How do St Mirren compare to the rest of the SPFL Premiership?

 How do St Mirren compare to the rest of the SPFL Premiership?

Warning: This article is statistics heavy. 

Expected Goals definition - 

Expected Goals (xG) works by "taking all shots by a team in the course of the game and looking at the probabilities [of a goal being scored]" (Biermann, 2019: 33). If you remember the beauty Ethan Erhahon scored at Pittodrie, a goal of the season contender, it has an xG of 0.02 - meaning that for every 100 shots, the model would expect a goal to be scored twice. A penalty has an xG of 0.76. The model is smart enough to distinguish between shots and headers, different body angles, and the proximity of defenders and the goalkeeper. Not to say xG is flawless, certainly not, but the limitations and benefits of it are best reserved for another article.

NB: Most other clubs are ahead of us in the games played column (3 in the case of Rangers, Hibs, Dundee United, St Johnstone, and Ross County). I will do my best to point this out where it is important to do so, although a lot of the numbers will be 'per 90 minutes', therefore rendering the amount of matches insignificant. All stats are from Wyscout unless another source is credited.

St Mirren drew 1-1 with Kilmarnock at Rugby Park on Saturday, in our 19th league game of the season - we're halfway there, as Bon Jovi once sang (the first and last Bon Jovi reference I'll make). Reaching this stage, to my mind anyway, is a natural time to reflect on how the campaign is going, beyond seeing us sitting in a pleasant 7th place. 

As mentioned, we have 22 points from 19 games. This is a points-per-game average which *should* be enough to keep the Buddies out of relegation trouble, either automatically or via the playoffs. However, if the postponed games against Hamilton and Motherwell get added back on, and Hamilton and Ross County continue their uptick in performance, then we are in trouble again. But have we been fortuitous to garner 22 points or have our displays merited more?

To answer this question, Wyscout quite helpfully has an Expected Points (xPts) tally. This is calculated using their xG data, which hopefully makes sense to readers after noting the xG definition above. Here's the SPFL table based on xPts (thank you very much to Ross for this graphic):

Note: If you're confused about why xPts isn't a round number, like the actual points tally, think of it this way. If the xG scoreline of a match is 0.93 - 1.21, the model doesn't award 3 points to the away team with the higher xG, as the single game variance of football means there is a real likelihood the home team with 0.93xG will grab a victory or a draw.

Interestingly, St Mirren would actually gain in terms of points, although drop down to 8th in the standings. What this table essentially means is we deserve more points than we've earned. The reversal of fortunes between St Johnstone and Dundee United can mostly be explained by Clark and Parish being two of the worst goalkeepers in the league, whereas Siegrist is one of the best. The Old Firm naturally experience a bit of a drop off since they are winning some of their games by fairly close margins. In fact, St Mirren had a higher xG than Rangers in our recent 0-2 loss (although that's another one for the limitations of xG argument, but I digress).

Defence

The most impressive feature of our game has been our defensive capabilities. We have conceded 22 goals, which is fewer than all clubs bar Aberdeen and the Old Firm. It should also be pointed out 7 of those came in the 3 games where Jak Alnwick was absent with coronavirus, with Bobby Zlamal deputising in an emergency loan from Hearts. We have only conceded twice from outside the box - Rangers, unsurprisingly, are the only club to have conceded fewer. This confirms that we're an effective, organised unit capable of harrying players and closing down spaces. In fact, our passes allowed per defensive action (PPDA) is 10.54 - fourth lowest in the league. Essentially, the fewer passes you allow the opposition to make means a lower PPDA number, indicating a proactive and robust defence with an emphasis on pressing.

The low number of goals conceded from lost distance adds weight to my argument Jak Alnwick has been excellent this season (see previous blog). We do outperform our xGA (expected goals against - the xG we concede) as it is currently at 25.72. This suggests we are perhaps too reliant on Alnwick and an average or below average SPFL keeper, such as Zlamal, would result in us collecting fewer points. 

Although we have a decent goals conceded column, we do allow the fifth most shots per 90, at 11.1. While nowhere near as bad as the atrocious Hamilton (16.15!), we should be aiming to emulate lowering to St Johnstone's 8.82 shots conceded average. We are, however, the second best defence at blocking shots, something which will ideally continue.

Another slight concern for me is the fact we've conceded 6 headed goals, which is more than 8 other clubs. This feeds into one of my preconceptions about us that we aren't always rock solid at defending set pieces and dominant aerial players can bully our defence - look at Whitehall's goal for Kilmarnock against us last game for evidence of this. We rank ninth in the division for aerial duels won.

Attack

Fans are enraged by our inability to score goals on a regular basis, with our strikers in particular attracting a lot of ire. On the face of it, scoring 15 in half a league campaign isn't acceptable - way below a goal a game, second fewest with only Ross County below us on 13. Sadly, the advanced metrics back up a large amount of the frustration. We've scored 15 from an xG of 21.92 - the second largest negative discrepancy between these figures in the league (once again, Ross County rank lowest). The upside here is that we aren't get lucky and overperforming, chances are being created, and if we can progress to the mean then we'll hit the back of the net more often and have more success (assuming our goals conceded doesn't regress simultaneously).

Speaking of upsides, we've scored 4 goals from corners - joint highest in the division. Slightly disproves my earlier point of not being brilliant in the air so let's keep that up. We've only been awarded and scored a solitary penalty this season, which is way below average. Not much you can do if blatant handballs are ignored right enough *coughs* Kilmarnock at Rugby Park on Saturday *clears throat* Hibs at home. Perhaps referees being more generous towards us will help too.

Interestingly, we actually rank fifth in shots per 90 minutes at 10.06. The problem arises in that our average xG per shot is 0.109, a figure which is below the league average of 0.12 and far behind the likes of Hibs (0.148), Livingston (0.138), and Rangers (0.135). To put it bluntly, we are not creating enough clear cut opportunities, instead prioritising low probability shots from outside the box. 

We rank fourth for shots outside the box (94) and would probably be third if we had played our additional games - St Johnstone are 11 ahead of us. While this will work occasionally, as the aforementioned stunner from Ethan Erhahon against Aberdeen aptly demonstrates, we need to prioritise more incisive passing in the final third to get more presentable chances to our much maligned forwards. My final comment on this would be that our player with the most shots in the SPFL Premiership, quite unbelievably, is Cammy MacPherson. While 5 of his strikes were direct freekicks, his tally of 0 goals from 26 attempts (with a total xG of 1.08!) is emblematic of what needs to change.

Passing

Appropriately enough considering where we are in the league table, we rank seventh for number of passes completed per game (323, which is miles behind Rangers at 574 but clear of Hamilton at 250 by a fair chunk). We're eighth for passing accuracy and tied fifth for possession overall, showing we certainly aren't a long ball team. The option to play high clearances up to our strikers is there, thanks to Jon Obika and Lee Erwin, and we do average the fourth most long passes per 90. That's certainly a Jim Goodwin team trait, not being afraid to mix it up if need be.

A separate article on our attacking woes sounds plausible but here are a couple of noteworthy facts as a precursor. Obika and Dennis (at 76.53% and 79.71% accuracy) have far better passing than Erwin (68.33%). Admittedly, Dennis has played just 69 passes compared to 240 and 213 for Erwin and Obika respectively, but perhaps a new "late in the game" partnership for Jim Goodwin to try is Obika and Dennis, who have yet to be fielded together for any significant number of minutes. This may help fashion the high quality chances we need if chasing a game.

As an addendum, Dylan Connolly (66.44%) has significantly lower passing accuracy than comparable players Ilkay Durmus and Kyle McAllister (72.99% and 71.36). While again there is a slight difference in sample size, this is something the Irishman needs to remedy with time on the training pitch. His pace is a lethal weapon Saints have thus far underutilised but there needs to be an end product to go with the line breaking runs. Retaining possession in promising areas will also go a long way to solving our problems in front of goal.

Thanks for reading - St Mirren Till I Die!

Lewis.