The Engine Room


The Engine Room is subsidiary to a review of season 16/17, and is the fifth chapter. Here I will consider the central midfield. This includes three distinct roles:

  1. Defensive or holding central midfielder – Bitton or Brown, and possibly Kouassi
  2. Box to box central midfielder – Armstrong or McGregor
  3. Classic Number 10 – Rogic

We have Henderson and Christie who could potentially play any of those roles, with Christie being a more natural Number 10 and Henderson a deep lying playmaker. As an aside, we sold Johansen whom we played as either 1. or 3. above for Celtic, but excelled this season for Fulham in position 2.! (see “The Stats Factory: The Changing Role of Stefan Johansen”).

Celtic under Rodgers have often player a 4-3-3 with Brown or Bitton holding and two other midfielders given license to break forward. This emerged in response to Rogic’s long term injury. When the Australian is available, the preferred formation is a 4-2-3-1 that has less rigidity than under Deila. One of the two deeper midfielders will have license to get forward more, and between them they will cover the attacking full backs.


2015/16 was considered by some to be the beginning of the end for Brown. Injury hit and said to lack the dynamism of old, would he survive a 10th season at Celtic under new manager Rodgers? A preseason meeting in London between the leaders of the football team seemed to rejuvenate Brown. Having missed 30 matches to injury the preceding two seasons under Deila, Brown has played more this season than any other outfield player. Out of 59 matches, he completed 90 mins 49 times. He was omitted three times due to suspension, and twice rested during demanding fixture congestion. Brown passed the 400 appearances mark, averaging more than 40 matches a season at Celtic.

I covered his recalibration as a more metronomic influence in “Great Scott 2.0” which I wholeheartedly recommend you read. Considering the work load of the remaining central midfielders, Brown deserves his end of season player of the year nominations through sheer endeavour. Across four competitions, he has played in 90% of all available minutes in a season that will go down as one of Celtic’s greatest.

Whereas Deila favoured two holding midfielders, despite there being no obvious need for such a structure in domestic matches, Rodgers has tweaked the makeup of the central midfield to include more attacking options. To the fore has been the reformation of Armstrong in his preferred central midfield position. I covered this in detail in “The Renaissance Man” which is another highly recommended read! No other player has come close to Brown in completing 90m matches nor overall minutes, and Armstrong is the only other central midfielder to play more than 50% of the total seasons minutes (5310).

The shifting of McGregor from an attacking wide forward to a progressive central midfielder has consolidated a process started by Deila and completed by Rodgers. Remarkably, he has only been missing from the match day 16 twice all season. Once was due to suspension. Whilst not a first choice, his game time has been greatly increased thanks to the injury to Rogic. McGregor is firmly established as a first team squad regular being able to fill a number of positions, even left back as in the Scottish Cup Final. Equally remarkably, his journey to usefulness is documented in another in depth article which cannot be recommended enough “Water Carrying for Galacticos”.

The Israeli international Bitton was favoured by Deila. A technically sound player, he offers perhaps a unique role as a deep lying play maker. Although not prodigiously creative, his passing skills allied to his height and tackling ability, render him the type of central midfielder more usually seen in Spanish or Portuguese football than Scottish. Certainly not forgotten by Rodgers, nevertheless he has had to come to terms with being more a squad player. He has been involved in 51 out of 59 match day squads. Injured for three matches, he has rarely been neglected. His role as the season reached its climax was more of a “special teams” player brought onto stiffen the midfield and close out games.

The Australian Nike Academy graduate Tom Rogic is perhaps the only true trequartista in the squad – a player specialising in the role behind the strikers and in advance of the orthodox central midfielders. A solidly built 6’2” frame renders him more a Riquelme than a Messi (I am not suggesting he is as talented as either, merely passing comment on style o play), but the player with the hint of fantasy to his game was only neglected from the match day squad on one occasion. Missing 17 matches through injury, Rogic struggles for full optimal fitness and completed 90m only eight times (31% of starts). However, he remains a player selected when fit, and whom Rodgers is prepared to reshape his team around. In another ground-breaking article (!) “The Rogic Effect” I showed as far as results and attacking performances are concerned, it makes little difference whether Rogic plays or not. However, we “look” a better team when he plays, or certainly more exciting.

After the five players mentioned above, only Henderson broke 500 mins. Christie was loaned out to Aberdeen from January to May 2017. Kouassi arrived for significant monetary outlay lacking match fitness with the Russian season having completed. Finally, the aforementioned Johansen was sold to Fulham and contributed a fine season for the Craven Cottage team. He completed a mere 59 distracted and ineffectual minutes for Celtic this season.


Although not surprising that Brown leads the defensive standing, Bitton has a very similar set of number. With more intercepts and challenges won than all the centre backs expect Sviatchenko, this is where the action is. Bittion’s style is more circumspect than Brown reflected in less challenges and intercepts engaged in.

Worthy of analysis is Brown’s defensive contribution in Europe. In the qualifying games he averaged 7.17 challenges won per 90m compared to 5.44 in domestic games. He managed to contribute offensively in the qualifiers with one memorable goal against Hapoel Be’er Sheva (the fifth, ultimately the tie winner). He managed 6 shots and created 2 chances in those six matches. In the Group Stages, he shifted to a more disciplined holding role and only contributed 1 shot with 0 chances created or attempted. He led the competition in intercepts – racking up 44 in 6 matches – 7.33 per 90m compared to 4.78 in domestic games. Brown had 0.5 defensive errors per 90m in Europe and 0.29 domestically. It was a very disciplined and determined Brown that played every minute of Celtic’s European campaign.

Bitton only completed 90m twice in Europe. His challenge won rate in Europe was 6.1 per 90m and domestically 5.98 per 90m. His intercept rate in Europe was 6.47 per 90m in Europe and domestically 3.6 per 90m. Bitton made 8 out of his 18 defensive errors in European matches – 1.48 per 90m. In domestic games, his error rate was 0.49 per 90m.

Brown, therefore, raised his game in Europe to a greater extent than Bitton and was less likely to be caught in possession in dangerous areas. Bitton does lead the challenges won success league overall – being the taller player he succeeds with more aerial challenges and I will differentiate next season. He also intercepts 1% more successfully than Brown overall.

Armstrong and McGregor both play similar roles being the central midfielder most likely to attempt to break beyond the striker. Armstrong is the quicker and more robust player and wins more than 1 challenge per 90m more than McGregor and is 12% more successful winning them. The agile Armstrong leads the midfielders in intercept success, with 53% of attempts resulting in Celtic possession. McGregor attempts more intercepts but only reclaims the ball 42% of the time.

Rogics’ defensive data lags the other central midfielders and his role is significantly less defensive than those players. He wins more challenges than McGregor although only succeeds in maintaining possession 27% of the time. Rogic is dispossessed more than any other Celtic player bar one (Dembele) – he loses outright 4.74 challenges per 90m and 7.44 challenges overall. Although physically strong, he can get caught in possession too easily. In Europe, he lost 10 challenges outright at home to Borussia Moenchengladbach, 8 in Barcelona and 7 in Astana averaging 7.13 challenges lost per 90m in Europe overall. This is lower than domestically probably due to the packed defences and lack of space he encounters at home. In a home games against Kilmarnock, he lost 14 challenges and 11 in the Scottish Cup Final against Aberdeen. He did score in both games! Rogic can be robust, winning 6 challenges in the September home game versus The Rangers.

Surprisingly, Brown is only the third most frequent foul offender in the group. McGregor, the smallest and least physical of the central midfielders did get a red card away to Hamilton Academical. A clumsy challenge rather than malicious one, nevertheless the opportunity was taken to dish out the red. McGregor commits half the rate of fouls Armstrong does, and Armstrong commits 36% less fouls than Brown. Rogic despite his advanced position in the pitch puts his six foot plus stocky frame to use and commits more fouls on average than Brown! Bitton is the worst offender but Brown has the highest card accumulation rate at 0.3 per 90m. Which puts paid to the press misconception the referees are lenient on the captain. Brown receives a card for every 5.4 fouls whilst Bitton is pinged every 7.7. Brown’s one red card late on at Ross County after some incredible refereeing to provide the home side the opportunity to score a late penalty, was rescinded.


No surprise that the number 10 leads the way in assists per 90m. Rogic has many assets as an attacking central midfielder including using his height and strength to hold the ball up like a withdrawn target man. He is also a reliable goal scorer as we will see below. He recorded only 7 assists, putting him only joint 8th in the team total assist league table. But per minutes played, he is 6th in the squad for assists per 90m. He recorded 2 assists in one game against Aberdeen in August. Against Motherwell in the League Cup and Kilmarnock at home in September he managed to create 3 chances in the same game. Rogic seems to take a long time to recover match fitness. A large man he struggles to complete 90m. On his return from seventeen matches out injured he managed only 1 assist at 0.2 per 90m to the end of the season. That compared to 0.31 per 90m prior to injury.

The admirable McGregor comes second on the assist league. He plays slightly more of an attacking role compared to Armstrong and attempts to create more chances than him. He is more accurate, his passing is 5% more exact than both Rogic and Armstrong. Arguably McGregor’s greatest asset is his ability to link play with quick accurate short passes, maintaining momentum to the attack. He started the season with a flurry of 3 assists in his first 5 appearances. He did not register an assist in his final 10 appearances of the season. He twice created 2 chances in one match – at home to Motherwell in the League Cup and at home to Inverness Caledonian Thistle in November. Remarkably, in the last 10 matches of the season, when no assists were registered, he created or attempted to create 16 chances. Next season I will capture expected assists, and I suspect in the seasons run in his xA would be more impressive than 0 actual assists (it is not the passers fault is the recipient misses the goal). In 5 matches I recorded xA towards the seasons end, McGregor completed 11 key passes (a pass resulting in a shot at goal) with an xA of 0.75.

Armstrong’s pacey all action style, everything done at speed, may contribute to a relatively low pass success rate. Also, his decision making is something that can be improved evidenced both in pass completion and his defensive error rate of 0.29 per 90m. Armstrong can be caught in possession and give away passes in dangerous central positions. Only 22% of the chances he tries to create are successful, the lowest of the peer group. There is another angle here though which is that Armstrong attempts more high-risk passes (difficult to execute, high reward if successful) than, I think, any other player in the squad. Next season I will attempt to calculate “packing rates” which will show the number of opposition players taken out of the game by a pass and receipt. I will predict Armstrong has the highest packing rate in the squad. Of his 8 assists, 2 came in the memorable 4-3 win at Motherwell in December. On two occasions, he created 2 chances – at home to Lincoln Red Imps and at Ibrox in December. At Dundee in March he gave away 24 passes and attempted to create 5 chances, all unsuccessful. “Unsuccessful” doesn’t do the risk-taking justice and I believe next season I will be armed with the data to show Armstrong’s worth to an even greater extent.

As for the two more defensive midfielders, it is remarkable how once again their performance data is so similar. Brown leads the squad in completed passes per 90m, with Boyata splitting him and Bitton. Seven times this season Brown exceeded 100 passes. The highest I have recorded for any player is the 146 completed passes at home to Kilmarnock in April. His passing success was 94% in Europe, only giving up 2 passes in Moenchengladbach. Bitton (141) and Brown (140) racked up nearly 200 more passes than Albion Rovers in January, although one may query the point of that exercise! The defensive and passing data does tend to highlight that playing both players at the same, time as Deila tended to do, means two very similar players in the same position. Certainly, for domestic games, that seems too conservative. One plus two from McGregor, Armstrong or Rogic seems to provide more pace and creativity.

Possession effectiveness (PEI) measures the proportion of all actions that result in possession being maintained for the team. Passing actions account for around 78% of all actions so clearly heavily influence this number. Rogic would expect to be at the bottom of this mini league given how far forward he plays. He will receive the ball under the most pressure and be playing with less space around him. Brown leads the pack with 90% given his high pass completion and challenge win rate. McGregor’s accurate passing more than makes up for his tendency to lose physical challenges and it is surprising he is a full 6% ahead of Bitton. Bitton seemed to struggle in matches against Aberdeen. Of ten games where his PEI was 70% or less, three were against the Dons. He also had an uncomfortable time at home to Borussia Moenchengladbach – the theme here may be pressing teams. Brown only failed to hit 70% PEI in four matches – all away from home at Hearts (August), Hapoel Be’er Sheva, Hamilton Academical (December) and The Rangers (April).

The creative Armstrong must balance the risk taking with the need on occasion to ensure possession is maintained in central areas where there is a risk of the opposition being able to break on you. Twelve times Armstrong failed to get to 70% PEI including four Champions League games.


If Armstrong looks imprecise in his passing (remember: high risk – high reward), the precision of his finishing this season has been a revelation. Once scoring 11 goals in a season for Dundee United in 2013/14, Armstrong smashed through that personal record. Most eye-catching was his ability to score from outside the box with 7 coming from distance – low probability efforts. He has greatly exceeded his xG of 10.1 with 17 goals. Either he has been fortunate and his long-range shooting will regress to the mean, or has exceptional technical ability in this regard. Given that his finishes have often been low and hard and into the corners, with few deflections, I would predict the latter but we will see next season! Armstrong, once established in the side in October, never went more than four matches without scoring – admirably consistent. Short accuracy of 39% is the same as Rogic who perhaps had a more prominent scoring reputation. Top scorer Dembele’s shot accuracy is 43%.

Rogic has the highest scoring rate at 0.49 goals per 90m, nearly 1 goal every 2 games, the gold standard for most strikers. He takes a lot of shots (4.21 per 90m) and only Dembele and Griffiths attempt more. With 39% accuracy only the main strikers have a better accuracy rate. Scoring 5 goals from outside the box he also exceeded his xG of 9.51 with 12 goals. Rogic is also 3rd in the squad with shots on target per 90m with 1.63. He went 11 matches without a goal with 4 matches prior to injury and 7 after. I suspect he doesn’t much care about that. The Australian playmaker had the distinction of scoring the winning goal in two Cup Finals in the same season. He netted the opener in the League Cup Final and you may not have realised he scored the 90+ minute winner to clinch the Scottish Cup, the Treble and the Invincible Season!

McGregor completes the trio delivering his best goal scoring Celtic season (he once scored 14 goals for a nearly relegated Nott County on loan). He also out performed his xG, managing 1 goal from 7 from outside the area. Less accurate in his shooting, he nevertheless completed a stellar run of 4 goals in the final 7 matches of the season. 8 of his 21 shots on target occurred in this run, compared to 1 shot on target in the previous 12 appearances. He had another run of 4 goals in 5 appearances in November and December. He may, therefore, be a player that goes on scoring streaks.

The Bitton / Brown comparison is fascinating once again with Brown having the edge in scoring potential. Bitton only had 3 attempts at goal from 36 shots from inside the box. 10 of his shots came from a range with a 2% or less chance of scoring. Whilst he has scored the occasional long-range goal, these are low probability efforts. Long shots and the threat of them are a necessary weapon especially against a packed defence. Bitton may not have the balance quite in his favour regarding shot decision making, managing just 1 goal all season.

With Brown, he chooses his moment well. All 4 of his Scotland goals throughout his career have been winning goals. This season he scored 2 key goals for Celtic out of his 3 efforts. The 5th goal against Hapoel Be’er Sheva effectively won the tie and opened the Champions League money box. He also scored a key goal at Dundee in October when a tired Celtic ground out the type of 1-0 win champions must do. In the middle of the Champions League campaign momentum was maintained. Although Bitton attempts twice as many shots as Brown and has twice as many on target, Brown had the higher xG nearly matching his seasons actual total.

As regards other attacking actions, Rogic is the dribble king succeeding in 1.72 per 90m behind only Roberts and Forrest in the squad. He is successful with 60% of his dribbles indicating good decision making around when to run with it. He is also fouled 1.8 time per 90m, almost the same number of time he commits fouls (1.78) – is Rogic the new Chris Sutton?? The other stand out number for Rogic is being dispossessed (losing a challenge outright + miss-controls) 5.56 per 90m, significantly more than the other midfielders. Only Dembele is higher and significantly so (6.87 per 90m). Both players perhaps dwell on the ball a fraction too long on occasion plus both are strong physical players and attract a lot of defensive challenges.

Armstrong completes around half the number of dribbles compared to Rogic but despite this is less easily dispossessed than McGregor. For an ex-winger, McGregor’s dribble rates are perhaps low. When you consider his excellent passing ability is makes sense for him to pick a passing option.

Brown again shades Bitton by these criteria. The tall, less agile Israeli is not perhaps optimally dimensioned for take-ons but does occasionally display excellent footwork and tricks to escape would be tacklers. Brown can still provide drive in his game and three times has succeeded with 3 dribbles in the same game, at home to Lincoln Red Imps, at home to St Johnstone in January and away to Dundee in March. He is the most fouled player in the squad. Drawing 5 fouls in the league opener at Tynecastle was a season high.

Finally, Rogic leads the way in total scoring contribution, averaging 0.78 non-penalty goals and assists per 90m. When you remove Sinclair’s penalties, this puts Rogic 3rd behind the two strikers in the squad. I doubt there is a higher contributing central midfielder than Armstrong in Scotland. And for McGregor to contribute a goal or assist at the rate of 1 every 2 games is evidence of pleasing development in his game.


The captain has made himself vital in Rodgers’ team. Cementing the deeper holding midfield role, comfortable slotting in to make a back three, or shifting left to cover Tierney, Brown is the defensive midfield rock. In European matches, he adjusted his game to try and keep possession, sacrificing his attacking instincts to concentrate on ball retention and ball recycling. He has recalibrated his role successfully but in the ruthless world of professional football, at almost 32, Rodgers must be succession planning.

That succession is unlikely to be bestowed upon Bitton. A player seemingly at odds with the pacier, pressing style championed by the manager, he has nevertheless become a player trusted to come on and help close out games. Defensively secure and strong in the challenge, Bitton’s main attributes are a tidy passing game. However Brown can play this role slightly better. We have not seen enough of Kouassi to see what role he will play.

Attacking threat has been consistently provided by McGregor, Armstrong and Rogic with 36 goals and 22 assists between them. Nicely complimenting the holding midfielders, Armstrong has blossomed as a box to box midfielder with a lethal goal threat.

McGregor has reinvented himself as an attacking central midfielder – a process started under Deila. Adding goals and assists to his game, he needs to develop strength and timing to be stronger in the challenge. His passing ability tends to get him out of trouble however. Will he become understudy to Armstrong, or continue to be a fungible resource capable of solving a number of problems from auxiliary left back to right winger?

Finally, we only have one true number 10. The unlikely proportioned Aussie is destined for cult status after a series of memorable goals (away to Kilmarnock last season, away to Motherwell to win the 4-3 game, a League Cup final goal, and of course the triple crown winner at Hampden in the Scottish Cup Final). Without a direct rival for his position, Celtic change shape if he is unavailable. His injury record and overall conditioning remain a concern but at only 24, he remains a prodigious talent. Cover for the “10” would be wise, however.

Leave a response

Your email address will not be published. Required fields are marked *