I recently wrote a blog about how Celtic had been performing this season and a small part of it focused on Craig Gordon. It turned out that the Hoops goalkeeper had the third highest save percentage in the division at the time of writing, behind only Aberdeen’s Scott Brown/Danny Ward combination and Neil Alexander of Hearts:
A throwaway sentence within that piece sparked a Twitter exchange that found its way to Dundee goalkeeper Scott Bain, a comment he has since deleted. He was understandably unhappy at being called out for having the second lowest save percentage in the league.
Funnily enough Bain has just been called up to the Scotland squad for the friendly with Czech Republic thanks to his performances this season. He could even make his debut this time, having not featured when last involved with the squad in May 2015.
I should start by saying that I think goalkeeper is one of the hardest positions to analyse. Using statistics to judge players is still in its relative infancy and it can be difficult to know which statistics are useful and which aren’t. Indeed, goalkeepers often have so few interactions that their individual data sets are quite small.
Key elements of being a goalkeeper such as organising the defence, commanding the box, sweeping behind the back line and judgement when catching or punching crosses are hard things to rate. The analytics community often decries this, but at times these so-called intangibles can be just as important as the performance elements that can be measured more easily.
I cannot deny that Scott Bain looks impressive when you watch him play. He has produced a number of spectacular saves for a team that has only been unable to carve out a comfortable top half position due to inconsistency. One really memorable stop came in the 93rd minute of the first Dundee derby of the season. With his side 2-1 down he tipped a close range effort onto the underside of the bar, quite brilliantly in fact. Less than a minute later Dundee went up the other end and snatched a last minute equaliser. In this instance Bain preserved local pride and played a huge role in earning an all-important point for his club.
While it’s often easy to remember spectacular saves like this one, or the more glaring mistakes that a goalkeeper makes, statistics can be used to help with the more routine aspects of performance that can often escape our memory.
It is important to state here, once again, that I believe using shots on target in any metric is flawed. At Stratagem we collect deeper information using our trained analysts and rate chances using our own definitions of Great, Good and Attempt. This information can give us a much better understanding of which goalkeepers are saving shots that they would be expected to, or not, as the case may be:
There are a few points to note from this table. While some goalkeepers, such as Jamie MacDonald of Kilmarnock, have had to face a lot of great and good chances, others have not. Craig Gordon has only had 41 great or good chances on target to deal with in the 26 games he has played. This naturally increases the variance and means that his numbers could shift quite considerably in the space of a match or two. However, it is still a good starting point and a much more accurate reflection than just using shots on target.
When looking more closely at Scott Bain, it is apparent that his figures still do not stand up too well, with a save rate of just 48.78%. This roughly means that for every great or good chance on target Dundee concede, every other one is going in. When the average conversion rate of great chances alone is around 40%, this is concerning. To look at a couple of goalkeepers who are similar in terms of numbers, Alan Mannus has only conceded 34 from an almost identical amount of great/good chances and Conor Ripley has let in just 31.
However, it is key to state that from both of these comparisons Dundee have conceded a much higher number of great chances, ones we would naturally expect the opposition to score more often:
The Dundee defence does appear to get breached with more regularity than that of St. Johnstone and Motherwell. In fact they are giving up great chances at a rate of 1.5 and 2 times more than these close divisional rivals, respectively. So while Bain’s save percentage may look less impressive in direct comparison, it is nowhere near as bad as it seems. Indeed, it’s the Dundee defenders that should be taking the majority of criticism for exposing their goalkeeper more than other teams of a similar level do.
What should also be highlighted from the above table is that attempts have been removed completely. These types of chance are much less likely to lead to goals because they include events such as speculative 30-yard shots on target that goalkeepers usually deal with quite comfortably.
Bain has an outstanding record in this regard, conceding only twice from 133 attempts faced. In short, it seems that if Dundee can force their opponents to shoot from range or poorer positions then Bain would be in his element. When adding context and again comparing his numbers to those of Mannus and Ripley, who have conceded 10/155 and 11/175 attempts respectively, we begin to see a different picture emerging.
Dundee United’s problems have been well publicised and many people have pointed to the goalkeeping situation as one of the main reasons they are rooted to the bottom of the league. But have things improved for them since Kawashima came in?
The figures say “no”, to put it bluntly.
Kawashima and Szromnik sit at the bottom of the above table, though it must be said that as both have only played eight games the number of great and good chances they have faced is very low. Zwick has a better record from 12 games, but is far from stellar himself:
Again, looking at the deeper metrics begins to show that Kawashima has been more exposed on the great chances he has faced. He has performed better when facing good chances than Szromnik and again this defence look like they are allowing great chances too regularly. Interestingly though, Dundee United have had to deal with less great chances on target than Dundee (33 vs. 42), but have conceded them at a much higher rate. This is slightly surprising given their respective positions in the table.
So in conclusion, it is probably fair to say that there are no significant goalkeeping problems at either Dundee club. This should make Scott Bain happy, at least!
I feel comfortable asserting as much because of my belief in great/good chances being a much more useful metric to consider than shots on target when measuring goalkeeping ability.
For more in-depth team analysis, trading ideas, statistical modelling, proprietary data and unrivalled trading tools/execution, start your 30-day free trial of StrataBet today.
Dave Willoughby (@donceno)