By James Dailey
My mentor in the analytics field taught me a lesson which I have found invaluable. To put it simply, he said it is best to always start with asking what question one is attempting to answer? Alan did an excellent job of answering the question in this piece last fall about what the actual results were for Celtic playing away on plastic pitches overall. My goal today is to answer other questions: why have Celtic supporters developed the idea that Celtic play poorly on plastic pitches, and is there any statistical evidence to make the idea a rational one?
I will start with a qualitative declaration – not all plastic pitches are created equal. Some have advanced to the point of being reasonable playing surfaces, while others are an abomination. Like grass pitches, they also come in different dimensions and are played upon in varying weather conditions which further complicate things depending upon how those conditions interact with the specific type and quality of the plastic pitch. To put it simply, this can be quite complicated.
It would be unrealistic to try and analyse all of these factors as they pertain to the expansive sample Alan used in his piece. I have examined the three pitches in the SPFL which Celtic have played the most upon over the data sample to which I have access, which dates from the beginning of the 2015-2016 season to present. I have focused upon SPFL league games only, which numbers 184 in total over that time. I think you will find the results of interest.

I calculated the average total xG (including penalties) for all league games over the period, which total 184. One can see the average of 2.20 per game overall and contrast with the 1.49 average for the 20 games at Livingston, Hamilton, and Kilmarnock. That sample represents 18.7% of total games over the period, so it is a decent sample size.
Various things can be ascertained by looking at the relative composition of those matches, or at least that was my hypothesis when I started doing this. Notably, just 3 of the 20 games resulted in Celtic having an xG over the 2.20 overall average. When I look further into the sample of 20 games, I noticed something else of interest. There were only 3 of 20 which were played primarily with two actual strikers, and two of those games are in the 3 which were above average in xG.
The average xG in those 3 games was 2.33, however, that sample includes the March 19th, 2016 game at Kilmarnock where the striker partnership was Kazim-Richards and Griffiths, where the xG was 1.36 in the 1-0 victory. The second of the 3 games was January 22, 2020 at Kilmarnock with an xG of 2.26 in that 3-1 victory. The third of 3 was the February 2nd, 2020 at Kilmarnock with an xG of 3.36 in that 3-1 victory. Both games this year were in a 3-5-2 formation with Edouard and Griffiths at striker. That 3rd game was also the only example we have with McGregor playing as a central 6 on a plastic pitch.
I also examined the record of The Rangers on plastic pitches since they arrived in the Premiership and they’ve only played with two strikers on a plastic pitch once: January 23rd, 2019 with 3.61 xG in a 2-1 defeat with Defoe and Morelos partnering.
This data set has helped me answer a couple of questions and cause me to ask another. I believe it is fair to say that Celtic have performed well below average on the plastic pitches at Livingston, Hamilton, and Kilmarnock overall. This underperformance has been pronounced and persistent and has particularly been true when playing a single striker system.
The big question I have is why do neither Celtic nor The Rangers, both of whom struggle at these venues against low block sides, experiment more with playing variations with two strikers?