Hmmm. I can see that they would not cherish the notion that they have only a defensive role to play, a notion that really characterizes them as negative players rather than creative ones.
Maybe they feel left out -- after all, these days fullbacks race down the wings and get involved in all sorts of attacking play -- and they score goals. So, occasionally do those big center backs, mostly at corner kicks and other set plays.
But goalkeepers are usually about 80 yards away when their team scores ... so, goalkeepers as creative players? That’s a bit of a stretch, guys.
Yet the idea of goalkeepers as goalscorers is reflected -- rather ridiculously -- by the stat people. If, for instance, you check the stats for English goalkeeper Joe Hart, you will find that, in the course of a 15-year career, Hart has scored no goals, not even one. This totally predictable drought is faithfully reflected in a massive table that finds it necessary to include several columns for “goals scored” all filled with “0" entries -- no fewer than 96 of them. Might was well add a column for “Saves” to Messi’s record.
The upshot of my ponderings about attack-minded goalkeepers was that I decided, a couple of weeks back, to do a survey. A very limited one, you understand. From the weekend of Nov. 4-5 I chose five games at random -- a German Bundesliga game, two MLS games and two English Premier League games.
Actually, not quite at random. I made sure to include a Manchester City game. Because coach Pep Guardiola, who undoubtedly wants his team to play attacking soccer, seems to be at pains to want a goalkeeper who can be a part of this style. More about that shortly.
So these were the five games I watched (boy, did I ever -- I’ve watched each of these games five times from start to finish, with plenty of pauses):
Borussia Dortmund 1 Bayern Munich 3
NYCFC 2 Columbus Crew 0
Toronto 0 NY Red Bulls 1
Manchester City 3 Arsenal 1
Tottenham 1 Crystal Palace 0
A pretty representative group, as it turned out: three home and two away wins; no massive scores, a total of 12 goals for an average of 2.4 a game; no penalty kicks; 2 red cards, but in the same game (one each for Toronto and the Red Bulls, for whom Jozy Altidore and Sacha Kljestan displayed their professional maturity by getting sent off. In the tunnel. At halftime.) -- so no 10v11 imbalance; and no upsets.
As for the 10 goalkeepers, there were four Americans, two Argentines, and one each from Germany, Brazil, Switzerland and the Czech Republic. Maybe only two stood out: Arsenal’s 35-year-old veteran Petr Cech, and the young Ederson (24), Guardiola’s special choice for the Man City job. Both of them left-footed, the only two lefties in the group and, as it happened, they would play against each other in the ManCity-Arsenal game. In another coincidence, the two Argentines played against each other … at the very English Wembley Stadium venue.
What I was looking for was (1) the number of occasions in which each goalkeeper took possession of the ball and had to distribute it; (2) the nature of the distribution -- in particular, whether the ball was played long or short; and (3) evidence for preferring long-ball distribution to short, or vice versa.
My analysis of the five games came up with 280 goalkeeper possessions. It was clear that most of these “goalkeeper possession” situations resulted either from the keeper receiving the ball via a back pass from a teammate, or when he took a goal kick. (Goalkeepers do not always take all the goal kicks -- but in these five games, they obligingly did so).
Thus, the first stat: of the 280 possessions, 115 resulted from back passes, and 75 were goal kicks. Meaning that 68% -- just over two thirds of the goalkeepers’ possessions resulted from those two actions.
The remaining one third -- 90 possessions which I classified as Miscellaneous -- were made up mostly of controlling, with hands or feet, loose balls and poor passes (51) and taking free kicks (9). Competitive situations accounted for most of the rest -- diving at opponent’s feet (6), possessing the ball when making, or immediately after, making a save, plus catching crosses, corner kicks, free kicks.
Rather too many causes for possession, I thought. To simplify matters, I decided to limit my inquiry to the back-pass and the goal-kick possessions. All 190 of them.
I wanted to know: what did the keepers do with those 75 goal kicks and those 115 possessions from back passes -- 190 occasions when they would have to use their feet to distribute? These 190 examples can, I think, give us some idea of the goalkeeper’s favored means of distribution: long or short.
Another simplification was necessary because I had trouble deciding at what point a short pass became a long one -- 20 yards maybe, or should it be 30 yards? I found that a better, certainly more useful, classification was to identify passes made to feet (nearly always within a 20-yard range), and aerial passes (most of the long ones).
These were the figures: of the 190 possessions, 80 (42%) were distributed to feet, 110 (58%) were aerial passes. A clear preference for long passes. That preference was significantly higher if only the 75 goal kicks are considered: two-thirds of them (50) were hit long.
Maybe that is simply habit. I have spent a long lifetime watching goalkeepers slam their kicks way downfield, with the seeming intention of distance rather than accuracy.
The accuracy of the short, to-feet passes is not in doubt. All of them went directly to a (frequently unmarked) teammate. What is the accuracy of the long aerial kicks? How often does the kicker’s team get possession, how often does the ball end up with the opponents? Well, I charted that, too, marking the end result of a long kick as either +ve (ball in possession of the kicker’s team), or -ve (ball with the opponents), or neither - occasions when the long kick resulted in an “M” for “melee”, with no clear and immediate possession for either side.
Of the 110 aerial (long) goal kicks and distributions, 13 (12%) resulted in a melee, while 41 (37%) were +ve, and 56 (51%) were negative. Put another way, only just over a third of the long balls retained possession for the kicker’s team. Over half of them immediately passed the ball to the opponents.
Ceding possession that easily does not sound like anything that is going to help a team’s attacking potential. Which is the way I tend to think. But there is another angle that deserves to be looked at. Possession is not everything. Of the 41 +ve long balls, i.e. those that retained possession, none was related in any way to anything close to the scoring of a goal.
That will surprise no one, as we have by now all seen plenty of occasions in which a team claims an overwhelming percentage of possession but loses the game. Long balls (what the English call Route 1), the antithesis of the possession game, can be dangerous. A fact that was demonstrated twice in these games, both occasions resulting from long-ball melees. A Toronto long ball (whacked upfield by goalkeeper Alex Bono after receiving a back pass) resulted in Red Bull defender Aaron Long having to defend desperately to hold off a 1v1 challenge from a Toronto forward, as they both raced toward the Red Bull goal. The fact that the Toronto player was Sebastian Giovinco will tell you just how dangerous the long ball was, and how skillful Long had to be.
Another close shave came in the 92nd minute of the New York City FC-Columbus game, when Columbus’ Kekuta Manneh, sprung loose by a flick-on of a long goal kick, raced unmarked into the New York penalty area, where the danger ended in banana-peel farce, as Manneh slipped flat on to his backside.
Two moments when a long ball (as opposed to a possession-based passing play) caused an immediate goalscoring threat to the opposition. Which seems to say that a goalkeeper can help his team’s attacking efforts just so long as he keeps thumping his goal kicks hard and long.
Now that sort of crudeness cannot, surely, enter into the thinking of Pep Guardiola. I think not. I would judge that, for Manchester City, Guardiola wants a goalkeeper who will not give the ball away on long kicks, a goalkeeper who will feed his teammates with sharp passes to feet, passes that will allow them to get on with their dazzling quick-moving, quick-passing game.
There is the added factor that Man City likes to play its game in the opponent’s half, leaving a lot of space behind them -- thus requiring a keeper agile and quick enough to deal with sudden counter-attacks. In short, a pretty damn good all-around keeper. Maybe Guardiola has found his man in the Brazilian Ederson. Ederson did not have a great deal to do in the game against Arsenal -- which was dominated by Man City -- but his statistics show a strong preference for playing the ball to feet (18 distributions, 16 of them to feet, two long and finding a teammate).
But dare one ask -- would the occasional -- unexpected -- long ball straight down the middle cause problems? The stats -- my stats, damn it! -- suggest it might.
Well, it would cause problems for me because I don’t like Route 1 play. I think the stats I’ve assembled show pretty clearly that the goalkeeper’s biggest contribution to attacking play can be that he shuns the long ball and plays the ball to the feet of teammates who know a lot more about creative play than he does.
But I’ve no wish to be churlish about goalkeepers. Of the 10 keepers I analyzed, the guy who -- for me -- came out looking good, really good, was the least likely candidate: the 25-year-old Paulo Gazzaniga of Tottenham. In the course of Tottenham’s minimal 1-0 win over Crystal Palace, Gazzaniga made three super saves. The TV commentators greeted them with “crucial save,” then “second brilliant save,” and for the third effort “unbelievable save.”
They were all of that, wonderful examples of the goalkeeping art (and forget all that stuff about an attacking role), dramatic, and immensely exciting to behold. What made Gazzaniga’s acrobatics even more special was that he was Tottenham’s third-string keeper, playing at Wembley only because No. 1 and No. 2 were out injured.