The Stat Man: Sunderland top the table for attacking effectiveness - & Burge is having an impact

It has been a couple of weeks since I graced your screens with my nerdiness but now I’m back, once again, to cast an alternative light on Sunderland’s season and to look to the future. First - an update on the alternative league tables, starting with the attacking side of the game.

Keep in mind that the data contained within the tables represents the 2019/20 season to date. Hence it contains Jack Ross’ eleven games as well as Phil Parkinson’s two. The impact of Parkinson was rapid and abundantly clear to anyone who watched Tuesday’s demolition of Tranmere (more on them later). However, his impact will take longer to appear in these tables.

Sunderland remain one of the teams who are least effective at turning possession into shots on target, languishing 17^th of the 23 teams both for possession effectiveness and attacking effectiveness. The outlook was even more grim following poor attacking displays against Lincoln and then Wycombe. But a tremendous performance against the men from Birkenhead turned that back around.

We see the lads make a welcome return to the top of the attacking quality table after a run of poor attacking displays, including two games in which we handed the opposition a clean sheet. Putting away five of six shots on target in a single game works wonders.

Teams who get promoted from League One tend to average five or more shots on target per game. More performances like Tuesday will serve us well, although there is no way we can expect to score from 83% if shots on target on a regular basis so don’t expect too many five-goal hauls.

Meanwhile, Wycombe continue to produce effective attacking performances despite being bottom of the possession table (not shown).

Defensively, Sunderland are midtable, but there are green shoots of improvement.

Since the introduction of Lee Burge, Sunderland’s ‘goalkeeper quality’ has improved from 0.64 to 0.68. Burge has faced eight shots on target and let in one. That’s an impressive goalkeeper quality score of 0.88. The kind of level Ipswich were attaining, until recently.

I said a few weeks ago that Ipswich’s exceptional goalkeeper quality couldn’t continue and, as predicted, it has dropped from 0.84 to 0.77. Funnily enough, they’ve started losing at the same time… Burge won’t continue at 0.88 and will also come back to the pack, but provided he settles at about 0.75, we’ll be in good shape for reducing our goals conceded enough to have a chance of automatic promotion.

Tranmere score low on all three defensive tables. As much fun as that 5-0 win was, they are one of the easiest teams in the league to score against. A special word for poor Southend, not only do they give away almost six shots on target per game, they also have the worst goalkeeper quality in the league. Combine this with their poor attack they look doomed unless Sol Campbell can pull off another miracle.

Now let’s have a look at the attacking and defensive effectiveness plots.

As I mentioned earlier, it’s too soon for Phil Parkinson to have an impact on what you see. Sunderland score a low-end promotion number of goals, but do not create a promotion number of clear chances. That red and white dot needs to start moving to the right. If Tuesday’s performance is representative of Parkinson’s future impact, we will move into that green ellipse soon enough.

Peterborough are way out on their own. As are Bolton, but in the opposite direction. I expect to see both drift back towards the others over time. If they don’t, then it could be a record-breaking season for The Posh (unless we steal Maddison in January…).

We witnessed that Tranmere weren’t great in attack on Tuesday night, but they are by no means among the worst in the division. This weekend’s opponents, Shrewsbury, are among the worst though. With Burge playing well and Shrewsbury having the second lowest attack quality in the division, another clean sheet looks likely.

Sunderland sit in the chasing pack defensively, sharing a crowded space with Gillingham, Bristol Rovers and Blackpool. If Lee Burge (or a rejuvenated Jon McLaughlin) can nudge the red and white circle down by reaching a goalkeeper quality of 0.75 and the defence can give away one less shot on target per game, then there’s every chance the lads could be sitting in a spot close to the one currently occupied by Ipswich before the season is over.

Another mention for Tranmere now. Only MK Dons, Wimbledon, Southend and Bolton have worse defences. And only Bolton and Southend have conceded more goals per game. Tuesday was great, Tranmere are not.

Talking of Southend, their position in the defensive effectiveness plot is as extreme as Bolton’s but for a different reason. Bolton give away far too many chances, but their keeper is decent. Southend, on the other hand, also give away too many chances (albeit fewer than Bolton), but their goalkeeper quality is very poor. Half of the shots fired at him result in a goal. I doubt Doncaster will be the last to put a hatful past them this season.

---

Looking to the Future

The alternative tables and effectiveness plots are useful tools for analysing overall team performance. Combined with a couple of thousand lines of code, an algorithm, and some machine learning, it’s also possible to use the information they contain to predict how the rest of the season might go. So I thought I might as well.

In a fraction of a second, starting from the current actual league table, each game that remains in the season is played and the most likely result calculated. The result is the league table in May. Or what the data says it might be anyway. Have a look.

I have learned through experience that this type of thing, and football statistics in general, can get neck muscles straining and blood pressure soaring. But before you start shouting at your screen that ‘the game isn’t played on paper’, ‘there’s no way Peterborough are getting 113 points’, or ‘Gillingham 7^th, don’t be daft’, please keep in mind the following:

1. It’s data and code, not my opinion. I do not think Peterborough will get 113 points!
2. The result will change every time a game is played because it introduces new data.
3. Sunderland have got a new manager and the result above was predicted from their 13 games to date, 85% of which were overseen by Ross. As Parkinson has an impact, the result will change, hopefully for the better.
4. It’s just a bit of fun, so let’s enjoy watching the predictions change over time and take it for what it is!