On Thursday 28th March the following podcast was recorded on the A Celtic State of Mind platform. You can view it here or listen wherever you consume your pods.
Today we wrap up the written record of this study.
I have presented several analyses. They all have a consistent result. There is one club that is a statistically significant outlier as regards being a beneficiary of refereeing decisions – The Rangers.
A pattern of assistance for one club.
The Rangers have seen rivals Celtic win 12 of the last 13 available Premiership titles and six out of the seven since they achieved status as a topflight club. If anything, it is Celtic that one might expect to have dominance in benefitting from decision-making given superior possession (average 68 percent possession to 62 percent by The Rangers) and attacking capabilities.
The Rangers could not be considered a suffocatingly dominant side over the rest of their league like e.g. Paris Saint-Germain in France and Bayern Munich in Germany, or, errrr, Celtic.
Furthermore, many criticise Scottish referees for poor performance, but if the part-time officials overseeing full-time professionals are merely incompetent, then decision-making anomalies should have no discernible pattern.
All clubs should suffer equally from incompetence over a large sample size.
Counting Penalties
This all started by counting penalties.
The Rangers' sagas (SAGAS – PLURAL) of not conceding a penalty were remarkable and difficult to explain (44 and 75 matches respectively since Crawford Allan was appointed Head of Refereeing Operations).
Simply counting penalties as a basis for “analysis” is inherently limited and I have used a framework that considers:
Were penalty awards correct or not as judged by a professional, verifiably neutral expert?
Were red card awards correct or not as judged by a professional, verifiably neutral expert?
Were goals allowed or disallowed correctly awarded or not as judged by a professional, verifiably neutral expert?
When awards were incorrect, what was the estimated impact in terms of expected points to the teams given the time of the award and the match score at that time?
What was the likelihood of teams getting and conceding penalties given their attacking profiles?
What trends were there in the distribution of penalties across multiple seasons?
What trends were there in the distribution of red cards across multiple seasons?
What trends were there in the estimated expected impact of red cards and penalties given game state across multiple seasons?
Given the available data, are these reasonable questions and a reasonable basis for a hypothesis set?
For each section of the multi-year analysis, the expected outcome was the same – “There should be no significant difference between distributions of data for Celtic and The Rangers”.
However, there appears to be statistically significant evidence - consistently high Z-Scores, over two standard deviations from the mean. The Rangers are the consistent statistically significant outlier.
The pattern is not of favour towards the “big clubs”, nor is it an “Old Firm” leaning. It is favourable decision-making, a pattern of assistance, for one club.
Penalties Again – Why They Matter
In particular, the awarding, or not, of penalties seems to have a distinctly one-sided pattern.
Specifically, Rangers seem to disproportionately avoid the jeopardy a penalty brings when awarded against you in football. Football is a low-scoring sport. Some argue penalties are disproportionately impactful to the penalised side. This is covered in The Athletic and by this piece in Sabermetric Research.
Essentially, if we take the example of a foul at a corner where two players are challenging for a headed ball (say Triantis of Hibernian and Souttar of The Rangers). If a foul and penalty are awarded in that situation, you penalise an approximate 5 percent scoring chance (a header from a corner has an approximate xG of 0.05) with a 77 percent chance of scoring from a penalty (penalty xG is 0.77). Exceptionally harsh.
And of course, if you are NOT awarding penalties, you are denying the attacking team that disproportionate opportunity.
The SFA appears to have a refereeing problem and it is recommended the next steps are to establish why this is, and then what can be done about it. This affects ALL clubs.
Celtic supporters can take direct action. If you are a shareholder, please register support for the following AGM resolution here - Resolution On Refereeing Standards – SENTINELCELTS.
------------------------------------------------------------------------------------------------------------------------------------------
Appendix 1 - Data Sources
All data used is in the public domain.
Here are the sources I have used.
Red Card information from the SPFL archive.
Specifically, Rangers red cards are summarised on the fitbastats side
Penalty awards from Transfermarkt.co.uk and the SPFL archive.
Analysis from an independent, expert referee is documented on the website Celtic By Numbers.
Video evidence for independent review comes from BBC Sportscene, the SPFL Youtube Channel, and occasionally social media clips posted by supporters.
“Touches In The Box” data was taken from FotMob with an initial historical data scrape from Wyscout (provided by a subscriber).
Fouls and possession data was taken from the BBC Sport website.
I used Wyscout’s method for possession adjusting to calculate fouls based on possession.
The Red Card expected points impact model is documented here. It is based on data from the SPFL archive and BBC Sport website.
The expected points model was provided by the American Soccer Analysis website.
Appendix 2 – Scope
The study focuses on SPLF Premiership matches only up to 3rd January 2024. Cup competitions are too variable in terms of the level of opposition; the impacts of extra time; the use of referees outside the normal Premier League cadre; and the impacts of playing on neutral grounds.
The Rangers achieved Premiership status in the 2016-17 season. Data from this point onwards is used for benchmarking purposes and to widen the sample sizes.
For comparative purposes, the study focuses on the 2020-21 season onwards unless stated. Between 2016-17 and 2020-21, as the new “Rangers” established themselves within the Premiership, Celtic were league champions with gaps back to The Rangers of 39, 12, nine, and 13 (season curtailed due to COVID-19) points.
Beginning with the 2020-21 season, onwards, the two rivals’ level of spending on wages approached parity. In contrast, in the prior four seasons following Rangers’ promotion to the Premiership, their wage bill had been 34%, 41%, 61%, and 80% of Celtic’s. The points gap during those four seasons tallied 39, 12, 9, and 13, with the 2019-2020 season truncated by the COVID pandemic.
Upon The Rangers’ wage bill reaching 92% of Celtic’s for the 2020-2021 season, they won the league by 25 points, winning their first title since promotion. The following two seasons Celtic reclaimed the title, with second-place The Rangers four and seven points adrift.
2020-21 also saw Celtic attempting to win their 10th consecutive Premiership title which would have been a new Scottish record. Both Celtic and Rangers had won nine consecutive titles in the past, with Rangers’ consecutive stretch occurring when they similarly enjoyed a dominant financial advantage over Celtic and the rest of the league. The amount spent on wages at Celtic remained below 70% of Rangers over the period. Before the 2020-2021 season, Celtic had also enjoyed an extended period of spending far more on wages than the competition.
In addition to the historic nature associated with which the club won the 2020-2021 league title, the two subsequent seasons offered guaranteed Champions League group stage football to the Premiership winners, and upwards of £25 million in additional revenue.
In short, there has been much more at stake since 2020-21 than just winning a flag, trophy, and a medal.