Championship Week 4: The fledgling fall of Fulham and Leeds United

I'd planned to give it another couple of weeks before publishing my first hit of Championship data but due to popular demand (who'd have thought that) I'll do it now.
Looking at the early results and table there are obviously a couple of big stories going on in the Championship right now at the bottom end - Fulham and Leeds.
Here's the table as it looks in the real world.
Not exactly happy reading for the two teams that play in white - three if you include Bolton in that too.

Pos	Team	P	W	D	L	GF	GA	GD	Pts
1	Nottingham Forest	4	3	1	0	10	3	7	10
2	Watford	4	3	0	1	9	4	5	9
3	Norwich	4	3	0	1	7	2	5	9
4	Wolverhampton Wanderers	4	3	0	1	3	1	2	9
5	Charlton	4	2	2	0	7	5	2	8
6	Sheffield Wednesday	4	2	2	0	5	3	2	8
7	Derby	4	2	1	1	8	4	4	7
8	Cardiff	4	2	1	1	5	3	2	7
9	Millwall	4	2	1	1	4	2	2	7
10	Blackburn	4	2	1	1	7	7	0	7
11	Bournemouth	4	2	0	2	8	5	3	6
12	Middlesbrough	4	2	0	2	6	5	1	6
13	Brighton	4	2	0	2	4	3	1	6
14	Rotherham	4	2	0	2	2	3	-1	6
15	Brentford	4	1	2	1	4	4	0	5
16	Birmingham	4	1	2	1	4	5	-1	5
17	Ipswich	4	1	1	2	4	5	-1	4
18	Wigan	4	1	1	2	4	5	-1	4
19	Reading	4	1	1	2	4	8	-4	4
20	Huddersfield	4	1	1	2	4	9	-5	4
21	Leeds	4	1	0	3	2	8	-6	3
22	Bolton	4	0	1	3	4	9	-5	1
23	Blackpool	4	0	0	4	2	7	-5	0
24	Fulham	4	0	0	4	2	9	-7	0

But what do our metrics tell us about how the first weeks have gone?
Both Fulham and Leeds are much hyped "big clubs", with foreign owners and managers who have come into the season under a massive amount of scrutiny for one reason or another. The data is very raw and still open to much variation but, at the moment it looks like one of these will have a happier ending than the other - assuming owners don't start doing wildly unpredicatable things like sacking managers five games in.
So with the caveat of the very volatile nature of the data in our minds - let's have a look at it.

Team	Position	Shots for total	Shots against total	Corsi	Shooting % For	Save %	PDO
Nott'm Forest	1	60	43	0.583	47.62	70.01	117.62
Watford	2	58	65	0.472	52.94	73.34	126.28
Norwich	3	76	47	0.618	25.00	83.33	108.33
Wolves	4	51	42	0.548	30.00	88.89	118.89
Charlton	5	50	46	0.521	36.84	70.59	107.43
Sheffield Weds	6	48	52	0.480	33.33	76.92	110.26
Derby	7	54	59	0.478	40.00	83.33	123.33
Cardiff	8	40	57	0.412	26.32	76.93	103.24
Millwall	9	50	55	0.476	26.67	86.67	113.33
Blackburn	10	45	62	0.421	46.67	72.00	118.67
Bournemouth	11	68	54	0.557	33.34	68.75	102.09
Middlesbrough	12	67	36	0.650	31.58	54.55	86.13
Brighton	13	63	37	0.630	20.00	62.50	82.49
Rotherham	14	52	42	0.553	13.33	70.00	83.33
Brentford	15	46	57	0.447	25.00	78.95	103.95
Birmingham	16	48	60	0.444	30.77	76.19	106.96
Wigan	17	36	40	0.474	57.16	61.54	118.69
Ipswich	18	62	50	0.554	26.67	66.67	93.34
Reading	19	43	49	0.467	44.44	52.94	97.38
Huddersfield	20	60	56	0.517	17.39	57.14	74.54
Leeds	21	27	59	0.314	25.00	46.66	71.66
Bolton	22	45	84	0.349	44.44	68.96	113.41
Blackpool	23	46	46	0.500	18.18	46.15	64.34
Fulham	24	59	56	0.513	13.34	59.10	72.43

First we see that at the moment the table follows PDO scores more closely than Corsi, a good indication that the sample has yet to settle down fully and this is still "early days".
However, there is some movement towards the more repeatable skills coming through.

We can see Fulham have been comfortably mid-table in terms of their share of the shots per game (Corsi) but have been getting absolutely hammered in the "luck" area (PDO).
Now, some of this outshooting might be coming through what's known as score effects - being behind early on in game a team will press forward looking for an equaliser, while the team leading will often go into a defensive shell to protect the lead. (Whether this defensive shell approach actually works is very open to debate.)
However, even if this is the case, it shows that Fulham are at least capable of dominating games and getting more shots away than the opposition - an encouraging position to be in at any point.

The key part in Fulham's predicament at the moment is that they simply can't buy a goal for love nor £11m of Ross McCormack - although the League Cup winner in midweek against fellow Championship side Brentford may get him going.
Does anyone really expect Fulham to still be scoring just 13% of the shots on target at the end of the season? That's less than half the league average of around 30%.
Defensively the west London team have also been burned - somehow conceding goals at double the league average - almost 60%.
Of course both these could be down to the quality of chances that Fulham are creating and conceding - but to have this wide a disparity must surely be very unlucky. (Consider it the reverse of when a West Brom or Sunderland or Swansea or other Premier League "minnow" is up near the top of the table a couple of months in - eventually every shot they take stops hitting the back of the net, opposition shots curl just inside rather than outside the post, and the team slides back down to its more natural (playing talent-based position somewhere in the middle of the table).



Now let's look at Leeds.
Sure the Yorkshire club have also been burned by the forces of luck with a PDO slightly worse than Fulham (and second only to Blackpool in the "Oh my god its how low?" stakes).
In this case its keeping the ball out of the net that's the biggest problem. For what it's worth Leeds may as well just brick up half the goal at one end of Elland Road and use that instead of a goalkeeper - it'd probably be more effective.
But, and here's the kicker, Leeds also has the worst Corsi rating in the league - and it's not even close.
On average Leeds takes less than a third of the shots in each game. So if Leeds muster 3 shots in a game the opposition has been pummelling six or even seven at the United goal.
Again, the data here is extremely early in the season - tactics can change and Leeds may have just been unfortunate enough to play against the better teams in the league so far, but this is most definitely not a good trend to be in.
Sure the luck may turn but even with a 100+ PDO over the course of a season it would take a minor miracle for Leeds to avoid relegation if this Corsi score stays the course.


Other notes:

Bolton - Yikes!
THAT Corsi with THAT high PDO and the Trotters are still only third bottom. This could get ugly quickly for Dougie Freedman if that continues. (Why on earth he left Crystal Palace - apart from a likely pay rise - I've got no idea.)
Today's game between Leeds and Bolton is already looking like a relegation six-pointer.

Also Watford - hmmmm. However I suspect that Corsi score has A LOT to do with the pummelling they took at the hands of Norwich when down to ten men very early on.

Middlesbrough and Brighton - again both burned by PDO but could be very strong teams if they maintain good Corsi rates and PDO regresses back up to 100, as would be expected.

And finally.... Cardiff.
The Bluebirds have the third worst Corsi score so far this season and one that if maintained would likely see a rapid descent down the table.
But, I suspect score effects and Ole Gunnar Solskjar's tactics have a role to play here.
In the three matches Cardiff got points the Bluebirds scored first and then largely sat back to defend.
The Wolves game however should serve as a bit more of a warning that getting outshot will eventually come back to bite you.
Cardiff visit Craven Cottage today. With Ross McCormack being an ex-Bluebird and Fulham due (at some point) a dose of good luck, as a Cardiff fan I fear the worst.


PS - I'm still working on getting the tables to be sortable, so you can manually click around investigating the stats at your leisure. Sadly, Blogger (or perhaps its my very poor html skills) is not cooperating.
Hopefully in future weeks I'll get the tables looking and working a lot smoother.

The real tale of the 2013/14 Championship season

The Championship kicks-off this weekend and I like that it has the first set of fixtures all to itself without any Premier League games going on.

But it takes at least a few weeks of games before the stats we’ll typically be dealing with to become useable or have anywhere near enough of a sample size to be significant.

So to start things off I thought I’d look back at last season in the Championship and see just who deserved to get relegated and promoted and if we tell a wider story of the campaign.

To do this I’m using data from WhoScored and FootStats.

Sadly not all the raw data I wanted is available there, but by using shots per game for/against and multiplying by 46 I can get to pretty much the right number – certainly suitable enough for this exercise.

First, here’s a reminder of the final table from the 2013/14 season:

Team	Pos	P	W	D	L	GF	GA	GD	Pts
Leicester	1	46	31	9	6	83	43	40	102
Burnley	2	46	26	15	5	72	37	35	93
Derby	3	46	25	10	11	84	52	32	85
Queens Park Rangers	4	46	23	11	12	60	44	16	80
Wigan	5	46	21	10	15	61	48	13	73
Brighton	6	46	19	15	12	55	40	15	72
Reading	7	46	19	14	13	70	56	14	71
Blackburn	8	46	18	16	12	70	62	8	70
Ipswich	9	46	18	14	14	60	54	6	68
Bournemouth	10	46	18	12	16	67	66	1	66
Nottingham Forest	11	46	16	17	13	67	64	3	65
Middlesbrough	12	46	16	16	14	62	50	12	64
Watford	13	46	15	15	16	74	64	10	60
Bolton	14	46	14	17	15	59	60	-1	59
Leeds	15	46	16	9	21	59	67	-8	57
Sheffield Wednesday	16	46	13	14	19	63	65	-2	53
Huddersfield	17	46	14	11	21	58	65	-7	53
Charlton	18	46	13	12	21	41	61	-20	51
Millwall	19	46	11	15	20	46	74	-28	48
Blackpool	20	46	11	13	22	38	66	-28	46
Birmingham	21	46	11	11	24	58	74	-16	44
Doncaster	22	46	11	11	24	39	70	-31	44
Barnsley	23	46	9	12	25	44	77	-33	39
Yeovil	24	46	8	13	25	44	75	-31	37

At the top Leicester romped to promotion with Burnley nine points behind. QPR won the play-offs after finishing five points behind third place Derby.

At the bottom Yeovil and Barnsley were cut adrift while Doncaster suffered relegation on the last day after Birmingham’s amazing comeback.

All the numbers would seem to make sense: teams that scored the most and conceded the fewest got promoted, those that did the reverse were relegated, and everyone else was in-between.

But did the best teams really get promoted?

And how bad were the relegated trio.

Let’s start with measuring the balance of play (in the form of shots taken) that each team accounted for over the course of the season.

All other things being equal, to succeed a team wants to be outshooting its opponents regularly.

To measure this we use Corsi or Total Shots Ratio (TSR) – I prefer Corsi as it’s where I first became familiar with the context, so I’m going to stick with using that, plus I like the sound of it better.

(I could go into a deeper explanation, but for the case of brevity I suggest you read the Stats Bomb definitions piece linked to on the sidebar if you want to know more.)

Team	Pos	Corsi
Leicester	1	0.58
Burnley	2	0.52
Derby	3	0.54
Queens Park Rangers	4	0.54
Wigan	5	0.55
Brighton	6	0.52
Reading	7	0.46
Blackburn	8	0.53
Ipswich	9	0.53
Bournemouth	10	0.52
Nottingham Forest	11	0.5
Middlesbrough	12	0.53
Watford	13	0.5
Bolton	14	0.48
Leeds	15	0.51
Sheffield Wednesday	16	0.53
Huddersfield	17	0.52
Charlton	18	0.44
Millwall	19	0.49
Blackpool	20	0.47
Birmingham	21	0.45
Doncaster	22	0.42
Barnsley	23	0.46
Yeovil	24	0.41

What does this table tell us? Well most of the teams’ Corsi values seem to line up fairly well alongside their actual finishing position.

As we can see, Leicester really did dominate, typically taking 58% of the shots in a game – way above what any other team could muster.

QPR were second best in that regard, and so from a Corsi perspective at least, worthy of that promotion place.

Burnley, however, finished joint ninth along with Brighton, Bournemouth and Huddersfield. So how did the Lancashire side manage to achieve such a clear second place when Brighton only made the playoffs and the other two were stuck snugly in mid-table?

We’ll look at the main reason for this later.

At the other end of the table, Yeovil and Doncaster were clearly the two worst teams in the division in terms of shot metrics - and by a long way.

But Barnsley’s 46% result, while not world beating, is better than Birmingham and Charlton, and the same as Reading, who all finished higher.

In Reading’s case, this would appear a massive anomaly that stands out and needs to be further investigated.

So let’s do that.

Another valuable measurement in analytics is PDO. It’s simple, but can suggest whether a team’s performance is sustainable or due largely to luck. Again, I recommend reading the Stats Bomb link, but in short, it adds up the team’s save % (saves made from shots on target faced) and shooting % (goals scored from shots on target taken).

Numbers will be clustered around 100, with those further away suggesting a greater element of luck – be it good or bad.

Team	Pos	SH% for	SV% for	PDO
Leicester	1	33.74	73.62	107.36
Burnley	2	33.03	77.16	110.19
Derby	3	36.84	69.23	106.07
Queens Park Rangers	4	29.41	74.27	103.68
Wigan	5	29.9	73.48	103.38
Brighton	6	31.25	77.27	108.52
Reading	7	39.33	76.47	115.8
Blackburn	8	30.04	66.84	96.88
Ipswich	9	29.56	66.46	96.02
Bournemouth	10	29.91	65.8	95.71
Nottingham Forest	11	31.46	65.41	96.87
Middlesbrough	12	28.57	71.43	100
Watford	13	34.42	69.95	104.37
Bolton	14	27.06	70.44	97.5
Leeds	15	32.6	63.78	96.38
Sheffield Wednesday	16	32.31	65.97	98.28
Huddersfield	17	28.86	67.98	96.84
Charlton	18	25.47	69.5	94.97
Millwall	19	25.41	64.93	90.34
Blackpool	20	26.57	64.13	90.7
Birmingham	21	29.74	68.91	98.65
Doncaster	22	26	70.59	96.59
Barnsley	23	25.88	67.78	93.66
Yeovil	24	24.31	72.12	96.43

Again, these figures line-up reasonably well with the table, but they really help explain some of those anomalies.

The biggest candidate here is Reading. Riding a massive PDO score of 115.80 (a full five points higher than any other team) we can see the Royals’ playoff push was driven by sky high shooting and save percentages.

This also helps explain a significant part of Burnley’s success too.

Unfortunately, there’s very little evidence that such high PDO scores are repeatable – i.e. that they are part of a hidden talent or certain tactic.

This leads to the suggestion that unless things change dramatically for both Reading and Burnley, this season could be a much harder fight for them in their respective divisions.

Lower down the table, it becomes clear that Blackpool and Millwall were screwed by PDO. The upheaval at Blackpool this summer means the team is highly uncertain, but both these sides deserved to finish a good few places higher than they did.
Barnsley too can count themselves somewhat unlucky to have been thrown so deep into the relegation tangle.

So to sum up the 2013/14 Championship season: Leicester were head and shoulders above the competition with QPR’s place in the Premier League, albeit via the playoffs, a valid one.

Burnley’s promotion party was fuelled by a mountainous PDO score, as was Reading’s ultimately unsuccessful playoff push.

Yeovil and Doncaster were just epically bad, while Barnsley fans can consider themselves just a little unlucky to be watching their team in League One this season.

As for Blackpool and Millwall; as far as they are concerned the only way is up for their PDO – probably.