the least, although he was in charge during the Whyte and immediate post-administration years.

The point of this blog, if there is one, is to suggest that perhaps we as Rangers supporters are unduly critical of our club. That we compare ourselves to others unfavourably, while measuring said others against lesser standards than we would judge ourselves. It’s my opinion that this causes a downward spiral of negativity that the support and the club struggles to extract itself from, and when this isn’t based on any factual evidence, this becomes double frustrating.