Friday, 17 October 2014

Championship Week 11: The quarter mile post

So having played 11 games and reached its second international break the Championship season is now basically a quarter of the way through.
So it seems an appropriate time to start looking at some of the league-wide trends and some new graphs for me.

First up, as always, is the real life table after those 11 matches.

Position Team Played Won Drawn Lost Goals For Goals Against Goal Difference Points
1 Norwich 11 6 3 2 21 10 11 21
2 Nott'm Forest 11 5 6 0 19 9 10 21
3 Watford 11 6 3 2 20 12 8 21
4 Derby 11 5 5 1 19 10 9 20
5 Middlesbrough 11 6 2 3 16 8 8 20
6 Ipswich 11 5 4 2 16 10 6 19
7 Charlton 11 4 7 0 14 10 4 19
8 Wolves 11 5 4 2 14 11 3 19
9 Sheffield Weds 11 4 5 2 11 8 3 17
10 Brentford 11 5 2 4 15 16 -1 17
11 Bournemouth 11 4 3 4 16 14 2 15
12 Blackburn 11 4 3 4 16 18 -2 15
13 Reading 11 4 3 4 15 17 -2 15
14 Leeds 11 4 3 4 11 13 -2 15
15 Cardiff 11 3 4 4 13 14 -1 13
16 Millwall 11 3 3 5 10 13 -3 12
17 Rotherham 11 3 3 5 9 13 -4 12
18 Huddersfield 11 3 3 5 12 20 -8 12
19 Brighton 11 2 5 4 10 12 -2 11
20 Birmingham 11 2 5 4 12 18 -6 11
21 Wigan 11 2 4 5 12 14 -2 10
22 Fulham 11 2 1 8 12 22 -10 7
23 Blackpool 11 1 3 7 6 14 -8 6
24 Bolton 11 1 2 8 8 21 -13 5

Norwich are sitting nicely top of the tree and their underlying stats back up that position very nicely indeed.
At the bottom we have Bolton and Blackpool, both with some horrible numbers but also, as we’ve been seeing, affected by awful luck (PDO) as well.
Fulham’s recent run of form has is genuinely deserved and were it not for having the second worst PDO score in the league the west London side would be very comfortably in mid-table.

I think it’d be quite nice to see what forces are driving the bus in terms of teams overall performance so far. Being a quarter of the way in to a seas you would hope that we are starting to see some performance (as opposed to luck) –based trends developing. But its still only 11 games in – by scientific principles still quite a small sample size.
So first thing I’m going to do is compare league points accumulated so far against shot share (Corsi/TSR).

What we’re looking for is an R2 value close to 1 (a complete positive correlation) or -1 ( a completely negative correlation). Anything around 0 means, well, its worth about 0.

As you can see, there’s barely any link at all between how teams are controlling play in general and how many points they have accumulated. An R2 value of basically 0.1 is pretty weak to say the least.

So let’s try doing the same only using shots on target share instead.

This looks a bit more encouraging, although it really isn’t strong enough to say that this measure has a massive baring on the results teams are seeing so far.

In that case, let’s have a look at PDO vs points won.

Well helllloooooo!
Compared to the other two that is pretty damn high.
Again it’s not a perfect match, but what it serves to show us is that at the moment points won and hence league position are still being dominated by luck factors.
Hopefully, as we follow the season through, this score will reduce and the shot share and sots on target share will increase. (Hopefully!)

So if we do see a regression back to the mean (of 100) of PDO scores, how can we expect the season to play out if teams keep performing as they are?
Well, being as it has a slightly stronger link to points at the moment, lets look at shots on target share vs PDO to see where our teams are.

The top two (Norwich and Forest) are green coloured dots, the four play-off teams are yellow, the three in the relegation zone are red, and the rest are blue.

There are a couple of very noticeable points that are due some regression one way or another.
First is the very top one – Forest.
With a very average share of the shots on target and a sky high PDO, Forest seem due for a fall down the league when the luck runs out.
Already we’re seeing some disquiet among Forest fans as the club has hit a dry spell in front of goal. It could get much worse for the early leaders.

In an opposite position is Brighton.
Having a great share of the shots on target but rotten luck, it’s just a question of how long Sami Hyypia can remain in charge at the Amex to benefit from the regression.
Once that kicks in, assuming these good underlying numbers are not purely due to score effects (or some horrible tactical inefficiency at one end or the other), the Seagulls could well race up the table.
And if they’re not too far off by then, even make a challenge for the playoffs or promotion.

In the left side of the graph we have the teams that are struggling.
As I mentioned earlier, despite the recent upturn in results Fulham are still being hit by brutal luck. Once that begins to reverse the west London side should be fine.

Looking on the good luck side of the x-axis, the two teams most in danger of joining Bolton and Blackpool are Leeds and Wigan.
Remarkably similar underlying numbers but Leeds managed to scrape a few more points together.
A sharp turnaround is needed in performance or this pair is going to find themselves sucked into a relegation battle shortly.

And teetering on the precipice is Birmingham.
Getting just 40% of the shots on target with a dead even PDO score (the only team in the league to do so), it’s safe to say that the Blues are right where they deserve to be.
It’s a relegation fight that could go right down to the wire again for them.

Too hot

There are a few fascinating games (statistically speaking) this weekend in the Championship.
Number one for me is down at the Amex where Brighton and Middlesbrough face-off.
Two teams that love controlling possession and a few weeks ago were both suffering the pain of harsh PDO.
Middlesbrough has since recovered, when will Brighton?
That should be a good one.

Too cold

Then comes the match at St Andrews and two teams with the opposite problem – neither one can control a game if its life depended on it.
This could be interesting for other reasons. Along with the obvious impact at the bottom of the table, it’s Neil Lennon’s first game in charge for Bolton.
Will Wanderers be lifted by this? Will Lennon’s tactics bring an improved performance?
And will Birmingham get sucked in to the relegation mire?
Of course a win for Birmingham brings them some much needed breathing space and would be a real blow to Bolton.
And surely someone has to get a shot on target!

Just right

Finally from my highlight trio is Cardiff vs Forest.
Looking at the table this seems a real mis-match, but going by shots on target share, this could be a very close game.
Both teams are basically at 50% of the shots on target share each game.
Forest have by far the stronger overall shot share (Corsi/TSR) numbers, but there’s very little difference when it comes to getting shots on goal.
And this of course will be Russell Slade’s first game in charge at the CCS – taking over after a particularly listless performance at Blackpool.
Does Forest’s PDO regression start in south Wales?

Dear Bill,

As an aside (or maybe rant is better), I’ve not done any graphs and charts work on Excel in a long time… like more than a decade ago when doing my degree, so this post has involved a lot of re-familiarising myself with that.
Or in fact, re-learning. Because Microsoft seems to have taken what was a relatively simple process in those earlier versions of Excel, and twisted it to become ridiculously complicated instead.
This to me feels like the sort of thing you have to intend to do, to make it that awkward. But still, I guess you learn something new everyday!

No comments:

Post a Comment