League Performance Regression? Yes. Here’s Why.

Celtic have won 20% less points in the SPFL than at this stage last season. To be fair, winning 97% of available points is almost an impossible target to exceed. Whilst this hardly constitutes a crisis, or even a slump, it is worth assessing what has changed, in the spirit of continual improvement.

The Glossary is an oasis of reason in a desert of insanity.

Team Performance

A look at the team level data reveals Celtic dominate possession in SPFL games to the tune of 69%. Completed passes by Celtic vs Opposition are

• Season 16/17 560 vs 194
• Season 17/18 596 vs 181.

Celtic’s controlled possession game has not, however, resulted in more shots on goal.

The opposition have fractionally fewer shots that last season. And indeed Celtic have conceded 0.68 goals per league game compare to 0.7 at the same stage last season. Very slight improvement.

So, more possession and the same shot productivity. What is happening to those shots?

Celtic’s have converted 1% less from all shots and the opponents the same as last season. Celtic’s shot accuracy has increased, with a 3% increase in shots on target.

However, Celtic are converting 7% LESS shots on target and the opposition are converting 4% MORE from their shots on target. If you think back to last season and the number of long range efforts from the likes of Armstrong, Griffiths, Rogic and Roberts. That was always difficult to maintain. Celtic are certainly scoring less averaging 2.18 goals per game. At this stage last season it was 2.7. As we saw above, although the opposition are having more shots on target, they are not scoring more goals, in fact 0.02 a game less.

As you should all know, it is Expected Goals that tell us more about chance quality and is therefore a better indicator of performance over the longer term.

Last season Celtic managed to over perform versus the Expected Goals model throughout the first half of the season. They only failed to exceed it once, during a tight 1-0 win vs Patrick Thistle in December. As can be seen below, the curve heads downwards towards end December as the number of games increased and tiredness set in. But Celtic kept doing enough to outperform their expected goals and win games, even if by 1 goal.

This season, you will notice the same downward trend towards the end of December, but there are now four times that Celtic failed to outperform the Expected Goal model. Three of those occasions resulted in failures to win, including the 0-4 reverse at Tynecastle.

We seem to have a theme. Celtic are increasingly dominating possession but not creating any more shooting opportunities. The opposition are performing largely as before. But Celtic’s conversion rate has decreased markedly, despite more accurate shooting.

In terms of raw results, the team found a way to grind out wins especially in the busy December month last season.

This season, after a stellar August, results have tailed off towards the end of the year. Celtic have stopped turning draws into wins in some cases. In short: Celtic are not taking their chances as well as in 2016/17. But why?

Injury / Fatigue

Celtic have played two more league games than at this stage last season and 38 overall. The SPFL looked to front load the league season to make room for the summer World Cup. Ha. (*Looks around the virtual room. No one is laughing*).

Is tiredness an issue? I do not have GPS or player tracking data unfortunately, so cannot assess distances run or sprints per 90m. But we can look at the circumstantial evidence. Appearances and overall utilisation first – data for ALL matches.

I don’t believe it can be a binary issue across the squad i.e. they are all fatigued or not. There are clearly some players for whom the rod has not been spared. Gordon, Tierney, Lustig and Brown are the only players above 80% utilisation across all games. I will discount the goalkeeper as I cannot believe he will be suffering fatigue.

It was clear from the last two games of December that the two full backs’ performances suddenly fell off the cliff. Lustig gave the ball away 38 times in two matches and Tierney 23. Their respective averages before that were 7 and 10. Remarkably, Brown has not really shown any signs of fatigue in his performances. Indeed, during the last seven games before the break he achieved his top 4 “most challenges won” across all games this season!

Sinclair and Forrest are around the 2500 minute mark. Which would mean they have completed every match if they played for Aberdeen or The Rangers, the nearest challengers. But after that players in the squad are in the 50% utilisation range, or less.

In short, I do not see fatigue as being a collective issue, but it would be a strong concern for certain individuals (Tierney, Lustig and Brown).

In terms of injury or unavailability (I’ve also included missing due to Internationals plus not being at the club – i.e. unavailable for selection- Rogic has missed two matches on this basis) the squad totals are:

• 58 matches missed in 2016/17 at this stage and;
• 85 missed in 2017/18.

Celtic have played more games this season, so what this means is that at this stage last season Celtic averaged 87% squad availability and this season it is 84%. So, a 3% drop in player availability overall.

Not all players are equal of course. I have selected what is my Starting XI and assessed their availability. (Some people will hate this. But guess what? My blog.)

This means that on average 2.3 first pick players were unavailable for every game. Is this “normal” or “high”? On Physioroom.com, the average number of English Premier League players out injured is currently 4.45 per club. They have a 25 player squad, therefore around half are first picks (three goalkeepers who are rarely injured). Therefore having just over 2 first picks out on average seem normal based on this sample.

In summary, injuries are slightly up, and in particular it is impacting the first choice players more than the bench / squad players. Some key individuals who have been heavily utilised could definitely use a break. But the injury and fatigue situation cannot be said to be abnormal across the whole squad.

The question is who is missing and what impact might this have on the team?

As we will see in the productivity data below, those playing more minutes have not, historically, got the productivity levels of those who have been missing.

Not All Players Are Equal

I showed on my season review of the European campaign, Half Term Report 1: No Hard Brexit Despite Performances, that defensive performance is a huge factor in Europe. Overall defensive organisation and strategy is brought into question rather than specific individuals. But as we have seen above, Celtic are conceding slightly less goals per game in the SPFL, and conceding slightly less shots per game. Domestically at least, the regression in performance seems to be down to a reduction in attacking potency, not defensive frailty.

And collectively, the attacking talent has seen an overall drop off in Scoring Contribution. This aggregates goals and assists. Assessing that as an average per 90m removes the vagaries of players playing different numbers of minutes.

Only McGregor and Dembele’s Scoring Contribution per 90m in SPFL games are up on last season. The rest of the attacking talents’ contributions are significantly down. Especially Griffiths. He as 7 assists overall but only 1 in the SPFL and this is dragging his contribution down significantly.

But what about Expected Scoring Contribution? Actuals can be boosted by scoring a few long-range screamers that cannot be sustained, or picking up a few cheap assists for passes on the half way line. xSC90 tells us the chance quality of goals and assists per 90m.

Many of the attackers have improved their xSC90 over last season. This is slightly good news. It is more worrying to not be creating chances than to be creating and not taking them.

Whether you agree with my first XI or not, I can demonstrate the most productive attacking players from last season by looking at Scoring Contribution per 90m.

If you look at the goals per 90m contribution of the main attacking players, only Forrest has improved his productivity significantly.

Having Forrest and McGregor as showing most improvement in Goals per 90m is great to see, but from a team perspective, you want to see the main goal contributors holding their previous performances at least – Sinclair, Dembele, Griffiths, Armstrong, Rogic.

With McGregor, his productivity is slightly up on last season, but we are simply seeing more of him due to injuries to others. Similarly, although Forrest’s goal output has increased, his assist productivity has decreased by 0.325 per 90m. Armstrong’s goal output has decreased but he is creating more chances for others.

A slightly improved McGregor, and a more prolific Forrest, but with less assists, does not compensate for the drop off in productivity of the main attacking players.

Here are the overall productivity numbers for 16/17:

Apart from McGregor with a small improvement, all other attackers have seen double digit decreases in overall Scoring Contributions per 90m.

This is not surprising when you consider than the most productive players (Griffiths, Dembele, Roberts, Sinclair, Rogic) have barely been on the park together. Based on productivity, I would have Rogic, Roberts, Sinclair and Dembele (better all-round player than Griffiths) as starters. Last season at this stage, that four had appeared together in 6 league games. This season, only once have three of that four started a game – the 2-0 win at Ibrox. No more than two of the four listed have appeared at the same time. Last season at least three of the four started 50% of league games.

Overall injuries are no worse than any English Premier League sides. But our injuries have happened to the most productive players, and stopped Celtic getting them on the park at the same time. The cumulative result is a huge drop off of chances taken. Forrest and McGregor have stepped up but, despite improvements in their output, are not producing to the rate of the players they have replaced.

One final point. Of the four most productive players, it is Sinclair who has been the most utilised this season. He has appeared in 75% of Celtic’s playing time, and none of the other most productive players have managed 50%. Yet it is Sinclair who seems to get criticised. His productivity level has dropped off yes, but not as much as his other prime team mates. Effectively, as the most used senior attacker, he has had to carry the burden of the attacking threat for the team. This makes Celtic easier to play against – and we have seen teams double up on Sinclair. Last season if he had a quiet game, Dembele, Rogic, Roberts, Griffiths would come up with a goal or assist. This season he has often been the only senior attacker on the field. It is time to lay off Sinclair.

Conclusion

Celtic have without question regressed in terms of league performance. Despite this, there is a healthy lead of 8 points. The performance remains on track to meet strategic objectives. But we seek continual improvement.

There cannot be one single reason to explain team performance over 22 matches. Many factors will come into play. The following will all have had a bearing to some extent:

• Injuries
• Number of games
• Fatigue levels per individual
• Improvement in opposition
• Opposition familiarity with Celtic’s style under Rodgers
• Natural cycles in player career performances vary
• Have Celtic refreshed the 1^st team squad sufficiently – build from strength?
• Last season saw significant over achieving and some performances are regressing to the mean (Armstrong)

There is no one factor. But as we have seen above, an important contributor to the decline in league form is that the most productive players have been the ones missing from the starting line up. And the replacement players, despite improving their own performances, cannot compensate for that drop in expected productivity of those missing.

The wrong players have been injured at the same time. Celtic need Rogic, Roberts, Dembele, Griffiths back AT THE SAME TIME to help Sinclair who has had to shoulder a disproportionate amount of the work load, and criticism.

This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.