Time for the good, the bad, & the ugly from yesterday’s Celtic vs Hibs game. I’ve tried to keep these posts fairly consistent and simple, as I recognize the metrics and nomenclature are new to many. Unfortunately, I think a productive analytical review of yesterday’s game requires more detail and complexity.
I typically share the pre-shot non-penalty xG data for games. Yesterday’s was 1.81 vs 1.71, which understandably feels wrong to many. I stated after the game on Twitter that my guess was around 2.5-1.0. So what happened?
I haven not shared post-shot metrics in the past, but it basically looks at the xG once the characteristics of a shot are included, such as its speed, was it on goal, positioning of the keeper, height, location in goal mouth, etc. This is yet another area where performance varies a lot game to game but typically reverts to the metric over time.
I believe a fair way to characterize yesterday’s performance was that the quality of scoring chances was close between the 2 teams, but that Celtic’s shooting was accurate and well placed, while Hibs’ was really poor. The post-shot xG was 2.82 vs 0.
As I stated in my pregame benchmarking Twitter thread, Celtic were likely to win, possibly with a comfortable score line, but that risks of variance would probably be higher than necessary given Lennon’s persistence with certain decisions. A pre-shot xG of 1.81-1.71 at home, vs a side of Hibs’ quality who decided to play a relatively open game, is not good. The openness of the game was certainly entertaining and like most, I am sure, I very much enjoyed the game.
However, normal variance could have easily resulted in a loss or a draw, and over 38 games, comparable performances most likely will. As I state regularly, as a one-off this would not be concerning, but a clear trend and pattern has emerged, and my analysis remains that decisions are not being made to address them. What does variance look like?
If we compare Celtic’s performance yesterday versus Rangers’ last week, we can see that with a far greater xG differential, Hibs were able to earn a draw.
Longer term, such as over a 38-game season, results tend to track underlying performance metrics. This is not to suggest that the victory and resulting three points was not important and welcome, but I believe it would be a mistake to take it as an indication that underlying issues unfolding so far this season have been resolved.
Now for the good- Frimpong is the obvious candidate and I give him one but with a qualifier. His athleticism is amazing, and his volume and efficiency stats reflected a dominant performance- 21 of 31 duels and 4 of 5 loose balls won, for example. However, his end product remains an issue- 0 of 5 crosses accurate, just 0.04 xA and he had 0.33 xG, but ballooned both shots. He will be frightening if/when he improves this part of his game.
Good and my MOTM goes to Bitton once again. He won 10 of 16 duels, made 4 clearances, and his ball progression was excellent. He made 9 progressive runs, 3 dribbles, and 11 progressive passes, which are levels Ajer typically posts in his good games (Ajer was relatively less involved yesterday).
Bad goes to the openness of our defending once again. I am not going to give a bad to Duffy, as overall I think he was fine. But, I think it is becoming noticeable and a reasonable topic of concern as to how he fits into a system playing a high defensive line regularly, and how and what a Jullien return will look like. Did we really sign Duffy to bench Jullien? If not do they fit the system together? I believe this may be of increasing concern as the first derby rapidly approaches.
Ugly goes to me- I allowed my emotions and temper to get the best of me the past couple of days and I crossed a line on Twitter. I have tried to stick with analytics using publicly available data and let the analytics speak for itself. I have gotten quite angry at what I believe to be terrible and damaging decision making by Neil Lennon. However, that is no excuse, and I crossed my own line by using unnecessary, inflammatory, and overly critical language. I apologize to Lennon and to anyone who may have read my comments.