clock menu more-arrow no yes

Filed under:

The Stat Man: Why can’t Sunderland convert performances into results?

New, comments

Given that the wheels appear to be coming off Sunderland’s push for automatic promotion, now feels like a good time to blow the dust of the numbers to see if we can uncover what is really happening.

Doncaster Rovers v Sunderland - Sky Bet League One Photo by Ian Horrocks/Sunderland AFC via Getty Images

Recently, seemingly uncharacteristically, Phil Parkinson has taken to mentioning statistics in his press conferences. His favourite has been the xG his side have enjoyed. He has rightly pointed out how good this was. Unfortunately, however, from a lofty average for the season of 1.94, Sunderland’s most recent two league games have resulted in xG figures of 0.69 against MK Dons and 0.74 against Doncaster. This has driven the average down to 1.74. Still in pole position - for the moment.

The manager expanded upon his stats based commenting a little during last week’s pre-Doncaster press conference when he said:

The one stat we fall short on is the percentage of our shots on target.

And that’s a key one where we’re 17th in the league.

Keep that quote in mind as we continue.

Post-Crewe was the last time I presented the performance wheels. A lot of football has been played since, so it’s high time I updated you. First we’ll have a look at the wheels for Sunderland’s individual games played.

It has been a couple of months since I introduced these performance wheels, so a quick reminder won’t hurt. The red circle closest to the centre represents the 10th percentile (relegation performance for that attribute), the white circle is the 50th percentile, or median (mid-table performance) and the green outer circle represents the 90th percentile (automatic promotion performance).

If a side has scored greater than or equal to the 90th percentile, the outer segment is coloured either green, for an offensive attribute, or red for a defensive attribute. The perfect wheel would be entirely filled with black and all outer segments would be coloured.

The offensive performances against Portsmouth and Ipswich were pretty poor and the defensive performance against Portsmouth - as if we need to be reminded - was also poor.

Other than that, the wheels are fairly healthy. But I want to draw your attention to something that I will expand on as we continue. The wheels for five of the six games show a poor score for Shots (1 o’clock-ish).

Next we’ll have a look at the wheels for every League One side that covers the entire season to date.

The sides with the best performance wheels are Sunderland, Hull City, Portsmouth, MK Dons, Blackpool, and Plymouth (in that order). But these wheels do not account for differences in the quality of opposition faced to date. The table that follows, however, does.

Sunderland remain on top, even when their scores are adjusted for the strength of the sides they’ve faced. Also doing well are MK Dons, Portsmouth, Hull, Blackpool, and Doncaster. How good MK Dons were against Sunderland last week, and how Blackpool dealt with Peterborough on Saturday, came as a surprise to many. But it hasn’t been so much of a shock for me because they have both been riding high in the performance table since the beginning of the season - even when both were sitting in the bottom four in the real league table.

How wonderful life would be for Sunderland fans at the moment, if the positions in the league table matched the performance table. Rather than being a clear first, as above, Sunderland are sitting eighth.

Why are the apparently excellent (in numerical terms) performances not being converted into league success? Let’s begin to answer this by looking at Sunderland’s season performance wheel in a little more detail.

This paints an overall healthy picture. There is plenty of black and not much white to be seen. And 19 of the 24 available automatic promotion attributes are bagged. Great stuff.

But it is in three of the five attribute failures that we find the root of our problems converting these apparently stellar performances into results.

Cast your mind back to the earlier Phil Parkinson quote when he implied that not getting a sufficiently high proportion of our shots on target was our only problem. There is some evidence in the performance wheel to support that notion. Observe the offensive SOT spoke at 2 o’clock-ish. It extends to a little outside the white median circle. To be precise, the result is the 0.542 (or the 54.2%-ile). This indicates that Sunderland are a mediocre League One side for getting shots on target. A true automatic promotion contender would be aiming to better the 90th percentile green circle.

That is where my support for Parkinson’s shots on target theory ends, however.

To get a goal, there needs to be a shot on target. And to get a shot on target, there needs to be a shot. If Sunderland’s only problem was not getting their shots on target, they would be generating automatic promotion amounts of shots. Look now at the offensive Shots spoke. It’s a little tricky to pick out in the image where exactly it ends, so I will tell you that it is 0.479 (or the 47.9%-ile). That means Sunderland are generating the number of shots that would be expected of a mid-table League One side.

So Sunderland are actually getting about the expected proportion of their shots on target and, from that, scoring about the expected number of goals. To check this, we can take the average number of shots per game of 12.33 and multiply it by the expected conversion to a shot on target, then the expected conversion to a goal. What we get is 12.33 x 0.33 x 0.33 = 1.34. Sunderland’s actual goals per game average is 1.33 - it couldn’t be much closer!

Rather than setting an expectation that Sunderland will hit the target with an unreasonably high proportion of their mediocre number of shot attempts, the focus should be on creating more chances. But that might mean altering the style of play. Something that seems unlikely to say the least.