Celtic may not need world class shot stopper to be goalkeeper in Ange Postecoglou's side

Is it analytically "crazy" that many supporters prioritise shot-stopping skills for the keeper positions?

Not really ... this is a situation in which conventional wisdom is seemingly accurate. However, Ange Postecoglou’s use of the keeping position is not conventional and, in fact, I believe Celtic are facing a situation in the upcoming season in which a below-average shot-stopper could still be a better fit for Ange’s system.

Ange Postecoglou’s playing style in Japan included a very heavy involvement of his goalkeepers in build-up play and supporting its high line and press-centric defending.

The following graphic shows the average number of passes made and received from a number of keepers for various seasons, to put this into perspective. Iikura and Il-Gyu were Ange’s primary keepers in his first two seasons in Japan, 2018 and 2019.

It's fair to say Ange expects his keepers to be vastly more involved in ball distribution, movement, and outfield play than most managers.

Here are some total action maps to offer some context - all are from the game with the highest total actions for each keeper in the respective seasons:

The first two are Park Il-Gyu vs Shonan Bellmore playing for Ange in 2019, and Manuel Neuer vs Bayer Leverkusen last season.

The next two are Brad Guzan of Atlanta United vs Nashville this season, and Fraser Forster vs Hamilton in December 2019:

Now here are Barkas’ from the friendly vs Bristol City and Midtjylland this summer:

Even in a game which included a significant period playing with a player less, we can see a difference.

Why include Guzan? He is averaging nearly 35 passes per game for Atlanta United this season, which is a very high level for a keeper. However, as we can see via his action map, the 36-year-old keeper is not nearly as dynamic as a "sweeper". His inclusion also is a segue into introducing the Goals Added (g+) model from American Soccer Analysis.

This model attempts to value the actions each player is involved in within a match, and how they contribute to goals scored and conceded. The hyperlink leads to the announcement that the model has been extended to include goalkeepers. Unfortunately, they do not cover the Scottish league, but I think some general inferences can be drawn from analyzing the model for MLS keepers.

Guzan’s passing and sweeping components within the g+ model so far this season have been 0.045 per 90 minutes played, which suggests that those aspects of his play have contributed to that many goals on average.

This is an important concept, in my opinion, as Ange Postecoglou’s keeper position at Celtic may be a bit of a "purple unicorn". 

What do I mean by this? Manuel Neuer and Brad Guzan faced 3.71 and 4.26 shots per 90 minutes in the respective seasons referenced.

Celtic keepers simply do not face anywhere near that level of goal threat domestically. Forster faced 2.38 per 90 minutes in league games in the 2019-2020 season, and Barkas just 1.79 last season.

READ MORE: Why Ange Postecoglou's Celtic brand is winning the fans over - Sean McDonald

This creates a very unusual situation for the Celtic keeper position, where shot-stopping skills offer unusually low value, and with the introduction of an Ange system, may dramatically increase the position’s value in passing, build-up play, and sweeping.

How have recent Celtic keepers profiled relative to this purple unicorn role? Here we have a radar comparing Barkas and Bain in league games last season, 2020-2021. The xG and goal metrics exclude penalties and own goals.

This shows Bain was around average versus other keepers in the league relative to shot-stopping and commanding his box. CCAA% is a metric to gauge how much keepers move to claim passes/crosses. At the same time, he was pretty good with distribution, though struggled passing when pressured.

Barkas was a poor shot-stopper but commanded his box well, all while measuring towards the very top across passing metrics - including under pressure.

For additional context, this radar compares Barkas from last season with Forster in league games from 2019-2020 season.

Here we see Forster was around average in shot-stopping and commanding his box, while rather mixed in passing metrics, and one of the lowest when being pressured. The shot-stopping metric is likely to surprise some, I suspect.

The difference between Barkas’ PSxG faced metric and goals conceded was just -0.07, with Celtic’s level of non-penalty or own goals conceded at just 0.45 per 90 minutes. As poor as his shot-stopping was, Celtic conceded fewer non-penalty or own goals with him playing at keeper than when Forster played during 2019-2020.

This introduces an interesting question.

Even if Barkas is destined to be a below-average shot-stopper, would he still be a better fit for Ange’s Celtic than a keeper like Forster?

I believe it is reasonable to expect a player of Barkas’ ball-handling skills to exceed Guzan’s g+ of 0.045 playing in Ange’s system, which would largely, if not completely, outweigh the -0.07 in shot-stopping underperformance Barkas displayed last season.

READ MORE: Celtic goalkeeper Vasilis Barkas has talent but he needs to show it fast

In addition, placing a keeper like Forster into this purple unicorn role, for which he is likely poorly suited, could drive more negative value than positive. Given his poor performance level while passing under pressure, how would he perform while being pressed regularly 35 to 40 yards from goal? How many goals could that end up conceding?

To consider the potential scale of poor keeper play in the passing and sweeping metrics, so far this season, the worst-performing keeper in MLS in those metrics has cost his team -0.10 g+ per 90 minutes. That keeper has only averaged 16.38 passes per 90 minutes.  

Ideally, Vasilis Barkas can revert back to being at least an average shot-stopper, which he had been over his 10,000+ minutes in Greece prior to his Celtic arrival, all while competently filling the “sweeper-keeper” role required in Ange’s style of play.

Even if Celtic decide an upgrade/replacement is preferred, then Fraser Forster looks more like a 'Shane Duffy', and very unlikely to be a purple unicorn.