There has been a lot of good work to highlight, so I’ll just dive in with a set of recommended reading.

Game States

Like most of the best analytics writers, Ben Pugsley is criminally under-followed, but his work of late for the Bitter & Blue SB Nation Man City blog has been fascinating. The primer post is here. The pertinent bit:

A very high 43% of all PL goals scored this year have been scored at a tied game state, thus breaking the draw and propelling one team into the lead. There are some interesting things going on here.

We know about goals scored at a tied game state but look at the drop off from evens to +1. There’s a whopping 57.98% decrease in goals scored between evens (going into the lead) and +1 (extending the 1 goal lead to a 2 goal lead).

Now, let’s look at the gap between +1 and +2 (a 2 goal lead to a 3 goal lead) The drop off in goals scored is 63.62% between those two game states.

The percentage decrease between evens and +1, and +1 and +2 is large and not too disimilar. In short the chance of scoring a goal falls off massively once a team has the lead, and falls off again when a team already has a 2 goal lead.

Overall 80.36% of all goals are scored at the crucial -1, evens, +1 game state.

Here again is an example of the benefit, as Simon Gleave put it to me last week, of walking before we run in football analytics. It also empirically underlines what most of us have known about football for ages: that most of the time, a two goal deficit is a game killer. You can read more of Pugsley’s work on Game States and other topics here, here and here.

Man United’s points total versus TSR, PDO

That last link is particularly interesting because one of the major mysteries this season has been how exactly Manchester United have managed to overcome a comparatively low Total Shots Ratio (a measure of team dominance) and a PDO on par with Manchester City. Pugsley gets into the thick of it:

Man City win the shots and TSR battle by some distance. United have scored 6 more goals than their cross-town neighbours and that is off the back of a ridiculously high scoring%.

United’s scoring% could be driven by talent, luck, both, neither, or Ferguson’s voodoo corner and crossing magic. We won’t be able to discover what is driving that scoring% (or nicking their save%) but it is covering a lot of gaps in Man United’s form.

Both clubs have an almost identical PDO’s, but the United PDO is scoring% driven and City’s is save% driven. Have both teams been unlucky to an extent, or do the constituent parts of their respective PDO’s tell us a little about the strength and weaknesses of each squad?

One of the questions that’s been annoying me about PDO, the percentage of shots on target that go in added to the percentage of shots on target saved, is at what point ‘luck’ ends and incredible individual or team skill at the higher echelon begins. Manchester United’s scoring percentage is a whopping 16%. It has been high all season and regressed only a little since 2012. Surely this can’t all be luck.

As Puglsey notes, whatever it is, we can safely say United are preternaturally good at scoring when offered the opportunity.

Points per pounds

I would classify myself a skeptic on making direct links between wage spending and transfer spending to table position, as in Total Team Value and other exhaustive models.

James Grayson has taken a look himself, using a much simplified model to ascertain expected points per pound spent on transfers and wages. You can read the results here.

Chelsea and Man City top the list, followed by Man United and Liverpool, Arsenal and QPR. To say it is roughly commensurate with the current table, or the table we’re likely to see at the end of the season, is the equivalent of saying the picture my son drew yesterday resembles my face. The outline is there, but it’s way, waaaay off.

Perhaps—worlds collide!—the TTV crowd might have a stylistic answer to Grayson’s model. I’d read it…