Are claims of Celtic's defensive left side being weak true?

Both left-sided positions were identified by many commentators at the start of the season as areas to improve. We need to just ‘get on with it’ at this stage of the season but one of the objectives I always have with Celtic By Numbers is to move beyond concepts of ‘stronger’ and ‘weaker’ and empirically assess what that means.

Note, that StatsBomb data is for the SPFL only. My data covers all competitions. Another context to have in mind is that team style and relative dominance over the rest of the league influence these numbers.

Full-back

StatsBomb provides two useful views. Firstly, defensive activity.

In defending, Johnston has the edge in making fewer errors, being more aerially dominant, pressing and counter-pressing, and defending opposition dribbles. All the aforementioned are symptomatic of Johnston being the more athletic player, and therefore this isn’t surprising. Despite all that, StatsBomb has Taylor with the slightly higher defensive action on ball value (0.02 to 0.00 – nope, me neither). He concedes fewer fouls and recovers the ball slightly more often. Overall, a clear advantage on the right mainly due to Taylor’s relative athleticism deficit.

Secondly, ball progression.

Very similar shaped profiles indicate that how Celtic use their full backs is similar on both sides. Taylor has more of an edge in certain areas here. He is more effective in carrying the ball forward in the dribble. And his passing under pressure is more assured. Otherwise, Johnston makes fewer errors again and has a much higher passing on-ball value (0.37 to 0.18). He is also safer in possession generating fewer turnovers and is dispossessed less often.

Covering a few different data points with my own data.

Johnston is generating nearly twice the xA despite their volume of chance creation being similar. His cross success-rate is also superior, as is his long ball completion per centage. My composite defending metrics – defensive action success rate and possession won from defensive actions – also show Johnston to be much more successful in defensive actions. Taylor is packed less often meaning Johnston is caught ahead of the ball more often, but otherwise, the more athletic Canadian is ahead in most defensive and ball-creation statistics.

As such, this is a good benchmark dataset for recruiting an equivalent left-sided full-back.

Centre-back

There is no doubt Scales has been a dependable presence virtually the whole season, whilst others have been either injured or not performed. Left-footed centre-backs are like hen's teeth and good ones even rarer. No one would have predicted that last season’s eighth-choice defender would be such a consistent presence in the side.

That is not to say many don’t feel Celtic need an upgrade to partner Carter-Vickers but is the gap as big as some think?

Some may be surprised that Scales has the higher overall defensive action on ball value (0.04 to 0.02). He also regains the ball more often and is more successful defending opposition dribbles.  Carter-Vickers is more successful in aerial duels and makes fewer errors, which is not a surprise.

In build-up, they have almost identical records suggesting for this view for centre-backs the team orders and style may be a more important factor than individual profiles. Scales’s passing on ball value is slightly higher to be fair, but he makes more errors and is less secure when placed under opposition pressure.

Overall, according to StatsBomb, there is not much differentiation between the pair. However, there is always more data…

My aggregated defending metrics have Cater-Vickers ahead with a defensive action success rate advantage (82 per cent to 76 per cent) and possession wins from defensive actions (93 per cent to 87 per cent). Both the American’s data points would profile him near the top of Celtic centre-backs across all my data. He is also superior in aerial win per cent with an advantage of 87 per cent to 76 per cent, a notable weakness in the Irishman’s game.

Carter-Vickers is more accurate with long passing, but Scales does complete far more pack passes per game. This may be a function of him being allowed the ball by opponents and I do not have up-to-date metrics for speed of passing. One feature of Scales is that although his error rate is higher, he does concede relatively few fouls. Both recover and lose the ball in their defensive third at a similar rate. This is an interesting comparison and lesson in the use of data sources. I think the StatsBomb data misses some key differentiators that if you were using the tool for scouting, may lead you to poor decisions. My take, of course.

Back to the topic. Scales is to continue to be applauded and will no doubt remain the first pick for the rest of the season. But his defensive metrics do lag in comparison to Carter-Vickers and his relatively poor aerial duel rate stands out. He does progress the ball well though although without the data to back this up, at quite a slow rate relative to the others in the team.

Summary

The objective here was to put facts together to illustrate the oft-opined view that the right side of the Celtic defence is stronger. The differences are often marginal but that is not untypical at this level of football. I conclude with the fact that the opposition would agree with the assertion that Celtic’s left is weaker than the right.

The prevalence of the yellow/red coloured shapes and indeed all shapes on the right compared to the left shows that opponents favour and find more success attacking Celtic’s left.