Is Sizzling Southampton's Start to the Season Sustainable?

I've been monitoring Southampton's underlying numbers this year and I was disappointed that they failed to annihilate Villa as any boost from that game would have sugar-coated their already historically impressive totals.  As it is, it's possible to show that they have had a start to the season that ranks very high in the pantheon of 'Great Starts to the Season' as recorded in the 'Enlightened Stat Era' (2009-now).  This era is defined as such because it covers the public availability of shooting numbers that can be built into metrics that are highly insightful & which can also be toyed with to create weird mini-stats that show that what Southampton have done is indeed extraordinary.

Factor in the pre-season predictions of demise (guilty!), the wholesale personnel changes & the size of the club when compared against other teams that have made outstanding & blistering starts and their achievement gathers even more credence.

But what achievement? Good point.  Starting a season well has little meaning unless it can translate into a tangible return at the end of a season so what chance they retain their position in the upper echelon of the league & usurp a more established visitor to the top 4?

Of course the key point against what they have achieved so far is their schedule which has meant they have avoided all but Liverpool & Tottenham of last season's top 7.  They lost to both.

For this article the seasons referenced above have been used (09-14) and that denotes 120 team starts of 12 games.

Total Shot Ratio (TSR)

  • Southampton rank 9th/120 with 63%
  • 7 of the top 9 went on to finish in the top 4 that season

Shot on Target Ratio (SoTR), SoT+/- and Goal Ratio

  • 2nd/120 SoTR with 72%
  • Of all these extra metrics, this (and the related 'Shots on Target +/-') are most predictive towards final points & position.  After 12 games it is comparative in predictive power to 'Goals For' and only behind 'Wins' & 'Goal Difference'

We can see how this impressive SoTR has translated into an equally impressive goal % & that historically, it stands very highly:



Similarly the shots on target +/- figure is historically high:



Of the teams on this chart, only Liverpool 09/10 did not finish top 4.  We can also see here that despite varying media perceptions of Man City & Chelsea this year, they rate identically on this metric.

Shots, goals against & saves

Whilst Southampton's goals total is impressive, to some degree it has been enhanced the the 8-0 romp against Sunderland.  There is some mileage here though; teams averaging 2.0 goals per game or better at this point have qualified for the top 4 on  the last 15/18 occasions.  Their shots for totals are no better than average; indeed they rate only 8th in the league here.  Where they really impress is at the other end of the pitch.  Defensively they are putting up extraordinary numbers, something that can be highlighted by this little nugget; they have conceded 4 shots on target in their last 5 games.

  • Shots against: 1st/120 (Arsenal 2014/15 also score extremely well here)
  • Shots on target against: 1st/120

And these figures are powering this:

  • Goals against: 2nd/120
  • Saves (low): 2nd/120

Save %

Not only are they effective at minimising opposition efforts, they've also been good at repelling those that do get through.  Their save percentage is not historically significant, but nonetheless leads the league this year.

Shooting Accuracy

Or, the percentage of shots on target in relation to shots.

  • For: 9th/120 (Swansea & Chelsea are also historically high)
  • Against: 6th/120 (Stoke are 1st/120 here)

Both solid numbers, but the key component here is the relationship between the two and is arguably Southampton's most impressive/least sustainable stat.  This cake can be cut a few ways, but I've just represented the difference between the raw percentages:



To score 13% where the previous high is about 10% is a noticable outlier & whilst the stat isn't particularly predictive, it can reasonably be assumed that the ability of Southampton to restrict their opposition's shooting accuracy in relation to their own will reduce as the season progresses.  Indeed, a seasonal high for this figure is only 8%.  Somewhere along the line 5 points will most likely disappear.

As we can also see, Swansea, Chelsea & Everton are all scoring above seasonal expected levels here.

Conversion rates

Simple shots to goals percentages:

  • For: 8th/120 (Swansea, Palace (!) & Chelsea also riding high)
  • Against: 2nd/120
  • +/-: 2nd/120

There are 8 percentage points difference between the for and against totals (14% & 6%).  In the course of the sample, the only teams to break 6 percentage points over an entire season are three Alex Ferguson teams & Man Utd & Man City occupy 8/9 slots above 5 percentage points.  The other team is Brendan Rodgers' Suarez fuelled 2013-14 squad.  Overall, these numbers appear to lack league-wide predictive quality but seem to represent the very good and the very bad at their extremities. Could this Southampton team join this list of good & great teams?

The near future tricky, but there is some hope that Southampton can retain their high position in the short to medium term:



They have been particularly good at home, winning 5/6 and conceding only one goal & five of their 'tough' matches are at home. Throw in winnable matches at Burnley and Palace & there's can be a genuine expectation that whilst their impressive numbers will almost certainly regress somewhat, there is clear scope to continue to pick up a good points return.   There are some precedents that suggest that what they have already done stands them in excellent stead towards the top 4:

  • The last 25 teams with +10 goal difference after 12 games finished the season in the Top 4. Portsmouth failed in '07-08. Southampton are currently +18.
  • The last time a team in Top 3 after 12 games finished out of the top 4 was 2007/8.
  • The top 16/17 Goal Ratio teams after 12 games finished 4th or above.  Southampton rank 2nd here & the only team not to finish top 4 was... Southampton 2013-14. 
  • 8 wins after 12 games? 9/9 top 4.

Problems & Top 4?

Southampton lack a deep squad and have been relatively fortunate with injuries thus far.  A series of quick games against tough opposition may well test their player resources and with key contributors throughout the first team, any succession of injuries will noticeably weaken them.

A significant proportion of their current success has centred around their defensive abilities & the Villa match was the first in which they committed a defensive error that led to a goal; and a loss of two points.  We can praise Koeman for instilling an effective defensive strategy & mindset thus far but having spent very little time in a losing position, their tenacity when under pressure has rarely been tested.  This will surely change as we go through December.

They have twice scored as a result of instantly returning terrible goalkeeper clearances.  This kind of goal is scarce & that they've benefited twice is a subtle indicator of how their season has progressed with a degree of charm.  Similarly, the generosity shown by Sunderland's defence has provided a notable massage to their attacking figures, some of which are not historically impressive.  Removed from analysis, their goals for is a much less impressive 1.3 per game and you lose 4 points from conversion.  Suddenly: not so impressive.

Despite these issues, I personally feel that it's worth taking a positive view of what has been achieved.  The betting markets give them around a 2/1 chance of qualifying for the top 4 and that, as I think I've shown, is not ungenerous.  Indeed anyone who has availed themselves of higher odds can be well pleased with their current position.

Moyes vs. Martinez - Which Everton Manager Has Been Better?

Yesterday, Gabriele Marcotti wrote a piece on ESPN FC about how difficult it is to evaluate managers. How a few lucky bounces can change the course of a career. How a wrong turn can leave can leave you marginalized, struggling to rebuild your reputation. He used the example of David Moyes, who until 18 months ago was "probably the highest-regarded British manager in the Premier League". Firstly, LOL. Secondly, while being "probably the highest-regarded British manager in the Premier League" is not actively an insult, it falls somewhere close to "probably has bigger lips than Garry Monk" on the spectrum of damning by faint praise. Nonetheless, it got me thinking that maybe it was unfair to judge Moyes against vintage Alex Ferguson, whose 2012/13 masterpiece (lucky or not) is starting to look like one of the best managerial performances in Premier League history. Maybe that's too high a standard. Pair that with the fact that he inherited 'Dad's Army', and you start to see Marcotti's point. Besides, it's not like Moyes' own successor has been lighting worlds on fire (despite a good deal more investment). So I'm going to compare him to Roberto Martinez, and I think it's a pretty reasonable comparison. Sure Romelu Lukaku is nice to have if you're Martinez, but don't forget how dominant Marouane Felliani looked in his last season at Everton. Other than that, the 2012/13 and 2013/14 Everton squads were, on a talent level, pretty similar. What I've done is make 'shot chart' graphics for Everton's last year under Moyes and its first year under Martinez. I employed the same technique that Kirk Goldsberry uses for his basketball shot charts, except instead of just showing the shots of an individual player over a certain period, I'm showing the shots taken/shots conceded by the entire team. Just a few of things in case you don't know the drill. 1) More hexagons = More shots taken 2) Bigger hexagons = More shots taken 3) Color depends on efficiency, which I based on shots/shots on target. Blue is bad, red is good. So first a look at the defence: emoyesdef Click to make them bigger... rmartdef Both Evertons were pretty good at limiting the opposing team to average or below average shooting from most areas of the pitch. Martinez's Everton benefited from opponents converting poorly from within the six yard box, perhaps because of Tim Howard's late-career emergence as the Great Wall of China, or perhaps due to plain old luck (it's probably a bit of both). But basically, things look kind of similar, which makes sense, because things haven't changed that much at the back. Leighton Baines, Sylvain Distin, Phil Jagielka and Seamus Coleman started pretty consistently for Everton both in 2012/13 and 2013/14. Only this season (2014/15) has John Stones destabilized the situation by soaking up a lot of Distin's minutes at centre-back (and perhaps it shows in Everton's increased profligacy). If you take a look at the numbers on the left-hand side of Martinez's chart (click on it to make it bigger), you can see just how similar these defenses are even in just in terms of how many shots they gave up all together. But this is where it gets interesting... In his article, Marcotti quoted some of Giovanni Trapattoni's most famous words, "A good coach who gets everything right can make a team maybe five percent better, a bad one can make it 30 percent worse. Sometimes more." I wouldn't consider Moyes at Everton a bad manager, in fact I'd say he was quite a bit better than your average. Sure he had his issues game-planning for big matches, and yes I always felt like he rode his players a little hard - leading to a seemingly perpetual injury crisis at Goodison Park. But generally, he sent out well-organized and motivated teams, which is what makes the charts below so interesting. Take a look... emoyesatt   rmartatt That's a pretty big difference! Martinez's Everton TOASTS Moyes' Everton in attacking efficiency. Even if you forget about the red-hot shooting from inside the six-yard box, it's pretty easy to see that the 2013/14 Everton team shoots more efficiently than their 2012/13 counterparts from almost everywhere. They get good shots from outside the box, and while they're still just about average from the heart of the box (that big yellow part), it's  a significant improvement on where they were a year back, wading around in all that BLURRGH (blue). Again that sort of makes sense when you think about how both managers set up their teams. Moyes' attack revolves around a lot of crossing from wide/deep positions. That means a lot headers and tricky volleys, which are difficult to put on target. It's a tactic he brought with him to Man United where his team famously crossed  the ball 81 times against Fulham's 6ft 7inch Brede Hangeland. Martinez's team rely more on throughball tactics, which lead higher-quality chances inside the box, and more shooting space outside it. Moyes isn't a bad manager, but maybe this is what "5 percent better" looks like.

Man City v Man United Player Positional Tracker

Man City 1 vs 0 Man United Here is our visualisation that shows the smoothed positions of players around the time as indicated. The locations are identified with reference to actions. Comments from Sam Gregory appear below the PPT.  Click on the gif to open in a larger window:


First Half

  • Rooney’s return to the United starting XI saw him take up an influencial central midfield role playing alongside Fellaini and Blind in the middle of a 5 man midfield.
  • Both teams started playing with one man up top but Aguero was much more effective than Van Persie. Van Persie was marked out of the match by Kompany and received very little service. Aguero on the other hand found plenty of space between Rojo and Smalling and was very mobile moving from side to side. His involvement is noticeably much larger than Van Persie.
  • After Smalling’s sending off Jovetic moved up the pitch and effectively joined Aguero as a second forward.

Second Half

  • For the first twenty-five minutes of the second half City absolutely dominated the ball. The ten United players left on the pitch had some of the smallest influence dots during this period that I’ve ever seen on a PPT. It wasn’t until the goal that United was able to put anything together in the second half. Yaya Toure and Fernando were particularly dominant to start the second half
  • During the last twenty minutes Di Maria, Rooney and Fellaini played much further up the pitch, with Di Maria playing in a more central position. This gave United something going forward and gave City a bit of a scare during the closing moments.
  • Added by Colin: Kompany and Demichelis swapped sides in Man City's defence for the final 20 minutes of the game.  Presumably this was an attempt by Pellegrini to contain Di Maria as much as possible. EDIT - @evolutionHPcal has suggested it may also have been to help Clichy with Fellaini's aerial threat.


  • City were dominant and fully deserved the three points in a game they should have scored more than one goal. Chris Smalling’s sending off was clearly the turning point, but City deserve credit for capitalizing on the advantage.