In August we unveiled a new model, using xGoals, to be used to analyse the Scottish Premier League throughout the season. We identified some early overachievers and underachievers, and had a look through the data to see what explained each team's performance.
Today we are going to see how the teams are progressing under our model, and have a look at a few notable performers.
Our xG model can be used to give a probability of each team winning a given match, based on the chances that occurred during the game. This also means that for each team, we can give an expected points tally for each match (see the August article for a fuller explanation).
If we compare this to the actual points tallied accrued, then we can see whether the table is an accurate portrayal of each team's performances. The graph below shows expected points against actual points. It should be pointed out that for the four teams who have a game in hand (Aberdeen, Celtic, Motherwell and Rangers), we have multiplied their tallies by 13/12 to make them comparable to the other teams.
The first thing to be noticed is that the league is splitting into a top 5 and a bottom 7. Given that the SPL actually splits into two groups of 6 near the end of the season, it's likely that that group of 7 will be fighting to be the one team in the top group.
Because the better teams should have better finishers and goalkeepers, we would expect them to generally be above the line, and the smaller teams should probably appear below the line. This is generally the trend we see here, with three of the top 5 comfortably above the line, and with three of the bottom 7 clearly below the line.
The most notable team on the graph are Dundee. We highlighted in August that Dundee were very unlucky to be bottom of the table, and really should have been safely mid-table. Well, that remains very much the case. Dundee remain bottom of the table, but our expected points tally actually puts them top of that group of 7, in theory meaning they are performing the best after Motherwell.
In August we also noted the overperformance of St. Johnstone, who were performing more like a mid-table side. Reality has now caught up with them, as they sit in the middle of the table, although they remain the side in the bottom group who are overperforming the most. St. Johnstone's regression to their expected points tally will be a comfort for Dundee, who will hope they do the same soon.
So far, we have been considering teams' xG as a bulk. But there are two ways to build xG - through taking lots of shots, or through focusing on creating high-quality chances. By comparing the number of chances teams are having with the quality of these chances, we can see what approaches each team are taking. Again, we have slightly modified the values for the teams with a game in hand.
Of the teams in the top group, it's interesting to see Dundee here, with an attacking profile close to Motherwell, one of the league's top clubs. We certainly can't blame Dundee's attacking tactics for their underperformance, given they are creating chances of the same quality as St. Johnstone, but more plentiful. This provides more evidence that these two Tayside teams are more similar than the table would have you believe.
In the bottom group, there are only two teams with low shot quality who are making up for it with a high volume of shots - Celtic and Hibernian. It is surprising that Celtic is in the lower group, as we would expect the league's best club, with many talented creative players, to be able to fashion high-quality chances frequently. Celtic's low position here may be a result of their opponents regularly defending deep and trying to make Celtic's chances as difficult as possible. In fact, we can see that there's barely any difference between Celtic and Rangers' total xG (shown by the grey contours), despite Celtic having 41% more shots than Rangers.
Meanwhile, the graph also highlights trouble for Partick, who are creating the least chances, which aren't of a high quality, leaving them well out on their own as the least threatening side.
We can look at the same chart as above, but for chances conceded. The higher a team appears on the graph the better-quality chances they are conceding, and the further to the right the more chances they are conceding.
Following Celtic is a close cluster comprising the teams just behind them in the table. Rangers are the one out of these four teams who have the highest xG against, which coupled with their high xG for means they are at least an entertaining side to watch.
At the right side of the graph is a group of four teams who are conceding lots of shots. This group contains both Dundee and St. Johnstone, further suggesting these two teams are quite similar in terms of xG. Dundee's defence is certainly their weakness, and the main reason they are in trouble at the moment.
However, there are still two teams with worse xG against than Dundee. Firstly, there is Partick, whose leaky defence coupled with their toothless attack is why they are firmly bottom of our expected league table. The other side is Hamilton, who are a curious case as they aren't conceding a stupidly high number of chances, but those that they do concede are of a high quality - significantly higher than anyone else in the division.
By plotting each team's average expected points across the season, we can see which teams are improving and which are declining. This is shown on the graph below.
Celtic have stayed clear of the rest since Week 3, although there have been a couple of times Aberdeen and Rangers have got close to them. Currently, though, it looks as though that will be a battle to finish in second behind Celtic. The momentum is currently with Rangers, as Aberdeen's performances appear to be slightly declining.
There is a tasty battle between Hibernian and Motherwell for 4th, with current form slightly favouring the Edinburgh club. There is even a chance that one of these two could challenge for a top 3 spot with a good run of form.
The next group sees six clubs all very, very close together. As we earlier mentioned, Dundee should be leading this group, but luck doesn't appear to be with them at the moment. In terms of longer-term trends, St. Johnstone may have started well, but they have been gradually declining, and could easily be dragged into a scrap towards the bottom. Meanwhile Kilmarnock should be happy with their recent improvement in performance, as they looked particularly poor around weeks 3-6.
And, finally, there's Partick, who look to be in big trouble, as all the previous graphs have helped demonstrate. They are marginally improving, but they have a long way to go before they are playing at the level of their competitors.
Reality vs. expectations
It's all well and good saying that a team is expected to score 2 goals a match, but this is just a prediction, and natural variation will occur - as well as factors not covered by the model, such as how good each team's players are at shooting (quite an important part of football).
If we divide each team's actual goals scored by their xG, then this tells us how much they are over/under-performing when it comes to converting chances. Similarly, we can do the same for goals conceded to see how ruthlessly each team is being punished by opponents. Plotting both of these gives us the graph below.
Celtic are sitting pretty in the bottom-right quadrant, as they have scored 15% more goals than expected, whilst conceding 20% less goals than expected. Partick can also be seen in this quadrant, as they have slightly overperformed in both metrics so far, explaining why they are currently not in the 12th place their performances have merited.
The obvious outlier, however, is unsurprisingly Dundee. They are slightly unlucky in defence, in that their opponents have scored 8% more frequently than expected, but the real story is Dundee's finishing. They are only scoring 71% of the goals that they ought to have, and this is the reason they are struggling at the bottom of the table. Either Dundee need to change their personnel up front, or they need to put in more hours in front of goal on the training ground.
For those who don't follow Scottish football, it's likely the only story you've heard about is Celtic's remarkable unbeaten run under Brendan Rodgers. This is impressive, although despite this run Celtic aren't as far away from the competition as you'd perhaps expect. Beyond this, there are a number of interesting battles shaping up.
The most intriguing story is Dundee's underperformance. Our model has them performing significantly higher than they have been, and if they sharpen up their shooting, there's a good chance we'll see Dundee turn their season around.
Other than Dundee, the teams have performed largely as we'd have expected following our August article. Will we see any teams make a break from their performances thus far and move up/down the league? Only time will tell.