DSTV Premiership Recap and Justice League Table - Round 7

Two wins in four days saw Sundowns return to the top of the DSTV Premiership after last week's leaders, Royal AM lost at home to Richards Bay by a goal to nil. Bottom of the log, Chippa United started their new era (after the sacking of Daine Klate) with a shock win away at Amazulu. Wins for Stellenbosch, SuperSport and Pirates saw them all move closer to where we expect them to be based on their performances.

This is what the Justice League Table looks like after 7 rounds:

The only changes in the Justice League Table from last week are SuperSport United moving from 5th to 7th position despite a great away win at Maritzburg United on Friday night. Njoti had a great chance at 0-0 to to put the hosts ahead which he put narrowly wide of the target whilst Gavin Hunt's men scored from two headers to seal the points. A few weeks ago SuperSport were being terribly unlucky in their finishing and it seems as if their luck is finally starting to turn, with them scoring from two of their three shots on target in the game. What was noticeable was how many times Bashera got onto the end of final balls for SuperSport United and his patience was rewarded when he scored in the 73rd minute.

At the foot of the table, Chippa United may have won their game against Amazulu but they were quite lucky to come away with the three points. They were literally handed a goal in the form of a penalty and the 31st minute of the game summed up how lucky Chippa United are, with a shot against the crossbar from Chidi Kwem and then the rebound was saved miraculously by the feet of Kazapua, Chippa's goalkeeper. Mbenyane actually decreased the likelihood of scoring Chippa's 2nd goal in the 40th minute by taking a touch when he had an open goal to aim for but ultimately Amazulu couldn't convert more of their 21 shots in the game and Chippa held on, despite their dodgy defending.

Here is a visual representation of the strength of each team in the DSTV Premiership, both in attack and defence:

Despite being only two points off the leaders, Orlando Pirates' performances need to pick up if they were to have any chance of competing with Sundowns over the course of 30 games. Sundowns' closest rivals in that regard is still Kaizer Chiefs (despite their league position) and Cape Town City.

Now let's compare the xG from the Justice League to the actual goals scored on the PSL log and see which teams are over-/underperforming:

As mentioned before in the case of SuperSport Utd, it is clear that there is already regression to the mean happening in this season, with the difference between xG and actual goals becoming less and less. However, Cape Town City still have a relatively big difference between the chances they're creating and the goals they're scoring (scoring 0.42 goals per game less than we would expect from an average team). This was exemplified when they played 5th-rated Sekhukhune United last Wednesday evening. City's two best chances fell to Goedeman and my favourite, Khanyisa Mayo.

Goedeman has racked up an impressive 2.8 xG over his 620 odd PSL minutes in his career but has failed to score yet and he was desperately unlucky when he hit the inside of the post when 1v1 with the goalkeeper. His underlying stats of 0.41 xG per 90 and 0.01 xA per 90 are superb, the finishing can be coached and practised but having the ability to get into goalscoring opportunities (xG) isn't always something that can be coached - so he's definitely one to watch. The less said about Khanyisa Mayo's effort in the 74th minute, the better. He should have buried that and perhaps that is also an example of regression to his finishing mean because last week he smashed in a more difficult chance on his weaker right foot and this week he missed an easier one on his left - as long as he keeps on getting in those positions, the goals will come.

Tshegofatso Mabasa particularly impressed me in this game with his unique ability of allowing the ball to run across his body, enabling him to get multiple shots on goal. He was unlucky not to score because of a couple of fine blocks from City defenders. His xGi over his PSL career of 0.52 xGi per 90 is also very good (especially in the PSL where forwards average around 0.33 xGi per 90). His finishing can improve, as he is currently finishing about 16% less chances than the average PSL player.

Now let's compare the xGA from the Justice League to the actual goals conceded on the PSL log and see which teams are over-/underperforming:

Chiefs remain the best defensive team in the PSL and also the team that have the biggest disparity between xGA and actual goals conceded (Khune still trying to right the wrongs of Bvuma earlier in the season). Khune didn't have much to do against Marumo Gallants and couldn't do anything about the great volley sent past him by Makudubela.

At the other end of the scale, Magoola and Richards Bay are keeping up their unbelievable run of conceding less goals than the model expects from them. Royal AM had 17 shots of which seven were on target but they didn't manage to score. Richards Bay almost put one in their own net (perhaps the only way to beat Magoola these days) but the ball dropped agonisingly wide of the post, very lucky. I don't believe Richards Bay's run of results is sustainable but when it's going for you and you can rack up points, like they have then by all means.

Looking ahead, I expect a close game between the bottom two on Saturday and despite Chippa United playing at home, Gallants have a slight edge according to the model, due to a significantly better defence (don't be surprised if this one ends 0-0). 12th hosts 11th when Arrows meet Swallows in what should be a close one too with Arrows having the slight upper hand in this one, whilst Sundowns should get past Amazulu by almost double the chances that Amazulu will create. Richards Bay and TS Galaxy are too close to seperate and Sekhukhune and Kaizer Chiefs should be able to get past their opponents this weekend. It's however important to note that the goal of the model and these articles is not to predict the outcome of games, because variance in short-term results will always scupper those effort. The goal is to identify trends and over the course of a season (or sometimes two seasons) identify teams and players that are performing better or worse than what our eyes see when watching the games, using probability math and statistical modelling.

I'll be back after the weekend's games with updates on the model and hopefully some goals to report on. We have a two week international break so in this time I'll be looking at the best performing players of the season and taking a deeper dive into some of the candidates for the title and relegation. As ever, if you have any questions let me know and we can discuss them.