Using analytics to look at the what’s the true reality for teams in the Scottish Championship

Mark Warburton, Mark Burchill, & David Weir

written by – Rangers Report

Last month we used Expected Goals Ratios & PDO to try & make sense of the how well teams were performing this season in the Scottish Championship.   Among the objectives of applying analytics to football is to get a sense of which teams are outperforming their underlying numbers (& vice versa, of course).  Basically which teams are destined to get better then their current results & which teams are in line for a decline in results.

In that post, there were two claims that the evidence suggested would become reality.  One was that Livingston was beginning to turn their season around & that they actually were a better team then their record suggested.  Since that last post – they have beaten Raith Rovers, earned a draw against Rangers, & lost close games with Morton & Hibs.  Four points out of four matches is progress, especially given the level of competition.  Let’s just say, their draw with Rangers wasn’t that big of a surprise if you paid attention to the fancy stats.

The other conclusion was that Falkirk was due to regress given their high PDO & the fact that their actual goal differentials were outperforming their Expected Goals Ratios.  Well, both numbers went up even more in the last month as Falkirk went undefeated in four matches (three wins & one draw).  That’s the thing about fancy stats like PDO…Falkirk’s play is destined to regress but sometimes it may take months, not weeks, for it to happen.  They are on a hot streak that is off the charts & will be difficult to sustain all season long.  Recently, we broke down the elements of PDO & wrote – “Falkirk is scoring on 41% of their shots on target & their goalie is saving 78% of the shots on target.  Both are much higher than the norm….are they too good for the competition or just too good to be true?”

After noting the play of Livingston & Falkirk, I felt it was time to revisit the stats a month later for the entire league & see which teams are beginning to trend upward & which are going he opposite direction.  You can see the similar post from October here.

Teams are ranked below based on Expected Goals Percentage (a reminder that xG% represents the percentage of total Expected Goals a team gets per game.  For example, when you combine Rangers xG for & xG against it is 2.58.  The fact that they earn 2.01 xG a game means thay get 78% of the xG in a game on average — xG/(xG +xGA).

Team xG per game xGA per game xG% GF per game GA per game GF% GF% – xG% PDO
Rangers 2.01 0.57 0.78 2.86 0.64 0.82 0.04 1.117
Hibs 1.73 0.72 0.71 1.87 0.67 0.74 0.03 1.081
Falkirk 1.20 0.98 0.55 1.93 0.87 0.69 0.14 1.182
Raith Rovers 1.37 1.13 0.55 1.27 1.40 0.48 -0.07 1.011
Livingston 1.12 1.04 0.52 1.13 1.60 0.41 -0.10 0.892
Morton 1.22 1.45 0.46 1.00 1.00 0.50 0.04 1.034
Dumbarton 0.77 1.01 0.43 1.00 1.93 0.34 -0.09 0.898
Queen of the South 0.95 1.25 0.43 1.06 1.44 0.42 -0.01 0.929
St Mirren 0.75 1.22 0.38 0.93 1.53 0.38 0.00 0.974
Alloa Athetic 0.63 1.91 0.25 0.47 2.27 0.17 -0.08 0.836

Now, let’s hone in on a few numbers.  First, PDO as it relates to xG% & then we’ll focus Goals For Percentage minus Expected Goals Percentage.  This number is particularly intriguing, because like PDO, teams should naturally gravitate towards certain normed expectations.

PDO v xG November 29, 2015 (1)

  • Notice that most of the teams are hovering around the league average, while you have four outliers – Falkirk, Rangers, Hibs, & Alloa.
  • First, Rangers & Hibs – given their high xG% you can make the argument that they are simply that much better than the rest of the league. Both teams get more then 70% of the Expected Goals in a game & the next best teams are at 55%.  The fact that they are both 14 points ahead of the fourth place team supports these numbers.
  • Then there’s Falkirk.  Their xG% is tied with Raith Rovers & only slightly ahead of Livingston.  But there they are, only six points behind Rangers & Hibs while eight points ahead of the fourth place team.  The fact that they have the highest PDO in the league suggests they soon may make that shift into the Lucky & probably not that good category with an emphasis on ‘not that good’.

What caught my eye when comparing the charts of stats from October to November was the GF%-xG% differentials.  Goals For Percentage is the same concept of xG% but with actual, real life goals.  Teams with a higher GF% then xG% are outperforming what the projections suggest they should do, while teams in the negative are underperforming.  It seems natural that teams with high differentials will see their actual goal differentials regress to the expected rates & conversely the teams in the negatives could be due for an uptake in actual goal differences.

The following graphs the GF%-xG% percentages from where they were at the end of October  & where they are now.

Using GF%-xG% to measure form

  • Notice that of the five teams that had scored higher GF% then xG%, four began to regress towards the middle of the graph.
  • Teams like Rangers & Hibs kept performing well (especially Hibs) but saw their actual goal difference get closer to their Expected Goal differentials.
  • Raith Rovers’ numbers just bottomed out, as they only gained two points in their past four league matches.  Meanwhile, St. Mirren’s numbers completely evened out – which is not a good sign  considering they, like Raith, only gained two points in four games.   Given Raith’s dramatic drop in GF-xG% you’d expect them to bounce back back.  But for St. Mirren the harsh reality is that the results probably won’t get much better then their current form.
  • Falkirk appears to be setting up for a mighty crash back to reality.  The disparity between their GF% (69%) & their xG% (55%) likely is not sustainable.  The question now is when will that bubble burst?  This kind of skewed production can go on for a while & the good news for the Bairns is that they have built themselves a real nice cushion in the league table when it comes to earning a play-off spot.
  • Livingston & Queen of the South are trending in the right direction & the fact that they have combined for nine points in the past month supports the underlying stats.  Furthermore, you can see why I predicted that Livingston were destined to get better.  Their differentials were too drastic to ignore – they are playing decent football & their results are beginning to match that.
  • Alloa is bad & getting worse.  This is where a collective lack of confidence amongst the players is probably making a bad team worse.

Lastly, I’ll leave you with a couple of graphs that now compares PDO to GF%-xG%.  Given that both sets of numbers should naturally move towards their averages it will be interesting to see the results.

GF%-xG% v PDO October

GF%-xG% v PDO November

That average number seems to act as a magnet – gradually luring teams back to a normalized ‘reality’.

If PDO is supposed to measure luck, the GF%-xG% seems to represent how close to reality a team is performing.

Teams like Livingston & Dumbarton are being pulled in the right direction, while clubs like Rangers & St. Mirren are coming back to earth while being pulled closer to reality.  Of course, the reality for St. Mirren is far different then the reality for Rangers.

You can follow Rangers Report on Twitter @TheGersReport 


3 thoughts on “Using analytics to look at the what’s the true reality for teams in the Scottish Championship

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s