In his excellent book The Sports Gene, David Epstein wrote of a phenomenon in the physiology of elite sports people that is the result of a huge increase in the numbers of armchair fans in the last century. As there has been a transition from people playing sports to lying about and watching others play instead, the financial rewards of sporting success have, according to Epstein, “slanted toward the slim upper echelon of the performance pyramid.”
In order to optimize the pursuit of these monies, certain body proportions have become concentrated in specific sporting fields – a process Epstein has called ‘The Big Bang of Body Types.’ In 1925, for example, Epstein writes that “an average elite volleyball player and discus thrower were the same size, as were a world-class high jumper and shot putter.” Now, of course, shot putters are generally to be found very tall and very wide, high jumpers very tall and very thin, and marathon runners very short and very thin. In the modern era of high performance sport, body type matters.
According to 2009 research by Alan Nevill, Roger Holder and Adam Watts in the Journal of Sports Science, this ‘Big Bang’ can also be identified in football. Analysing thirty seasons worth of data from 1973-74 to 2003-04, the authors found “a significant increase in professional footballers’ height, body mass, and BMI” between the start and end of the period. In 2011 research published in the International Journal of Sport Physiology and Performance, Gal Ziv and Ronnie Lidor found that the average professional goalkeeper is both taller and heavier than the average professional defender, midfielder, or forward. So footballers are generally getting taller, and goalkeepers are the biggest of the lot. Nevill et al. went as far as to suggest that their findings should encourage football coaches and talent scouts to “pay attention to… body shape when selecting potential players for their squads.”
So that’s that then. Goalkeepers are the tallest players, so scouts should recruit tall players for the role. I find this a questionable conclusion for the researchers to have come to as it smacks of what Simon Kuper and Stefan Szymanski called, in their 2009 book Soccernomics, ‘sight-based prejudice.’ As has been made famous in the Moneyball book and film, the Oakland Athletics General Manager Billy Beane noticed that baseball scouts were typically suspicious of recruiting any player who wasn’t tall and handsome.
Kuper and Szymanski pointed to the anecdote of a Brazilian agent who had flogged Brazilian players to teams in the Faroe Islands and noted that “it is much easier selling… a crap Brazilian than a brilliant Mexican.” In another anecdote, one club was said to have reported that their scouts disproportionately recommended youth players with blond hair, a cognitive bias believed to be linked to their light hair colour having caught the eyes of the scouts and leading to them being recommended for reasons totally irrelevant to their footballing abilities. Kuper and Szymanski concluded that instead of recommending players simply on the basis of their abilities, “scouts look for players who look the part.”
The logic follows that footballers who, for whatever reason, don’t look the part will accordingly be undervalued in the transfer market with respect to their actual abilities. Signing a tall goalkeeper for the simple reason that everyone else has tall goalkeepers as Nevill et al. appear to recommend is exactly how sight-based prejudices begin. If a tall goalkeeper does perform better than a short goalkeeper, then a club should by all means sign them. A goalkeeper should never be signed, however, simply because they are tall.
One such goalkeeper around whom a debate of sight-based prejudice appears to have sprung up is Arsenal’s David Ospina. Signed from OGC Nice in July 2014 for an estimated £2.8m in the aftermath of the World Cup, Ospina has divided opinion amongst Arsenal fans. Standing at just 1.83m tall, the Colombian was, alongside Hull City’s Allan McGregor, the shortest goalkeeper in the Premier League in the 2014/15 season. Amongst goalkeepers to have played 16 matches or more (1440 minutes +) throughout the season, the Colombian conceded 9cm to the league average, and 19cm to Sunderland’s Costel Pantilimon, who, at 2.02m, was the league’s tallest.
Ospina himself has commented on the added pressure he feels as a result of his relatively diminutive stature, stating that “At first, you laugh… When I’m in the tunnel and I see the opposing team’s goalkeeper, I have to look up. In England, they’re all two heads taller than me!”
Since dislodging Wojciech Szczesny from the Arsenal starting XI midway through the 2014/15 season, the reception to Ospina among Arsenal fans has been far from entirely positive, with Tim Stillman at Arseblog.com having brought the Colombian’s aerial handling into question as early as February. Others have focussed squarely on Ospina’s height as being an issue, with a cursory glance at tweets from Arsenal fans including the suggestions that “Ospina’s game is technically sound [but] he’s just too short,” that “Ospina is short and at his peak,” and that “everyone’s better than Ospina, he’s too short to be a goalie.”
Although these tweets that I mention are all offerings from the past week, this viewpoint has been salient among Arsenal fans for a while. Indeed, Rob Bateman (@Orbinho), the Director of Content at Opta, was led to tweet in June that “Ospina suffers from the typical scouting/fan faux pas in that he doesn’t look like a goalkeeper” – clearly suggestive that Ospina, and other short goalkeepers for that matter, suffer negatively from sight-based prejudice despite being, in terms of ability, no worse off for their stature.
The popular Arsenal tweeter @Arse2Mouse responded to Bateman, and suggested that the criticism of Ospina was merited and not simply the result of height-based prejudice, as he stated that Ospina was fond of “parrying into dangerous areas, [and] coming but failing to claim….” Bateman replied, and suggested that Ospina was no different to any other keeper, but that as Arsenal fans he and @Arse2Mouse “just don’t watch them week in week out.”
The statistics from the 2014/15 season, derived from the ‘Safe Hands Rating’ (data from squawka.com) that I created in an attempt to objectively rate goalkeepers (a metric not without flaws which will be the subject of another blog post in the near future), appear to side with @Arse2Mouse. It would seem in the 2014/15 season that Ospina was indeed aerially fallible when compared with the other goalkeepers in the Premier League. Claims % = Successful Claims/Attempted Claims.
But to what extent is this down to the fact that Ospina is relatively short? Did he fail to claim 7.3% of the crosses he attempted to catch in the 2014/15 season as a result of his height, or just because he was simply relatively poor at catching?
The idea that short keepers are inferior at claiming crosses when compared with their tall counterparts is a fairly commonly made suggestion, and, intuitively, would seem to be central to the very idea that tall goalkeepers are better. The goalkeeping coach and scout Bob Warby has commented on this on an article at keeperportal.co.uk. Warby noted that as he is of the opinion that height is not important for a goalkeeper, and instead prefers to judge young goalkeepers based on their handling abilities, he is placed “firmly in the minority.”
It is certainly a fairly common perception that taller goalkeepers are able to claim crosses that shorter goalkeepers are simply unable to reach. The heady positioning of the likes of Hugo Lloris, David de Gea, Lukasz Fabianski and Adrian – all goalkeepers shorter than the league average – and the lowly positioning of Szczesny and Simon Mignolet in the ranking of Claims %, however, appears to suggest that height is far from the be all and end all when it comes to successfully claiming crosses.
So what is the relationship between height and cross collection? In all analytics it is important that the factor being analysed is repeatable (“the true holy grail of soccer statistics” according to the OptaPro Blog), and an analysis of the repeatability of cross claims % from one season to another by goalkeepers in Europe’s Top 5 Leagues is a blog post that I have planned for the coming weeks. It is worth noting that the seasonal data for Ospina is fairly unreliable – a by-product of his having played only half a season in comparison with many of the other goalkeepers in the table. All four of his failed catches could be down to bad luck, for example.
Sample size is important to correct this flaw, as the larger the amount of crosses that a goalkeeper attempts to catch increases the chances that random factors such as luck and the quality of crosses faced will be smoothed out. Accordingly, I have decided to only consider for analysis the 40 goalkeepers in Europe’s Top 5 Leagues who over the past three seasons have attempted to claim the most crosses. This throws up a sample in which every goalkeeper considered has played between 5628 (Mattia Perin) and 10260 (Stephane Ruffier) minutes of Top 5 League football in the past three seasons, with the range in attempted claims spanning from a minimum of 220 (Diego Lopez) to 400 (Brad Guzan). The top 10 by Claims % of this sample appears as follows:
First of all, it is a startling fact that over three seasons of La Liga and Premier League football, Thibaut Courtois – incidentally the tallest player in the sample – spilt just one of the 289 crosses that he made an effort to catch. After that, however, it seems that height really is an irrelevant factor by which to predict a goalkeepers Claims %, even over three seasons of play.
This suspicion can be confirmed with the help of a scatter graph and a regression line when the data for all 40 goalkeepers in the sample is considered:
There is a fair variation amongst the goalkeepers in terms of both height and in terms of Claims %. Heights vary by 18cm between Courtois and Geoffrey Jourdren of Montpellier at 1.81m, although it is worth noting that this is still sufficient to place Jourdren at somewhere around the 70th percentile in height for a young male. Meanwhile, there is a margin of 6 percentage points separating Courtois from Reims’s Kossi Agassa, the worst in the sample (93.561%) in terms of Claims %, which across a sample of 300 shots would amount to around 19 more completely botched claims for Agassa than for Courtois.
Statistically, the relationship between Height and Crosses Claimed is virtually nonexistent, with the data points apparently randomly scattered. The R² measurement of 0.0068 means that only 0.68% of the variation in Claims % can be accounted for by a goalkeeper’s height, with a regression analysis exposing the fact that the relationship between height and Claims % over three seasons worth of data is horribly statistically insignificant, with Significance F = 0.928 (statistical significance = <0.05). The correlation coefficient of the variables, incidentally, stands at 0.0825 – sufficient for a categorization of either “zero,” “trivial,” or “weak,” depending on whose definition is applied.
To conclude, then, if one was to believe that a tall goalkeeper will be able to claim a higher percentage of crosses simply by virtue of their being tall, this analysis would indicate that they would be sorely mistaken. It would appear that Bob Warby’s approach – to assess potential in young goalkeepers by way of their handling abilities rather than by their heights – is entirely correct.
Epstein, David. (2013) The Sports Gene: Talent, Practice and the Truth About Success. (London: Yellow Jersey Press).
Kuper, Simon and Stefan Szymanski. (2012) Soccernomics: Why Transfers Fail, Why Spain Rule the World and Other Curious Football Phenomena Explained. (London: HarperSport).
Nevill, Alan, Holder, Roger and Watts, Alan. (2009) ‘The changing shape of “successful” professional footballers’, Journal of Sport Sciences, Mar., 7(5): 419-26.
Ziv, Gal and Lidor, Ronnie. (2011) ‘Physical Characteristics, Physiological Attributes, and On-Field Performances of Soccer Goalkeepers’, International Journal of Sport Physiology and Performance, Dec., 6(4): 509-24.
All data contained in this article is sourced from http://www.squawka.com/football-player-rankings, with all the graphics and data analyses created and performed by the author in Excel.