I could no longer cope with being constrained to a thread of 280 character tweets to get across my numbers based observations and judgements about Sunderland AFC, so I had a word with the head of the merry band of volunteers at Roker Report, and he foolishly agreed to let me have a little corner of the site in which to go full nerd.
Those of you who follow me on Twitter - I’m @MobileMackem - will know that I produce weekly alternative league tables for League One. With those tables, I attempt to illustrate why Sunderland (as well as all the other teams in the division) are under or overperforming. I like numbers. Numbers tell me the truth and, contrary to what many believe, statistics are not akin to lies. At least not when in the hands of those who understand the numbers and have no politically motivated message to deliver.
Anyway, this tweet caught the eyes of many last Sunday (September 22nd, 2019):
#SAFC are making progress under Ross, but the question is whether after 72 games he is taking too long to find the full potential of a good squad on a huge League One wage bill, writes @sturayner https://t.co/yRCz5Ec8oL— The Chronicle (@ChronicleSAFC) September 22, 2019
The response, from most, suggested that its author was in a small minority in thinking that SAFC are “making progress”. However, perhaps that rather negative response to Stuart Rayner’s assertion was fuelled by the still raw memories of the previous day’s pitiful performance at Bolton? Or perhaps it is just wrong?
Before considering the question of progress further, allow me to introduce the alternative league tables I like to use to inform my opinions.
Possession is probably the least important of the metrics I will present, but it provides valuable context for the discussion which follows.
At the higher levels of the game, possession is important. Liverpool on 0.63 (63%) and Manchester City on 0.61 (61%) occupy the top two positions in the Premier League table. A certain team from the north of the River Tyne have the lowest possession in the Premier League with 0.34 (34%). The last time I looked, they were close to the bottom of the league. I’m implying a strong correlation without having actually measured it.
In League One, possession is of little importance. Contrast Liverpool and Manchester City with Rochdale and Portsmouth who occupy 17th and 20th places in League One respectively. It has been a feature of this season that Rochdale play attractive possession-based football and they certainly impressed against Sunderland. But, ultimately, how much good is it doing them? Not a lot.
‘It’s not how large your possession percentage is, it’s what you do with it that matters’, or so they say. Just ask Wycombe Wanderers who are 21st in the possession table and 3rd in League One. More on what is going on with that soon.
Next let’s discuss a much more important possession-based metric – that of ‘possession effectiveness’.
A team might like to play a possession game, like Rochdale for example, but what matters is that team’s ability to turn possession into clear goal scoring chances. You have a winning formula if you play a possession game and you create a hatful from that possession.
Sorry Rochdale, but I’m going continue to use you as an exemplar.
They were our possession league leaders in the previous section, however, now we look at their position in the ‘possession effectiveness’ table, we see they are a lowly 20th. I also highlighted above that our favourite panto villains, Wycombe Wanderers, occupy a lowly position in the possession table. But when it comes to ‘possession effectiveness’, they sit 2nd. Impressive work from Gareth Ainsworth’s men to turn 44% possession into an effective (and no doubt extremely aggressive) attacking display.
What of Sunderland? We are a mid-ranking team for possession (11th of 23 with 51%) – fine. Our possession effectiveness, however, gives cause for concern. Only Rochdale, Southend, MK Dons and Bolton are worse at turning possession into clear goal scoring chances. Which leads me on to the next metric, that of attacking effectiveness.
This is simply the number of shots on target per game. It’s a metric I’m keen on because it’s easy to understand and it is a very good predictor for which teams will be successful and which will fall short.
As a rule of thumb, promoted teams tend to average five or more shots on target per game in League One. It’s still early days in the 2019/20 season, but, for the moment at least, Ipswich, Coventry and Peterborough are looking tidy. Unfortunately, Sunderland are not.
It will come as no surprise to anyone who watches Jack Ross’ Sunderland every week, that we don’t get a lot of shots on target during the average game. Even with that in mind, it is a little alarming to find that we are keeping such company as Rochdale, who are 17th in League One and AFC Wimbledon, who are 22nd on similar ‘attacking effectiveness’ scores.
With that in mind, why are we currently 5th in the League? The bulk of the answer to that question becomes apparent in the next metric – attacking quality.
Sunderland are currently 5th in League One, largely because our finishing is so good. 45% of the shots on target we have during the game result in a goal. Until the previous two games, Sunderland were in first place with over 50%, however, a combination of poor finishing (looking at you McGeady, Wyke and Grigg) and good keeping have knocked us into second place. Also, Oxford won the lottery last weekend putting six of their seven shots on target against Lincoln in the onion bag. That’s an exceptional performance, the impact of which, on this table, will be diluted over time.
Two further interesting observations from this table are the positions of Southend and Coventry. The former currently sit 21st in League One but have an attacking effectiveness of 0.44 (just behind Sunderland in 3rd). The latter are 2nd in League One and have an attacking quality of only 0.30. The difference? Southend create only 2.78 shots on target per game while Coventry create 5.56. The lessons here for Sunderland are that Southend are potent if you give them a sniff in front of goal despite their lowly position. And you can expect to spend a lot of time defending against Coventry.
Turning attention back to Sunderland, our generally high quality of finishing masks problems elsewhere. We seem to rely almost entirely on moments of quality from players of quality. The danger of that approach is that sometimes McGeady will take the worst penalty in his career and Maguire’s 30 yard screamer will end up in row Z.
Worryingly, the situation was very similar in the 2018/19 season.
Next let’s discuss defence.
Similar to ‘possession effectiveness’, ‘defensive effectiveness’ simply measures how many shots on target each team allows their opposition per game.
The meanest defence is that of Coventry, with Peterborough and Ipswich not too far behind. It will be a surprise to no one that easily the worst defence is Bolton’s with AFC Wimbledon just a little better. When we beat Accrington, Rochdale and AFC Wimbledon, it was stated by many that we won despite playing poorly, much to the anger of a few. The position of all three of those teams in the ‘defensive effectiveness’ table would suggest they are rather easier to create chances against than the average. Combine that with Sunderland’s attacking quality, and you have a positive result (without the need for a good performance).
I won’t labour the Bolton point too much and the data is skewed by their deadline day business. But it is interesting to observe that they give away an average of 8.5 shots on target per game. Sunderland managed 5 on Saturday and put one away. An attacking quality of 0.20 on the day meant that, unfortunately, the ‘paper on the cracks’ came unstuck that day.
Turning our attention to Sunderland’s position in the ‘defensive effectiveness’ table, we see that we are a relatively healthy 8th with a score of 3.45. I will discuss this further later, but this is actually a sign of improvement on last season. But this improvement is not leading to an improvement League performance, why is that?
For the answer, I’m afraid I need to point the finger in the direction of an area of the pitch where quality shone brightly last season but has since dimmed.
Ipswich haven’t scored a whole lot of goals yet this season. They rank a lowly 16th for attacking quality (although they do create a lot of chances). But they’ve conceded only four goals while facing 25 shots on target. That leads me to conclude that their goalkeeper is performing as exceptionally well, hence a score of 0.84.
Over the course of the 2018/19 season, the top three goalkeeping performances came from Charlton in first place with 0.762 goalkeeper quality, Sunderland second with 0.758 and Coventry third with 0.737 (Lee Burge anyone?). I expect Ipswich to settle closer to 0.75 over time.
Now consider the contrast between Sunderland’s goalkeeper quality score last season of 0.76, and the score so far in 19/20 of 0.65. Keep in mind that, for the entirety of both seasons, Jon McLaughlin has been between the sticks.
We found, when discussing ‘defensive effectiveness’, that Sunderland are doing relatively well. They are allowing an average of 3.45 shots on target to the opposition per game. If we compare this to the 2018/19 season, during which we allowed 4.21, we can see that, without doubt, Sunderland have improved defensively. Yet we’re conceding more, why?
Unfortunately, this comes down to the drop in ‘goalkeeper quality’ from 0.76 in 18/19 to 0.65 this season. Sunderland rely heavily on quality at the front and at the back. When one or the other drops off, so do our results.
There is a case to be made for giving Lee Burge a run in the team if we want some clean sheets.
Comparing 18/19 to 19/20
I want to look at ‘attacking effectiveness’ and ‘defensive effectiveness’ from a slightly different angle next by comparing some teams from the 18/19 season to our current season. To this ‘attacking effectiveness’ scatter plot, I have added Sunderland 18/19, Sunderland 19/20 and last season’s promoted and relegated teams.
The happy quarter of the plot to be in is the top-right where the attack is both effective and exciting. You also get promoted when you’re close to the top-right corner, as both Luton and Barnsley will attest.
The bottom-left quarter is an unhappy place to be. That is where you’ll find teams who have fans that don’t have much to cheer and who get relegated.
I mentioned earlier that teams who get promoted tend to have five or more shots on target per game. Luton and Barnsley are comfortably above this threshold. Sunderland in 2018/19 were not. And in 2019/20, we are getting an average number of shots on target that is less than the four teams relegated last season. We’re also scoring fewer goals. Not good.
There is, however, no danger of relegation because, although we are getting a relegation level return in shots on target, we occupy a position in the top-left corner (which I have labelled as ‘effective but dull’) thanks to our high attacking quality. No “progress” here.
With the same group of teams, let’s examine ‘defensive effectiveness’ in the same way.
This time, the happy corner to occupy is the bottom-left ‘chilled out fans’ zone. Chilled out fans know that the opposition aren’t going to get too many good chances to score against them and, when they do, the keeper will probably make a save.
Luton and Barnsley had pretty relaxed fanbases last season. And, to be fair, Sunderland also occupied a position in the bottom left quarter. However, it’s clear that there was a large gap between the 18/19 version of Sunderland and the two promoted teams. Not quite good enough, as we know.
Turning our attention to Sunderland in 19/20, there is good news and bad. The good news is that I have finally found something I can agree with Stuart Rayner about. The defensive performance has improved as the Sunderland 19/20 datapoint has moved much closer to the levels needed for promotion. However, there is also bad news because we have edged towards the ‘Lee Camp’ zone. In other words, the goalkeeping performance has dipped, thus confirming my preliminary conclusions in the earlier discussion.
Finally, I’ll turn my attention back to where I started with the tweet from the Chronicle. Have Sunderland made progress under Ross? You won’t be surprised to learn that there is no straightforward answer. Defensively, we have improved but this has coincided with a drop in goalkeeper performance. Attacking wise, we are are poor at turning possession into clear goal scoring chances and, consequently, we rely far too heavily on moments of quality. The number of clear chances created has dropped from 4.50 last season to 3.45 this season. That certainly could not be thought of as “making progress” because, if we’re going to rely on moments of quality, the number of opportunities need to be maximised.
Many have observed that the Sunderland of both last season and this, are slow, predictable, ineffective and dull to watch. The numbers we have looked at today seem to bear that out. When the ‘moments of quality’ abandon Jack Ross’ Sunderland, they seem to have little else to offer.
Join me again in the coming weeks as I take a closer look at other aspects of Sunderland’s performance, trends over time and maybe even some suggested solutions to their woes.