Why Celtic Should Win The Title

A regular reference point for performance discussion on The Huddle Breakdown (subscribe and never miss an episode) is xG difference.

That is: the difference between Expected Goals For and Expected Goals Against.

The higher the difference in your favour, the more likely your performance will be sufficient to win the game. Football is low scoring and so, as Chelsea discovered last night, you can dominate possession and chance creation, but be undone by one moment of absolute brilliance (the Modric pass to Rodrygo) or bad luck (Rudiger slipping). The impact of misfortune becomes less the greater the gap between the quality of chances you are creating versus the opponent. Hopefully, that is fairly intuitive.

A case study for bad outcomes is Celtic last season.

20/21

The Huddle Breakdown was born during Celtic’s worst season in 10 years. Plenty of material then!

A consistent observation was that not only were Celtic not creating enough better-quality chances than their opponents, they were giving up too many chances as well.

Here is the 6-match rolling average xG differential across domestic matches in that fateful season:

There are three aspects to this we should concentrate on.

1. The red trend line shows that xG Difference across the season varied just either side of 1. That is, in domestic games, Celtic’s xG was around 1 higher than their opponents. That trend improved over the season as COVID issues, injuries worked themselves out and Lennon left. But also, harsh truth, improve too late when the league was all but over.
2. What is more important than the trend direction is the average differential. A differential of 1 is (my judgement), too small in a low scoring game, to exclude the negative impact of variance (injury, sendings-off, penalties against, player error, 30-yard screamers) having a bad impact on results over the long term. Simply, there isn’t enough “insurance” in the xG differential to absorb unfortunate events. We know last season, with the defensive alignment we had, unfortunate events were many.
3. Look at the variability shown in the height and depth of the Green line. This shows high volatility of both xG for and against. Unpredictability can result in the occasional 6-0 win, but as likely is the 0-1 loss to a bottom 6 side. In layman’s terms, the team lacked consistency.

21/22

Now let us consider Celtic’s 21/22 performance domestically. I am excluding Europe as, for the purposes of this article, I am considering the likely league title destination. There is a wider discussion to be had about Europe.

Anyway, Celtic’s 6 game rolling average xG Differential:

Again, let’s focus on the three components:

1. The trend line is flat and even trending down very slightly. Both xG for and xG against are trending down over the season. We have spoken about this on the pod. My opinion is that Postecoglou, after the heady early days of going “full Ange”, that he has become slightly more pragmatic and traded a bit of offensive potency for improved defensive solidity.
2. The trend is for Celtic to have 1.7 xG MORE than their opponents in each game over a 6-game period. This is hugely encouraging as a 1.7 xG difference is more likely to mean you can absorb the occasional honest mistake, rush of blood or squiggler from 40 yards.
3. After November when Postecoglou started to improve the defensive aspect of the team, notice how small a difference there is between the Green line heights and depths. That is consistency. Celtic are now consistently over 1.7 xG difference over each 6 game period. And right at the end of the season, that is trending nicely upwards (4.82-0.05 xG in one game helps!).

Let’s look at Celtic’s nearest opponents this season for reference.

This covers SPFL matches only for The Rangers this season. Let’s go through our 3-point check list:

1. The trend line is quite flat but trending downwards slightly but more steeply than Celtic’s. However, as we progress through our analysis, we can see focussing on that hides a lot of variability.
2. The Rangers xG Differential has never been higher than 1 all season and is now trending less that 0.8. That is worse than Celtic last season and opens up the team to exposure to great/poor goalkeeping, great/poor finishing or good/bad luck variance. An example would be the 3-3 away draw with Ross County (xG 3.49 – 2.16) where McGregor errors led to a draw.
3. Most importantly, potentially, is the shift from predictability under Gerrard – notice in August through end October very little variance in height and depth of difference, AND the xG difference consistently over 1. Since Van Bronckhorst took over, we see high degrees of variance against a decreasing differential trend. The Rangers are becoming more inconsistent and generating less of a xG differential under this manager.

Summary

Celtic and The Rangers are going in different directions are regards Expected Goals for and Against trends this season.

For Celtic, Postecoglou has a consistent style of play that the players are increasingly implementing into consistent performances, helped by a reducing injury list. More vitally, Celtic’s base xG difference in domestic matches is around 2 goals. This means that they are able to absorb the occasional slings and arrows of outrageous fortune.

As regards the Blues, the highly predictable, but effective, style implemented by Gerrard/Beale which was yielding consistent xG differentials (albeit less than Celtic’s), but enough to maintain a challenge, are becoming a more chaotic under Van Bronckhorst. That variability may result in occasions like Sunday – 4-0 win, but are also likely to see them fall victim to misfortune or the impacts of an off day being exacerbated as we saw in Celtic’s recent 2-1 win at Ibrox (Celtic had an xG advantage > 1).

Dare I say it, Van Bronkhorst’s side have more in common with Lennon’s Celtic. And his tactical choices in the two recent Derby matches support this comparison.

xG Difference does not mean Celtic will definitely win the league.

But, give me a higher differential consistently delivered, to high variability around a lower differential, any day of the goddam week!