Hawkes Bizarre – What can Szczesny’s decision making tell us about luck?

If you said to the average football fan that there’s no skill involved in around a quarter of the saves professional goalkeepers make they’d say you were bonkers. No doubt anyone of you or us would be hopeless if we were stuck between the sticks in a Premier League game. There’d be guaranteed laughs (almost as many as you’d get by watching Hugo Lloris this season).  But it’s all relative. What we’d like to know is whether there’s any difference between goalkeepers’ shot stopping abilities once they get to Premier League level.

Last time out we explained how we’ve developed a model to rate goalkeepers based on the location of shots they face. Having gone into great detail to cover as many angles and distances in the original model, for these tests we’ve regrouped locations into larger areas. All those areas with similar save rates have been latched together to give the biggest sample size possible without losing the accuracy of distance/angle of shot.

We’ve split the areas into 4 main zones:

As you can see, the average chance of a keeper making a save from a shot on target from Zone 1 is 44%. There’s more chance of them calling heads or tails in a coin toss than there is of them making a save from here. The plot below shows the main keepers’ record over the last 3.5 seasons when facing shots from Zone 1:

The thick black line in the middle of the plot is the Premier League average. Anyone below the line is below par, anyone above it, is, erm, above par. The curved lines at the top and bottom of the chart represent two standard deviations from the mean. If any of the keepers’ red dots were above or below these lines then you could be pretty sure they’re showing some real skill (or total lack of it) when dealing with these Zone 1 shots on target. None of the keepers actually are. Vorm, however, does stick out from the group. There may be more on this in future on Differentgame.

There’s a school of thought that we shouldn’t be evaluating keepers based on their shot stopping abilities because measuring save % isn’t reflective of true ability. @JamesWGrayson (the author of that piece) is a constant source of great literature on football numbers and you should follow him on Twitter and read his blog. The ‘funnel’ plot above is inspired (swiped?) from his previous works.

After our piece and James’ follow up post, Wim Reckers, keeping coach and scout for SC Heerenveen in Holland’s Eredivisie, made the point on Twitter that goalkeeping was about ‘a whole lot more than shot stopping’. Of course, he’s correct. Wojciech Szczesny’s performance against Napoli this week provided evidence of this:

Now, you may look at Higuain’s goal and simply see a good finish. And you’d be right. It’s fired neatly into the corner. However, Szczesny’s decision to take several quick steps forward as Higuain receives the ball back to goal looks bizarre to us. We’d love to know why the Pole does this – there is a defender covering Higuain as he receives the ball. Watch it again, and you can see that if Szczesny doesn’t have an inexplicable rush of blood and simply stays where he is, he could have saved the shot whilst having a cup of tea. As it is, he simply doesn’t have time to get down to the shot as it passes underneath his hand. This is the danger of leaving the Keepers’ Comfort Zone.

Szczesny’s decision to come early for Callejon’s goal is possibly worse. The looping ball is never even in front of Callejon until the very last moment. How on earth does Szczesny think he can get to it first? He makes Callejon’s decision to shoot for him. The Spaniard strikes the ball when it is bouncing at waist height. There is no other possible shot on – especially as Koscielny is covering. If Szczesny stays home as he should, there is no way Callejon makes the attempt at goal look so easy.

Here’s a Dennis Bergkamp quote:

“When I played in Holland, I always tried to lob the goalkeeper. People used to say, ‘Oh, you’re always only trying to make a nice goal’. But I said, ‘listen, if the goalie is a little bit off his line, how much space do you have on his left or right? It’s not a lot. And how much space do you have above him? There is more. It’s a question of mathematics.”

 

The goals in that Arsenal v Napoli tie illustrate why we think  shot-stopping envelopes a whole range of processes going on within the goalkeeper’s skill-set. Shot-stopping is about concentration, positioning, decision making and foot movement before we even get to the agility part of the equation. Where the ball is, is a huge factor in this process. Where defenders are also greatly effect all these processes too. We feel that with enough data in the bank, the model does a good job of capturing this. The model does not simply measure a single skill-set called ‘shot-stopping’.

Once we move out of Zone 1 and into Zone 2, we can perhaps begin to see these processes taking effect on an individual keeper’s save %. Here’s the plot for Zone 2:

Although once more we see no keeper outside the curved lines (which may statistically indicate any kind of special ability) the spread between keeper’s save % is much greater here. There’s over a 20% gap between top man Mignolet and our old mucker, Wojciech. That translates as both keepers conceding pretty much the same number of goals (27-28) from Zone 2 despite one (Mignolet) facing twice as many shots. The data covers around 100 games and over 3 seasons of football for both.

We think that when the ball arrives in Zone 2, the goalkeeper has the biggest amount of decisions to make and this accounts for the greater variance in save %. Of course luck still has its roll to play – an attacker still has to pop the ball on target in the right way to score. We don’t think keepers influence this a great deal (and we’ll be using data to study this as the season progresses). Sure, in the example above, Callejon makes dinking the ball over Szczesny look easy. It isn’t. The simple fact is, all keepers make small (and large) technical mistakes every week. The mistakes aren’t horrendously exposed all the time because the attacker still has work to do in getting his own wotsit together.

We’ll look at the story from Zones 3 and 4 next time out. In the meantime, try and watch out for goalkeeping mistakes next time you watch a game. It’s good fun. Honest.

Follow us here on Twitter.