I’ve graduated from extreme scepticism to healthy interest regarding StatsBomb’s on-ball value (OBV) metrics concerning Celtic. As mentioned, their video primer explaining the concepts is worth the time.

Very simply, it seems to use StatsBomb’s highly sophisticated expected goals model to ask the question ‘what value of xG did that action add to the team’s overall xG?’. This can be positive or negative.  My opinion is that for passing actions (75 per cent of all on-ball actions), packing concepts are superior given the wealth of data this provides about both passer and receiver. xG becomes abstract and theoretical when describing a pass from centre back to full back 80 yards from goal.

Packing is a practical count of opponents taken out of the game. Intuitively, that feels more useful. However, this is not a theoretical methodology article you will be glad to know. Instead, I want to take a historical view of OBV.  Given the breadth of metrics and players, I will split this into two separate pieces.

In this piece, I will report the OBV values for the defensive players and in the second piece the attacking players. Only players with 900 minutes or more are included. All OBV values are a per 90-minute average.

Goalkeeper OBV

Goalkeepers’ on-ball actions are quite specific given they can use their hands and given the nature of fielding crosses and saving shots. Passing is, of course, a common discipline.

Goalkeeper OBVs are quite low. Given passing is the highest volume activity and given they are furthest from the goal, their passes add low values to xG-driven OBV. The top goalkeeper in the SPFL Premiership last season had an OBV of 0.07 – Kelle Roos from Aberdeen.

Only one goalkeeper has had a positive OBV for Celtic since StatsBomb started publishing the data. That was Scott Bain in 2018-19.  Some forget that Bain replaced Craig Gordon in Brendan Rodgers’ third season near to New Year and on merit. I wrote at length on this on Celtic by Numbers. It may look freakish now but Bain’s campaign that season was excellent. Also, given the xG basis of OBV, the metric is likely ‘penal’ for goalkeepers.

Perception dissonance may be triggered by Joe Hart’s numbers. The extent to which he, through sheer willpower, dragged his performances to be the best season of the three at Celtic in his final professional campaign cannot be forgotten.

But the more prosaic reality is that he was an ageing and declining player who happened to have a magnetic and glowing presence and personality. Being human, we are attracted to the latter, to the exclusion of dwelling on the former. Numbers are cold like that.

In terms of benchmarking, a goalkeeper OBV of 0.00 would seem to be ‘good’. However, other, better, metrics are available for evaluating goalkeepers.

Read more:

Defensive action OBV

I’ll split this into full-backs and centre-backs. Defensive actions include tackles, interceptions, blocks, fouls etc so not passing. Full-backs first.

At first glance, this doesn’t feel intuitive. Was Diego Laxalt the ‘best’ defender at full-back Celtic has had over the last six seasons? Was Anthony Ralston Celtic’s top defender last year? Mikel Lustig – negative defensive action OBV? But, remember this is defensive action only. Our emphasis for Celtic full-backs is often to focus on their attacking outputs.

The other outlier is Alexandro Bernabei with the lowest OBV of -0.16 – that does feel appropriate. Given this covers many of the not-so-noticeable in-game actions, perhaps it is not a surprise it doesn’t feel like it meets the eye test. Am I being charitable?

Let’s take a look at the centre-backs.

According to this metric, the most effective defensive action campaign was Dedryck Boyata in 2018-19 and the least effective was Mortiz Jenz in 2022-23. The second-highest OBV was Stephen Welsh in 2020-21.

Given the relatively low scores for Cameron Carter-Vickers, I suspect this is a weakness in OBV being an accumulation statistic. For example, Liam Scales does much more defending than Carter-Vickers. This is due to how teams attack Celtic. Scales does most of this competently. If we use, for example, defensive action success rate, which is a percentage and therefore not dependent on the accumulation of actions, we have Scales at 76 per cent and Carter-Vickers at 84 per cent. That seems more in line with the eye test.

I have a robust framework for evaluating defensive actions, but I accept that OBV may be picking up something missed elsewhere. However, there needs to be some semblance of meeting the eye test.

Ball progression

I am going to combine passing OBV and dribble and carry OBV into an overall ball progression view. Note that dribbling or carrying the ball is of relatively low volume especially compared to passes. For example, Carter-Vickers completes 82.9 passes per 90 minutes but only 2.94 ball carries.

We’ll do full-backs first.

This view seems closer to the eye test and the outlier of Jeremie Frimpong exemplifies that. He is an elite-level runner with the ball, and this is what that looks like when compared to other Celtic full-backs. How underrated was a fit Emilio Izaguirre? He has the best balance of averages for both passing and carrying the ball.

The good news is the current crop – Greg Taylor, Johnston and Ralston – score well for passing if not for carrying which as I have said is a relatively low-volume activity. To counter the defensive action OBV’s, less well-remembered players such as Laxalt, Jeremy Toljan, JonJoe Kenny and Bernabei are all towards the bottom left quadrant of sadness. So, maybe OBV IS meeting the eye test it is just that our minds don’t tend to remember the more prosaic basic defending activities.

Here are the centre-backs.

Again, this is better aligned with what I see. For example, all of Kristoffer Ajer’s seasons are clustered in the top left, indicating a high level of ball-carrying but average passing outcomes.

Welsh’s 2021-22 campaign was of a high standard but maybe not as well remembered. StatsBomb likes Scales. Both he and Carter-Vickers are in the virtuous top right quadrant for last season’s efforts. Indeed, the American has improved by this metric season over season.

Summary

OBV is a fascinating addition to the advanced metric canon. It highlights focus on areas of the game we as fans do not perhaps store so easily in the ‘moments’ memory banks. Especially the more mundane defensive aspects. Equally, as regards defensive metrics, notoriously difficult at the player level given the collective inherent nature of defending, it seems there may be a volume bias in the calculations.

All that being said, in the crucial area of ball progression Celtic defenders rate highly. We know we need a new goalkeeper with younger, more agile capabilities. Next time I’ll go through the midfielder, wingers and strikers in the spirit of testing the OBVs and also revealing the underlying performance data.