A short (lol) piece on assessing the risk versus reward dynamic amongst the central midfielders.
Many supporters appreciated that “giving the ball away” is not always a negative per se, but the corollary to attempting high risk or creative passes. Losing the ball is worth it sometimes if there are chances being created and opponents being taken out of the game, to some extent.
There are data points that help us understand this better. (Of course!)
Deep Progression
This means getting the ball forward from deeper positions in midfield.
The risks associated with not making a pass from a central area are far higher than losing the ball in the opposition box. It is highly likely that if you lose the ball in central midfield, your attackers and number eights are “out the game” – that is, not goal side of the ball.
The upside is that with a set zonal defence, a high pack passing score indicates you can thread passes through opposition lines removing their players from the game (they are now the “wrong” side of the ball).
I measure this using pack pass score – opponents taken out the game and pack turnover score – your own teammates taken out the game by giving the ball away.
The “safest” passers here are also the most creative. Aaron Mooy and Callum McGregor have the highest pack pass scores and the lowest pack turnover scores.
As regards McGregor, that is as we would wish it. He is the deepest midfielder and we’d expect he’d have the lowest turnover rates. The fact he has such a high pack passing score is evidence of part of his “special” skill set.
It may be a coincidence the two older players have the best metrics here – decision making should improve with age you would think.
Reo Hatate has the next highest pack passing score but also easily the worst pack turnover score. Only Alex Bernabei’s is lower in the squad. However, for those who would say that the deal with Hatate is “high risk / high reward” on this evidence, that is the quadrant he would align to.
I would caution a bit with Matt O’Riley’s numbers as there is a bit of noise due to him playing 12 matches in McGregor’s position. However, his pack passing score last season was 48.91 so it has gone down this season. David Turnbull’s is down from 52.78 so regressed even more. Being mainly a substitute may impact this albeit we know the subs under the five-sub rule generally experience a data boost dividend.
The judgement to be made is “how many turnovers are too many”?
Final Third
Let’s look further forward.
Final third losses are where a player loses possession in the final third of the pitch. This may be due to an incomplete pass or cross; being tackled; miss controlling the ball or committing a foul.
I have crossed this with expected assists (xA) from open play per 90 minutes. Note this is from open play to remove the (mainly) Turnbull noise of chances created from set pieces that others e.g. Hatate do not generally take.
I think we allow McGregor to be the outlier here given his deeper role in the team and therefore he has less propensity to be involved in last third actions.
Of the more attacking players, as mentioned above, the upside to Turnbull’s substitute role is that against tiring opposition there is a dividend in chance creation and goal threat. Last season Turnbull’s xA from open play was 0.28! Turnbull also gives the ball away the most so is firmly in the “high risk / high reward” quad.
O’Riley’s xA is a little depressed due to his stint in McGregor’s role (xA was 0.37 last season) but he and Mooy trend very similarly.
Hatate posted 0.18 xA last season so 0.25 this IS an improvement. He is a young player experience wise so that is encouraging albeit from a low base for a Celtic creative player. He also gives the ball away in the final third less that the other attacking midfielders. So, in the final third compared to his direct peers, he is more low risk / low reward.
Ball Security
Back to basics now.
Ball security is my attempt to highlight midfielder safety with the ball. As mentioned above, the cost of transitions in central midfielder can be high.
To plot this, I use basic open play pass completion and defensive action success rate (DASR). The latter is an aggregated metric for measuring success in defensive duels and other defensive actions.
Again, it is reassuring McGregor is in the low-risk category with the highest pass completion and highest DASR rate.
Of the more attacking midfielders, Turnbull is the next most steady passing but has the lowest DASR of this group.
Again, O’Riley and Mooy are trending very similarly which is odd given stylistically they seem quite different players.
Hatate is to the left meaning he has the lowest open play pass completion and second lowest DASR.
Summary
I hope this study emphasises the extent to which McGregor has adapted to the pivot role whilst still offering upside risk to forward passing.
Overall, Turnbull profiles more like a number ten than a number eight. Which is a slight issue as Celtic don’t really play with a number ten!
O’Riley and Mooy profile very similarly across the board but we have the complexity of O’Riley’s numbers being built whilst playing McGregor’s position for 12 matches. Overall, I’d characterise the pair as having a balance between risk and reward in their play.
The most interesting one is Hatate. He seems more effective in deeper areas given his good and improving pack passing data. As an add on, he scores well on ball progression with the ball at feet also. The downside is he gives the ball away (turnovers and basic passing metrics) more than the others and often in dangerous areas. His final third attacking data shows he has by far the lowest (but improving) chance creation stats – so there is little evidence of final third “reward”.
Further follow up work for me would be to assess whether O’Riley and Hatate have improved over last season (on this evidence it would be “no” and “yes” but context etc).
Some will reduce this to a binary point scoring exercise. If that is your thing, I doubt you’ve read this and definitely won’t have made it this far!
For what it is worth my personal view is that the central midfield area still presents the manager with one of his biggest challenges versus what a Champions League set up would look like.
Ryan Malone says
The eye test for Hatate for me is that he seems much quicker than the rest of the midfield and moves play around an awful lot more, stretching the opposition. Not sure how that can be accounted for in numbers. Saying that additional speed you hope would lead to more successful pressing stats.
A bit like how Maeda tires out his opposite number ai think Hatate brings some ‘choas’ to the middle that the others don’t. It ‘feels’ like he bridges the best, and quickest, between McGregor and the final third, and carries the ball with more intent and pace while the others are more pass masters, which helps add a different dimension, again not sure how that would be put in numbers.
I would say opposite to Starfelt who’s balance and gait is uncomfortable on the eye and the heart, Hatate always seems balances and glides around the pitch which probably benefits the eye test.
celticbynumbers@btinternet.com says
Ryan we can all rationalise but it is what it is. I have no end of metrics that would provide a platform for “risk taking” and “creativity” to shine and in few does Hatate come out on top of his peers. I think he is great to watch but as i’ve said before i measure all the other “stuff” too.
Interesting you mention Starfelt and both he and Hatate defy the data in terms of the managers (and many of the fans) perceptions. In my years doing this, that is quite rare.
Damian says
If O’Reily played 12 matches in McGregor’s position should we not expect that to have strengthened his score, given that – as I think you were saying with McGregor – he was positioned deeper in those games and so probably in a better position to score better by these metrics?
I’ve never really warmed to Turnbull. He strikes me both as quite a stupid individual and quite a stupid footballer. I may have been swayed by an interview with fan media folks at the end of the league loss season where he was asked (in an admittedly smart-arse-ish way) about taking shots with low XG. But not only did he not recognise the term ‘XG’, he didn’t even seem to understand the logic of the concept when it was explained to him. This may to speak to his experience working under Lennon to that point, but I keep watching him with those specs on and he always strikes me as a poor decision maker (he’ll flick the ball over opponents and do show off nonsense when it would be simpler and more productive to play a simple pass etc.).
I thought your comments on the podcast about Hatate a few weeks ago were brave and interesting. But I do rather subscribe to the ‘risk/reward’ gamble theory with him. It also strikes me that he best embodies exactly what the manager wants. He’s had lots of opportunities to drop him and never has. The style of play seems to revolve around playing the dangerous pass as close to instantly as possible, then create a volume of opportunities. At any level, these chances will not come off more often than not. But, I very much respect your data too.
Peter says
Another excellent analysis Alan, thank you.
I would describe the Hatate situation as Organised Chaos, I think this is the coaches accepting the risk and reward scenario….
Do you know if the Celtic back room guys do this type of analysis ??