The woes of a team can never be down to a single player; but whilst we like to think each position is equally valuable, there are positions that – if weak – are more crippling to the team than others. Those positions are arguably a lead centre forward, central midfielder(s) (which is critical to control), centre back(s), and goalkeeper. These represent the spine of a team and are indeed the foundation on which a first 11 is built.

Each man does matter; but just as the old adage in American Football that defense wins championships holds considerable weight and truth, the same can be said of the universal version of football. That is not to say an attack is not important, but a solid defensive structure more easily compensates for a faltering attack than a fluid attack does for a poor defensive spine. Pep Guardiola has found the going tough at Man City this season. Who knew he’d be baptised in fire 21 games in? Close observers of the catastrophe that has developed over the recent past at the club, that’s who. The very base of the spine will be the focus of this piece. Fast becoming the football equivalent of American President-elect as a comedy goldmine, former Barcelona keeper Claudio Bravo must answer a few questions as to why such an expensively assembled squad is outside the top four.

How bad has it been?

Monumentally bad is the answer to most. Is this true? One graphic lays the foundations of a harrowing story.

Oddly, five of the worst save percentages since 09/10 are from this season in a league wide trend. Bravo is second worst of that group.

Bravo has the fourth worst save % of any goalkeeper in the Premier league since 2009/10. We must understand the embryonic stage that goalkeeping stats are in. Compared to the fairly accurate measures for other positions, the defensive ones need to consider many more almost unquantifiable measures such as organisation of a defense, reading of the game, and positioning (although this does contribute to save %) for a goalkeeper, amongst others. Still, unless Bravo has been facing Steven Gerrard screamers from 10 yards out all season, this is borderline embarrassing for a club of Man City’s standing and aspirations.

The Man City playing style presents different challenges too. The high defensive line leaves him open to breakaways and counterattacks. This doesn’t absolve Bravo completely. Examples of world class keepers who have had to ply their trade facing similar challenges are Hugo Lloris and Manuel Neuer. Under Andre Villas-Boas, Lloris came under some fire for his apparent clumsiness. He was though, faced with disproportionate amounts of breakaway attacks for which he required to stomp out as a sweeper-keeper. It produced a few mistakes. Again, he wasn’t blameless in his first season in England – something that has proved difficult for top quality keepers before. A look at the save records of these keepers does tell a story – albeit one excluding adjustments for shot quality faced. The vanquished Joe Hart also enters the analysis.

Saves, save ratios and errors: 12/13-16/17, weighted estimate

David de Gea has made the most saves per game – indicative of what Manchester United fans know – that the defense has hung him out to dry way too often. He’s had to protect United alongside an ageing and still marauding for some reason Patrice Evra, Rafael da Silva (enough said), Jonny Evans, Tyler Blackett, and Marcos Rojo amongst others. Even with those colleagues, he’s made the least errors and is in a virtual tie with Manuel Neuer for the least errors leading to a goal. De Gea’s stats include saves for his first season where he faced constant criticism and the wisdom of his purchase was questioned.

Claudio Bravo, though, was required to make more saves at Real Sociedad. At Barcelona, he was solid all round and showed no signs of major weakness apart from a relatively high amount of errors compared to the esteemed sample of keepers here. The two seasons which should be honed in on, though, are Lloris 12/13 and Bravo 16/17. Both are often (were) being criminally left out to dry by their teams’ style of defending and almost concede a goal every other shot. Hart does have the worst shot/goal ratio. This is severely skewed by his exploits in 12/13 and 13/14, though, and since then that ratio has been 2.12. He is the most error prone and reinforces the thinking behind the keeper change.

We must delve deeper. In lieu of concrete shot difficulty measures (there are some proxies), the amount of defensive errors committed by teammates will further illuminate the dark room of goalkeeper rating.

Defensive Errors

Isolating the defensive errors not made by the keepers themselves, we get an idea of defending quality and the amount of undue distress they have had to face. Neuer benefits heavily from the supreme dominance of Bayern Munich in possession and defensive quality. Lloris’ defense has let him down the most, and Bravo and Hart fall in the middle. However, going back to the seasons of interest we see that our theory carries weight and it does not bail Bravo out.

Defensive errors made by players other than the GK: 12/13-16/17, weighted estimate

Lloris’ teammates committed 0.65 defensive errors per game, of which 0.17 lead to a goal. That would be roughly 25 defensive errors, leading to seven goals over the 12/13 season. He made his errors, but the suicidal line which made for comedic errors and drubbings galore and AvB’s sacking hurt the French man significantly. Just by these numbers alone, he stands right alongside Neuer and De Gea as likely three of the top five keepers in the world.

As for Bravo, City’s defense is catastrophic by perception, but they have had so much of the ball that these errors have not quantified as much as the more under pressure defences of Man United and Spurs. However, 0.13 goals per 90 mins or a total of three have been caused by errors made by players not named Bravo. Also, eight defensive errors total have been made so the volumes of errors, although egregious and costly, have not been a major factor in the Bravo trainwreck.

Cannons Unleashed

This is also a difficult one to quantify. However, some approximations can be made from the fast-developing analytics world in football. Expected goals and expected save differences again paint an ugly picture for Mr. Bravo. This also serves as an indicator of keeper quality.

Shot Stopping Indicators Indicators (Source: Simon Greaves – @analytic_footy)

It’s no coincidence that Man City, Liverpool and Watford lie at the bottom of the left-half of the graphic. Any team in the bottom half on both sides has had keepers saving less than expected given calculations made based on shot position etc. In the same vein, Tottenham and Chelsea falling at the very top is no surprise and just indicates to a great extent the shot stopping skill of their keepers this season. Man United and De Gea lie in the middle, 11th overall. De Gea is effectively saving what is expected on average and it does tie in to the feeling that he has been below his usual superhuman levels. However, other facets of his game have improved.

Good (bad) in this sense simply means that the defense has conceded a fewer (greater) amounts of quality shots, so no need to convulse at the thought that Liverpool and Man City had a good defense. Lucky (unlucky), by the same token, means that the keeper has saved more (less) than is expected. There is one caveat which makes it more difficult to pinpoint the culprit. Expected goals cannot factor in defensive pressure and, therefore, one has to be mindful that lax blocking or covering leaves the keeper exposed as we have seen Bravo a couple times. Amazingly, even with that he can’t be absolved because of his own positioning which has allowed for shot stopping to be that more difficult.

Condolences to those teams falling in the lower right quadrant (bad and unlucky). Clearly this is where Leicester and the relegation strugglers would have fallen, right?

Chance quality Indicators (Source: Simon Greaves – @analytic_footy)

This graphic indicates that the higher up (lower down) you fall the higher (lower) quality shots you have faced. Bravo hasn’t faced the cannons that would excuse such an abysmal save ratio. Four teams have faced shots with higher average expected goals, and it does explain the defensive plights of these teams too. It also suggests below-average defensive performance, defensive organisation, and a vulnerable defensive tactical setup which concedes higher quality shots.

Finally, the further to the right a team falls the more chances you concede. At least Liverpool and Man City concede less chances than anybody else which prevents a major worry from being devastating.

Critical decision ahead

Of course, Joe Hart is imperfect. His errors have also been costly and he is often too casual. Man City perhaps needed to move on from the inconsistent England international to move to the next level but, if you replace someone, the replacement performing far worse than the replaced does call the decision into question. Bravo didn’t give fair warning based on his Barcelona stint but, to be fair, his defense has been a letdown.

However, we cannot tell him ‘Bravo’ for a job well done. He could have the distribution of Zinedine Zidane and Xavi Hernandez; his shot stopping has been so meek that it overshadows any possible benefit.

P.S. His passing accuracy is also the lowest since he moved to Barcelona at 78% and with good reason. There is a trick to this stat in passing distances that will be considered when we tackle the keeper rating minefield. His pass completion was 55% and 62% in his last two seasons at Real Sociedad – for example – where his passes were at least one and a half times longer and therefore more difficult on average. We must be mindful of taking statistics without proper context.