The (Partial) Death of Possession Football

A supposedly harrowing week ended with two comprehensives away victories. Success in Stockholm secured European football until year end. Triumph at Ibrox opened up an early SPFL lead, and punctured a seriously inflated Scottish media balloon.

All good for Lennon and his new look side, but attained in a manner the Celtic support have not been accustomed to during the Treble Treble seasons. Simplistically, Celtic had minority possession in both ties, and seemed happy enough with it.

Are we now seeing Lennonball realised?

Possession? Meh.

The Rodger’s era was characterised by ever more patient and possession-based football. Having 70%+ possession was not unusual – it happened 27 times in 63 matches last season.

The principles are simple. The opposition cannot score if they do not have the ball. Wear teams down by attrition, seek positional overloads especially in wide areas, wait for errors or failures in concentration and strike. Increasingly, within the Rodgers reign, Celtic would not take a shot until optimally placed to do so. It all makes sense and is the natural progression for a Guardiola acolyte.

Football does not stand still, and whilst genuinely new thinking is rare, stylistic mores do persist for a while but evolve.

Think back to 2003 and the Champions League final between AC Milan and Juventus – a 0-0 draw with all the hallmarks of classic Italian defensive excellence. Leap forward 8 years to 2011 and Barcelona passed Manchester United to defeat in arguably the high point for Guardiola’s “tiki taka” football. Another 8 years forward to 2019 and Liverpool and Tottenham Hotspur attempted to out press and out “pack” each other in a largely formless mess of a final. Possession was not the primary goal, catching opponents short with quick forward passes was the norm.

There is a “rock-paper-scissors” nature to football tactics. Formations and styles come and go – three at the back is suddenly back from the 1990s. Pressing is more prevalent as players are generally fitter and more athletic. Football is adapting to advances in sports science in this regard.

Rodgers was very much of the Guardiola school but was big on pressing in his first season in particular. This faded over three seasons I believe because of 1. Player fatigue and injury and 2. Teams being wise to it and simply “sitting in” leading to ever more “sterile” possession.

Lennon is clearly a more pragmatic coach as regards playing styles. And it is this that we will assess.

Possession, in and of itself, is not a great predictor of success but we are here to talk about playing style.

We can see that Rodger’s Celtic were getting towards 70% possession as an AVERAGE. It was 60% under Deila. It is falling under Lennon. I would expect Celtic to still be possession-dominant against the majority of Scottish opposition. But in general Celtic are happy to cede some possession if it turns the opposition.

Passing, Schmassing

Again, passing stats won’t tell you who should win but tell you a bit about team style.

Pushing up towards 600 passes per 90m, Celtic under Lennon are now just under 500.

And not only are there significantly less passes, they are less accurate too.

This is entirely consistent with Lennon’s stated aim to make Celtic go a bit more direct at times, and be less predictable in their build up.

A note on the Deila era as a sidebar. I would characterise this as being a bit of everything but all done at much lower pace than more recent squads. Players like Johansen, Bitton, Brown, Commons, Stokes, and the array of striker misfits did not have the pace the current side has. Celtic under Deila simply played slower football. Under Rodgers it was possession orientated, but also quick – faster passing and faster transitions.

Anyway…….

Passing less successfully sounds like a negative but we need to explore more data to assess the effectiveness of this approach.

Shooting

For some, the only stat that matters are goals. And I’d be a hypocrite if I didn’t pander to the old school here given my age! Goals are certainly up – 2.69 per game is just under a goal per game and better than last season (1.75) and more than in the Invincible Treble season (2.34).

But as hopefully all readers know by now, that does not speak to chance quality nor repeatability.

Let’s start with shots – are Celtic taking more and what of their quality?

Total shots are up slightly under Lennon and from 16.7 to 19 over last season. Shots inside the box are steady at around 10 per game. Shots outside the box has increased to be higher than any time under Rodgers. Lennon has spoken of taking more shots. Taking long shots is, however, not a great long-term strategy as the further away you are from goal the less chance you have of the ball going in (there are other variables of course).

So, all good that Celtic are taking more shots, but even better if more of them were from inside the opposition box.

Celtic’s Shot Accuracy is usually around the 40% mark – i.e. 40% of all shots are on target.

The stand out season was the 16/17 Invincible treble season. Despite a very average 37% shot accuracy, Celtic scored with an impressive 37% of all on target shots and 14% of all shots taken. That was the season when Celtic scored over 20 goals from outside the box, and many of them were game changing.

This season, Celtic are converting an impressive 41% of on target shots, and a 5-year high 17% of all shots. My question is – is this sustainable? It looks more extreme than the Invincible season (perception = “everything went in for Celtic”). We know Celtic struggled to sustain that in the subsequent two seasons. So, permanent uptick, or regression to the mean to follow?

Well, the good news is that expected goals are up as well.

In addition, we know that Celtic under performed against xG over the last two seasons

As you can see above, under Rodgers’ last two seasons, removing penalties, Celtic scored less goals than expected. Under Lennon that trend has reversed. Expected goals from non-penalty shots are remained pretty static over the last three seasons (down slightly to 2.037 this) but Lennon is seeing a scoring rate of 2.69 per game – nearly one per match better than last season.

So, is this just luck, and Lennon benefitting from the football Gods allowing Celtic’s shot performance to move back to the mean? Or is there something else at play here?

Chances

The good news for Lennon is that the quality of chances created has increased, but very slightly. In all of Rodgers’ seasons, Celtic averaged 15 chances per match. Under Lennon this has gone up slightly to 16.

Expected Assists (xA), a measure of chance quality, has increased from just under 2 per 90m to 2.023. Again, a modest but encouraging increase.

Other indicators, such as total number of touches in the box, are flat. This was around 25 under Rodgers, and remains so under Lennon.

Celtic aren’t creating more chances from passes into the Danger Zone. This has actually decreased from over 6 to 4.22 under the current gaffer.

And despite the obvious management desire to get more crosses into the box, predictably, this isn’t having a positive impact. Cross attempts have gone up from 16.86 per game to 17.62 (not as much an increase as you may have imagined), but cross success is down with only 1.69 crosses being successful per game.

Lennonball

How to make sense of all this? Goals up, shots up, but evidence of overperforming to xG rather than underperforming, and with minimal improvement in chance quality or volume.

The problem for Rodgers became the predictability of the play. His tactical weakness was an unwillingness to change approach. In retrospect, gaining two draws against Guardiola’s Manchester City in his early Champions League career at Celtic was the worst thing that could have happened. This relative success emboldened Rodgers to believe he could go toe to toe with European elite. Indeed, Celtic’s best performance of last season, a 2-1 win over an excellent RB Leipzig side, is an example of doing just that. It was also an exception.

There were the 3, 5, 7 goal maulings in Europe. And SPFL matches were becoming increasingly sterile. Teams knew how Celtic would play. And although individual quality would usually tell, there were an increasing number of blanks (9 in the SPFL) as teams could set up to counter exactly what they expected.

Not so under Lennon. From 703 completed passes and 89% pass completion in destroying St Johnstone, to 205 passes @ 69% completion at Ibrox. Possession football is not discarded but more tactical flexibility is in evidence.

The way Celtic allowed The Rangers almost total ball dominance on Sunday (no Celtic player completed more than 9 open play passes in the second half – and that was Ntcham who only played 24 minutes) is instructive.

Celtic then looked to play long passes to the attacking talent knowing that if they could get Johnston, Christie, Forrest and Edouard in play with minimal defensive shape between them and the goal, then chances would follow. Despite 38% possession, Celtic created the better chances and had the higher xG.

From 70 minutes in, the home side had exhausted their patient passing game, created very little, and Celtic threatened to run riot on the break.

And it is providing the environment where chances can be created in transition rather than at the end of long bouts of possession against a set defence, that is the major differential under Lennon. If you can get a shot off against a defence that is under manned and disorganised, you have a better chance of scoring than if you took the same shot from the same place against an organised defence. The expected goals models would tell you that.

Consider this:

22% of chances Celtic create are from a “fast break” scenario – that is, against a relatively disorganised and stretched defensive alignment. Last season that was 13%

Not surprisingly, the xG from Fast Breaks has increased from 0.601 to 0.994 per 90m this season.

Celtic are 65% more likely to score from a Fast Break than last season. Remember – fast break chances are generally higher xG values.

Lennon, therefore, has made Celtic less predictable. He has also created a side that can break against other teams more effectively than over the last few seasons. This will often create better quality chances. Fast, skilful player with good decision-making ability are crucial here. Needless to say, Christie is central to this approach both defensively and offensively (another article).

A Warning

Increasingly, under Rodgers, Celtic conceded less and less. More possession, the opposition cannot score.

Under Lennon, there is more risk in the game. The “Basketballification” of football if you will. As with Liverpool and Tottenham Hotspur, the coaches’ attitudes were – I back my defenders to be better than your attackers. I won’t rely on always having 9 behind the ball. The opposition are completing more passes, but with less overall accuracy – Celtic’s press is forcing more longer passes.

xG for the opposition is going up as well. From 0.845 to 0.945 per 90m.

The opposition are, conversely, having slightly less shots and less touches in the Celtic box. They are creating less chances but their xA is going up (from 0.711 to 0.802).

See a pattern here?

You reap what you sow with this approach.

Whereas under Rodgers, 13% of opposition chances were from fast breaks, it has now doubled to 26%. Celtic are offering up no more chances than previously, but increasingly they will be from better probability positions, on the break.

You may recall that Celtic found Lennons’ Hibernian sides quite easy to break against.

Overall in matches, there are less challenges (Ibrox aside) and passes attempted per tackle (PAPT) is down from 13.24 to 7.71 from opposition sides. Another indicator of Celtic’s pressing improvement.

The danger is once you beat the press, you generally create better chances.

Summary

Lennon’s Celtic are less predictable. I am heartened that, especially in tough away fixtures, creative solutions will be found in defensive shape and organisation.

Celtic are simultaneously benefitting from

1. Correction to underperforming against xG. Or “luck”. Last season 4% of shots outside the box resulted in goal – this is normal. This season 11% of shots outside the box have resulted in goals. I’d posit this is not sustainable.
2. Forcing better environments for chances. More chances as a result of creating fast break scenarios. Either through pressing or varying passing style.

It is imperative the defence becomes settled and organised as under Lennon, they will more often have to defend without midfield cover. It is also imperative attacking players compliment the system – you can see why Johnston is preferred, and why Elyounoussi has been recruited.

The opposition, however, will occasionally get better chances too. Against Cluj, for example, a team set out to play an even more extreme template of the Ibrox performance, you can see what will happen if the opposition get the breaks and win the second ball in transition.

It’ll be a bumpy ride!