On-ball observations Part 2: Celtic's offensive OBV numbers crunched

I’ll concentrate on ball progression and shooting with a defensive twist to include the central midfielders (and focus on Callum McGregor).

Defensive Action

I will restrict this view of the defensive action OBV to the central midfielders only. The “midfielders” cover holding, box-to-box and number 10s).

The first aspect that strikes me is that the overall height of the bars in 2018-19 is greater than any of the subsequent seasons. This makes me wonder: did Brendan Rodgers have better players then or were they more effectively deployed defensively?

A midfield of Scott Brown, younger Callum McGregor, Ryan Christie and Tom Rogic probably had more “legs” than an older McGregor, Reo Hatate and Paolo Bernardo. Matt O’Riley is the obvious exception.

But overall defensive action OBV across the group did decrease from an average per player of 0.04 in 2018-19 to 0.02 in 2019-20, 0.01 in 2020-21, 0.02 in 2021-22, 2022-23, and 2023-34. Small margins, but could we see Rodgers implement greater defensive solidity into the midfield for the coming season?

Losing O’Riley would be a blow as was losing Aaron Mooy last season. Both are great examples of all-round midfielders. Not necessarily the top rated at any one discipline but competent in all.

Other notable eye-catchers is the decline of Brown. And if you want to see what “downed tools” look like in data form, see Olivier Ntcham’s 2020-21 contribution!

McGregor dipped in 2020-21 also as did most. But if we isolate his numbers, it is incumbent upon me to highlight the downward trend.

His numbers perked up under Neil Lennon as he adapted to his new role and had to cover an ageing Brown. But otherwise, the trends are all one way. Recruitment and strategic thinkers take note.

There is a “but” and I’ll come to that.

Ball Progression

The ball progression view is a composite of the dribble and carry OBV and the passing OBV.

A bit messy this one but let me try and break it down.

If O’Riley leaves, Celtic are short of passers who get the ball into good areas. Note David Turnbull, Mooy and O’Riley are in the middle to bottom left. Note also further evidence of McGregor’s decline in this facet.

Reo Hatate’s passing is a bit Marmite as we know, but he does bring ball-carrying capabilities otherwise sparse in the current midfield.

Some more athletic “do everything” types like O’Riley and Ryan Christie would be near the top of my wish list.

Statsbomb really hates on Bernardo. His passing OBV is in the extreme negative. I think this is synonymous with his low pass packing data. He does not provide enough line-breaking passes. As I have said before, that is a show-stopping void that would preclude me from supporting a permanent move for him other than at a bargain price.

Overall, there is a clear ball progression deficit looming for this squad in the midfield area.

Turning the attention to the wingers.

On the face of it, the current crop does well by these metrics. However, I mentioned a “but” above and this view highlights what I am about to say.

OBV seems to “reward” volume accumulators. That is, the basic methodology adds or subtracts an xG value to each on ball action. The more actions, the more numbers get added (and subtracted).

On the Huddle Breakdown we often talk about “volume” players such as Jota and Luis Palma. Their numbers look good through simple volume. You need to supplement this view with “quality” indicators such as average xA per chance or average xG per shot to get the full picture.

Palma is an outlier in this view, as was Turnbull on the ball progression view. Both are accumulators of many actions but do not confuse this with consistent quality.

Nicolas Kuhn also features well within the virtuous top left corner. Which further supports the point I am making.

Daizen Maeda is consistently near the low end of both scores, and he is not a volume accumulator. But he does score well for average xA per chance created for example. Quality over quantity.

Ideally, you want both and Jota fits into that category. His two seasons are pleasingly close to each other. A perfect confluence of volume, quality and consistency.

More of that, please.

Shooting

Statsbomb reduces goal threat down to a single shot OBV. Given it is an xG model behind the metric, one might expect this to be the closest to the eye test but let us see.

Starting with the wide forwards.

Again, the striking aspect of this is that the average scores in 2018-19 and subsequent seasons are much higher than under Ange Postecoglou and now Rodgers.

A downgrade in quality? It certainly isn’t coherence of system as we know under the Australian the right players played in the right positions and were drilled to within an inch.

The players recruited by Postecoglou may not be as effective under a different manager, of course, and we’ve seen some evidence of that. And last season was probably a recent-history low for wing quality.

At the moment Celtic does not have a wide player with the “second striker” instincts of a Scott Sinclair or Moi Elyounoussi, or even peak James Forrest.

That goal threat from wide areas is another clear area to improve.

What of the strikers?

The curious case of Kyogo Furuhashi. The highest shot OBV in the data set, and the lowest. That despite his 2022-23 campaign yielding over 30 goals.

Kyogo also “suffers” from the inbuild model bias for volume over quality. You only need to see Leigh Griffiths’s comparative data to grasp that point.

And we know that Kyogo was not best utilised to maximise his strengths last season.

Once again though, compared to prior seasons, the striking area appears to need a refresh, either in terms of upgrade in personnel or better adoption of the system. Or both.

Summary

I have focussed on one set of metrics which is by definition, reductive, and does not pretend to give you an easy “who is the best” answer. We need multiple dimensions of volume and quality to form that view.

OBV is a welcome addition to the advanced metric canon. It has limitations though. Rewarding volume over quality appears to be the major one.

Despite this, what we can see is that by a range of metrics the outcome final scores have gone down for this Celtic squad versus prior seasons.

The need for Rodgers to get quality over quantity into the squad is made despite these metrics being biased in the exact opposite direction!

This is another reason I love tooling about with these stats.