We Need To Talk About Leigh

C:\Users\Alan\Documents\Football\Celtic Stats Analysis\Images\griff and dembele.png

The received wisdom for many years is that teams need four good strikers to choose from. Celtic have had two. This apparent folly was called out by fate when both Dembele and Griffiths were injured at the same time in January. An invincible treble suggests the squad coped – will it again? In the final player review of the 2016/17 season, I look at the two strikers.


With Ciftci failing to rack up enough minutes to constitute half one game, and young Aitchison given 21 minutes, the story of Celtic’s strikers is the story of a much-admired young Frenchman, and a controversial Scottish player who, thankfully, scores enough goals to keep the impact of his occasional behavioural issues at bay.

I put Dembele and Griffiths head-to-head in “A Striking Success” (it’s really good – you should definitely read it!) and concluded “With Rodgers’ iron words of warning, delivered in a smooth velvet glove, regarding what it will take for him to reach his potential, will Griffiths settle for this role in the team or rise to Rodgers’ challenge and reclaim his place by upping his performance?”. Griffiths was struggling to recover his fitness. Between 10th Sept 2016 and 16th April 2017, he missed 16 matches in batches of 3 or 4. In February, Rodgers was not happy with his efforts to recover full fitness. Although not explicit, the inference was that Griffiths lacked conditioning. This, in turn, inevitably, led to speculation around his personal habits and their impacts on his ability to maintain the fitness levels Rodgers demanded.

Dembele was never omitted from the match day squad when fit and only unused once when available. This was in the Scottish Cup Final, when Rodgers trusted his charges to defeat a pressed-out Aberdeen without the need to call on his recently returned young striker. The 40-goal striker from 2015/16 sat on the bench 15 times, unused.

At the time of my previous article, Dembele was in imperious form. Between 5th February and 5th March, he scored 12 goals. 6 goals were scored in 90 elapsed minutes at St Johnstone and at home to Inverness Caley Thistle. Additionally, as my article showed, he was highly effective compared to Griffiths in the high-profile games (i.e. in Europe, in Cup Semi Finals and Finals, and against the top four in the league). He lacked the overall scoring contribution of Griffiths, but at 20 years old was already proving himself a big game player with a winning goal in the League Cup Semi Final, a winning penalty in the last minute against FC Astana, another penalty in the League Cup Final and 5 goals (including penalties) in Champions League.

Four months have since passed, which is an eternity in football. With Dembele ending the season injured, my confirmation bias takes an absolute pasting as we revisit the great striker debate. Which is fine. It is one of the main reasons I started to do this blog – to test the evidence before my eyes and try and provide fact based context to Celtic performances.

Goals, Goal, Goals

We’ll start at the pointy end, where it matters. Dembele scored 32 goals overall, including 8 penalties. I discount penalties from further analysis as the probability of scoring a penalty is the same irrespective of the taker, at around 78%. Therefore, I focus on Non-Penalty Goals (NPG) as including penalties artificially inflates perceptions of finishing ability. That being said, Dembele just out scored a 20-year-old Brian McClair who bagged 31 in 1984/85 in his debut season. Even Larsson as a relatively seasoned 26-year-old “only” scored 19 in his debut season. Without penalties, Dembele still scored more on his debut season than Sutton, Vennegor of Hesselink, Hooper and McDonald. By those measures, Dembele has had a fantastically successful debut season.

Dembele rattled in 161 shots over the season, and hit the target with 40% of them, the equal best in the squad. Managing 1.71 shots on target per 90m, his 24 non-penalty goals were scored at an impressive 0.64 per 90m – well above the bench mark 1 goal every 2 games for a striker. His xG was 19.91 and so 24 goals represents an uplift on the expected, in line with most of the squad.

After scoring 40 goals last season, Griffiths is not the first to suffer from the capricious nature of professional football by playing the backup role to the hugely talented Frenchman. Whilst clearly not having the game time to rack up the sheer quantity of goals, nevertheless Griffiths outscored the rest of the squad with a rate of 0.7 goals per 90m.

*Note: I have not adjusted the 14/15 nor 15/16 data to remove penalties – this would reduce the overall accuracy %

Last season was Griffiths’ least productive of the last three. He reached the Larsson-esque milestone of 40 goals overall last season, 36 non-penalties. His shots per 90m was almost 1 per game higher than this season. Despite taking so many shots, he achieved an accuracy of 48% which is extremely high over a full season. The accuracy rate fell this season although joint best in the squad, with Dembele. It is equally remarkable that Griffiths maintains such accuracy given the location of the shots he takes.

Shot Location

We have seen that Dembele has the better shot accuracy. To understand this better, we need to look at shot location.

45% of Griffiths shots are attempted from outside the box. The average % chance of scoring from such a shot is around 3% or less. More sophisticated xG models will consider the number of defenders in the way, the height and pace of the pass, precise distance, and angle. But generally taking shots outside the box has a very low probability of success. Griffiths significantly out performs that “average”, scoring 6 goals (or 12% all attempts) outside the box. For comparison, Armstrong hit 19 shots on target from 64 attempts (30%) and scored 7 goals (just below Griffiths at 11% conversion of all shots outside the box).

Dembele scored only 1 goal outside the box, and may reflect that 38 fruitless attempts and 24% of all efforts are numbers he may wish to reduce next season. His conversion rate of 2% is around the industry average, suggesting he is no more than competent when attempting from long range.

Where Dembele comes alive as a striker is in the box, in a central position, where you want your team to be having efforts from. Shot quality is a key learning from xG models as, empirically, the case for ensuring efforts are taken from a position on the field with the best chance of scoring is clear. In a low scoring sport, and where you may get 5 shots or less on target, you need to take those shots from positions on the field you may actually score!

Dembele scored 21 of his 24 non-penalty goals from a central position inside the box – that is – positions parallel to the goal and no further out that the 18-yard line. 68% of all his efforts inside the box where from similar positions. He attempted 32% of his in-box efforts from positions wider than the breadth of the goals, but only scored 2 goals from those positions. Overall, he scored with 25% of all efforts from central positions inside the box. This beats my crude xG model by almost double – that is, the % of shots from central positions in the box resulting in goals is 13%.

C:\Users\Alan\Documents\Football\Celtic Stats Analysis\Images\dembele.JPG

Griffiths attempted shots wherever he finds himself in the box, with only 52% coming from a central position. Compared to Dembele, Griffiths managed only 6 goals from central in box positions (Dembele 21) and only converted 18% of all chances central in the box (also beating the xG model but not by as much). Griffiths really is an unusual striker. His unpredictability is probably an asset – for example his first time shot at Ibrox to make it 2-0 was hit from an unpromising position wide left but was hit with such power, and so early, the ‘keeper would have been fractionally taken by surprise. Against lesser teams this shot strategy reaps benefit, but against better organised defences and higher level ‘keepers, the more such teams will be happy to let Griffiths shoot from the positions he does.

Griffiths had 2 shots all season from Zone 1 – that is, in front of the goal no more than 6 yards out – gold territory for an attempt. He scored 1 goal from this position. Dembele managed 10 efforts from the danger zone, scoring 6 times.

C:\Users\Alan\Documents\Football\Celtic Stats Analysis\Images\griff goal v hearts#2.JPG

Many of you will be aware of the buzz surrounding Kylian Mbappé of Monaco. The 18-year-old striker scored 27 goals for the free-flowing French champions. He is rumoured to be wanted by all the richest clubs in the world and is the hottest young striker on the planet. Mbappé attempts 96% of his shots from inside the box and he finds the target with 66% of them. Such ruthless efficiency also displays clear headed decision making around shot selection. He only requires 3.36 shots per goal. Dembele achieved a goal every 6.71 and Griffiths 7.25.

Both players, but especially Dembele who aspires to play at the highest possible levels, can take lessons from the younger Mbappé. Dembele is highly effective when taking shots from the higher probability positions and his focus should be on finding space within them. For Griffiths, I suspect there will always be a sense of he is what he does. His maverick style has been hugely successful for Celtic and he is by far the most prolific striker in Scotland. But hopefully Rodgers has a culture of continuous improvement and he can surely make better shot decisions on occasion. Does he want his productivity to turn heads in Scotland only, or across Europe?

Shot Conversion

Two very different styles of striker, with different approaches to finding the goal.

C:\Users\Alan\AppData\Local\Microsoft\Windows\INetCache\Content.Word\s conv rate all.jpg

C:\Users\Alan\AppData\Local\Microsoft\Windows\INetCache\Content.Word\s conv rate on.jpg

Dembele just edges the all shot conversion metric by 1%. Since Dembele takes the overwhelming number of his shots from in the box and centrally, it is not a surprise he scores with 38% of all shots on target compared to 35% for Griffiths.


Dembele is more involved in games, with 44 action events per game compared to Griffiths 37. He completes more passes and is significantly more accurate with them (by 9%). Griffiths has the lowest pass completion rate in the squad.

But Griffiths is something of a phenomenon, as we have seen with his unusual shot profile. In a squad containing Roberts, Rogic, Armstrong, Forrest, it is he who leads the team with 0.66 assists per 90m. His 15 assists are behind Roberts (18) and Forrest (16) but as he played less time than them, his rate per minute is much higher.

An aside, I am quite generous with assists. I follow the Opta definition for a “Fantasy Assist” and award them where players are fouled for free kicks that are then scored directly from; for players taken down for penalties even if they score the penalty, for example. This may differ from “official” statistics.

The fact he often takes corners and free kicks, enhances Griffiths’ assist rate. Towards the end of the season he formed a particularly dangerous partnership with Boyata. In one game against Aberdeen 4 clear chances and 1 goal resulted from the Griffiths corner and Boyata header routine. This combination also saw one great chance in the Scottish Cup Final missed. He managed 4 assists in 145 minutes of action at home to Hapoel Be’er Sheva and St Johnstone away early in the season. With Dembele injured, Griffiths registered an assist in 5 consecutive matches towards the end of the season, and created 9 chances.

He attempts to create 2.2 chances per 90m and manages to create almost 1 clear chance every 90m. Allied to his low pass success rate, Griffiths is likely a founding member of the “If you don’t buy a ticket…”club.

Dembele tends to stay in more central positions than Griffiths, and does not tend to take set plays. Nevertheless 0.35 assists per 90m is higher than Sinclair and all the central midfielders. His assists have been accumulated at a steady and consistent rate throughout the season. He managed two in one match once, versus Motherwell, being brought down for a penalty and providing a pass for Forrest. A cameo showcase of his abilities was against FC Astana at home, when he won the ball with strength in a dangerous area and used skill and tricks to dribble into the box, drawing the foul and a penalty in the final minute. His resulting penalty goal was his first for the club and sealed progress against the highest stakes. Strength, power, skill, composure, confidence. And although I discount penalties from the goal scoring analysis, his penalty style is revealing. When the referees whistle blows, Dembele does not react immediately. He takes the penalty when HE is ready, disrupting the goal ‘keepers’ rhythm. It is tellingly thoughtful and strategic thinking. His one blemish in 9 attempts from the spot came at the Nou Camp at 0-1 down where, for once, at the pinnacle of the game, nerves seemed to afflict him.

C:\Users\Alan\Documents\Football\Celtic Stats Analysis\Images\dembele2.JPG

Dembele does like to pick up the ball deep and drive at defenders, something he attempted 83 times over the season. It remains an area for development, as although successful 0.96 times per 90m, and drawing 41 fouls, his dribble success lags significantly behind the other attackers (he is successful with 43%).

Griffiths does not attempt as many dribbles, and is slightly more effective completing them (48%).

The young Frenchman is Celtic’s only traditional “target man”. As such he is the long ball option. An effective style he has adopted in that when involved in an aerial challenge, he rarely jumps. Rather he positions his body to where the ball is likely to land using his strength to maintain the field position. It is then usually easier for his to control the ball, being rooted on the ground rather than off balance in the air. Given Celtic’s lack of long ball options (Rogic, when fit, or Lustig wide right) for aerial balls, he usually attracts a lot of defensive attention Consequently, Dembele has the highest dispossession rate in the squad – losing overall possession 6.87 times per 90m. This is due to combination of the number of duels he is involved plus his propensity to take players on in crowded areas.

Griffiths despite being the smaller, slighter man, has developed his own knack of being able to time his leap to win headers against far larger defenders. It is then a lottery as to where the resulting flick or header goes, partially explaining the low pass completion rate. But both players are capable of performing as the lone striker albeit suffer from high dispossession rates for different reasons.

“Go 4-4-2 and just play them both”

There will forever be a strand of British football fan that believes all ills are fixable by going 4-4-2 with a big man and little man up front. The Championship Manager armchair fan has this belief system crafted from watching the years of British dominance of European competitions in the 1970s and 1980s.

Satire aside, given the different styles of the two players, wouldn’t this work? Dembele is a strong target player, likes to play the width of the box and holds the ball up well. Griffiths is the quick, unpredictable deeper striker, dangerous from any angle or distance. Surely they wouldn’t take the same space or make the same runs and could form an understanding? Football, of course, is not as simple as that with eight other outfield players to fit into a coherent system.

Rodgers started the season believing that he could make the team work with both in the starting line-up. They started together in four of the opening five matches of the season. Griffiths started wide right away to Lincoln Red Imps and was as unproductive as any other player that awful day. Otherwise, with Griffiths the slightly deeper forward, they played together at home to Lincoln Red Imps, away in FC Astana and away to Hearts. Although appearing together occasionally as one would come on as a substitute, with Griffiths generally filing the “10” position, they only started two more games together – at home to Inverness Caley Thistle in November, and at home to Hamilton Academical in December. Both have suffered short periods of injury and so this may also have contributed.

Overall, my estimate is that they have been on the pitch for 661 minutes at the same time. During those periods, they have combined for 4 goals as follows:

  1. H Hapoel Be’er Sheva – a Griffiths corner headed in by Dembele
  2. N League Cup Semi Final v The Rangers, in the last-minute Griffiths crosses for Dembele to flick the winner
  3. H Inverness Caley Thistle in November, a Dembele shot is saved and Griffiths scores the rebound
  4. H Hamilton Academical in December, Dembele rolls a pass to Griffiths to score the only goal of game

Four goals and four assists in 661 minutes equates to a combined Scoring Contribution of 1.08 per 90m. Not bad, but the overriding memories are of little evidence of link up and understanding. Both players are quite individualistic in different ways. Furhtermore, individually, their Scoring Contributions are impressive.

Scoring Contribution (Individual)


The two Celtic players have the best scoring contributions of any players in the top flight in Scotland, and Griffiths’ sheer productivity is untouchable. He averaged 1.27 goals and assists last season, and over his career it’s 0.84. Dembele managed virtually 1 goal or assist per 90m on his own and his career average is 0.74. This appears to be a case of the sum total being less than the sum of the parts. Rodgers reduced their on-pitch game time together as the season progressed. They have the ability to achieve productive outcomes in very different styles, individually.

The thrust of my February article on the pair was that Dembele was proven significantly more productive in what I called “Big Games”. Those being European ties, Cup Semi Finals and Finals, and games against the top four in the league. Revisiting this analysis at seasons end, I also refined it to remove penalties from the calculations as explained above. The results surprised me.


Griffiths productivity is now higher in the Big Games than his season average! He was particularly productive in the Champions League Qualifiers with 4 goals and 2 assists. He did not contribute any goals and assists in the Group Stage, completing only 42 minutes. In a small sample, this may be telling.

Dembele’s numbers suffer when penalties are removed – he scored 4 in those matches. Nevertheless, scoring hat-tricks against The Rangers (head, right and left foots) and St Johnstone (as a substitute) and 2 against Manchester City in the Champion’s League Group Stage were memorable performances.

But Griffiths continues to confound and deliver. And we should be grateful to have them both.


Finally, to round out the analysis – defensive actions.

Griffiths out performs Dembele at defensive actions shocker! Hmmm, yes and no. Dembele is involved in a lot of challenges in the offensive third. Due to the amount of personal attention he gets, and with potential to improve his decision making, he loses a lot of those challenges. I don’t differentiate between attacking and defensive challenges (maybe I should). Dembele’s numbers are heavily influenced by the number of challenges he loses in opposition territory. In terms of purely defensive actions, Dembele is notable for the interceptions and clearances he manages from opposition corners. Griffiths defensive contribution is book marked by his error in letting Hayes run off him to score for the Dons in the Scottish Cup Final.

What this tells me, frankly, is I need to differentiate between defensive and attacking challenges! Another addition for next season.


A perception would be that whilst Dembele, at 20, is heading for the upper echelons of the game with his next move likely to be to one of the elite European clubs, Griffiths would be considered a journeyman were it not for his Celtic career. At 26, his international career has only just started to gain traction, and after a 40-goal season he has found himself second choice at Celtic. With 85 goals for the club and 195 in his career, he would no doubt be a success elsewhere but you would imagine Celtic represents his chance on the biggest stage. Therefore, both players have strong personal career motivators.

Griffiths had a minor melt down in May at Partick Thistle when substituted. Playing well he realised this was a great chance to cement a place at the expense of the injured Frenchman. Rodgers was not amused, but as the cliché goes – it showed how much Griffiths wants to take this opportunity at Celtic.

Both can considerably improve their games. For Dembele it is about judging when to lay it off rather than run with it, and to temper the desire to shoot from distance and be more of a menace in central areas. For Griffiths, it is overall decision making in terms of pass selection and shot selection. He does not get into dangerous positions enough, that is, positions with high goal probability but can score improbably goals.

Both have been injured and missed 25 matches between them – almost half a season. The timing of Dembele’s injuries seemed to result in him being over looked at end of season award time. Therefore, can Celtic rely on youngsters and adaptable wingers in this position as back up? If a 3rd striker was purchased, how could they improve on the productivity and promise of the incumbent pair? Both striker, when fit, provide an automatic Plan A and Plan B between them. As we enter the 2017/18 season it looks like Rodgers will continue to use Rogic, Sinclair and Armstrong and their respective strike ability as mitigation. It remains one of his better problems to have.

Leave a response

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