Hindsight, they say, is the predictive analyst’s best friend. To make a prediction, we collect data (the past), look for trends, and attempt to apply them to the future. When we are wrong (and when we are right) we go back to our model and revise it based on our new data.
The stat-du-jour is xG: Expected Goals. Expected goals looks at all shots in the past, determines what percent of them have been scored from what positions and assigns a numeric value to that shot. In simplest terms, shots inside the 6 yard box, for example, are scored at about a 40% rate. Shots outside the 18 yard box about 3%.
From there you can get more finely detailed with the analysis and make the stats a little more accurate: was the pass before the goal a through ball (that raises the chance of a goal), was it a headed shot from a corner (that lowers the chance of a goal), did the player win a dribble right before the shot, and so on. But if you don’t have access to the Opta data that informs such an analysis you can go old school and simply look at how close the shot is to goal. The closer, the better.
Arsene Wenger has been known throughout his career at Arsenal as the manager who “tries to walk the ball in”. Wenger does that because Wenger knows that shots outside the box are low xG (this is not a keen insight, everyone knows this) and he would rather his team retain possession and work to get a good shot than have the player (ahem, Elneny) shoot from distance. In essence Arsene Wenger is OG xG.
Back in January I said that Arsenal had the best underlying stats of any team in the League and that there was no way Leicester could win, unless they tightened up their defense. That opinion was based on how many shots Arsenal were taking from inside the 18 yard box, how many Big Chances Arsenal were creating (a big chance is a shot situation where you expect the player to score, like a 1v1 with the keeper), and how few of those chances Arsenal were conceding.
That opinion was also based on Leicester’s stats. They were creating shots just fine, but they were conceding far too many good shots to win the League. I also posted a controversial opinion that Tottenham’s underlying stats were scary and that because of their defense, they could be fighting it out with Arsenal for the title.
What happened at Leicester was extraordinary (I don’t include own goals in any of my tallies, this may confuse some people but just try to play along).
If we divide the season into two halves we see that Leicester went from being a team that conceded 1.3 goals per game to a team that conceded 0.58 goals per game. That’s almost 0.75 goals per game fewer.
They achieved this miracle by limiting the opposition’s shots in prime areas (the area from the penalty spot in to the goal) and making them take shots from slightly further out. Their shots in prime allowed dropped from an unsustainable 94 in the first half of the season to 61 in the second. That brought them on par with teams like Arsenal in terms of controlling the opposition’s shots.
But they also got a HUGE boost from Kasper Schmeichel. He went from saving 51% of the shots on target in prime (20/39) in the first half of the season to saving 73% of the shots on target in prime (16/22), As a result, Leicester allowed 19 goals from shots in prime in the first half of the season to and just 6 goals from shots in prime in the second half. If Schmeichel had saved at the same rate in the second half as he did in the first half, Leicester would have allowed 5 more goals. That would have been the difference in any one of their six 1-0 wins that they chalked up in the second half of the season.
So, what happened to Arsenal?
As you know, Arsenal created more shots in prime and more big chances than any team in the League, 97 big chances this season and 207 shots in prime. As a point of comparison, Leicester were second in the League in big chances with 93, and only created 158 shots in prime.
Over the course of the season Arsenal’s problem wasn’t creating the chances, Arsenal’s problem was finishing. But critically, Arsenal’s shot creation dried up in the second half and to make it worse instead of getting better at finishing (people kept waiting for Arsenal to “revert to the mean”) Arsenal went from bad to worse in the second half of the season.
Arsenal created just as many (3 more) shots in prime in the second half of the season as they did in the first. This makes perfect sense because that is Wenger’s football philosophy at work. Remember he is OG xG. But what dropped through the floor was the number of big chances they created, down from 58 to 39. That’s a 19 shot drop over 19 games. That’s one big chance per game fewer than in the first half of the season. And if big chances are converted at 40%, which is about the correct rate, that’s almost 8 fewer goals.
Arsenal’s finishing from shots in prime dropped from 25% in the first half to 17% in the second half. This could be attributed to the drop in number of big chances. Those two stats have a lot of overlap.
Oddly though, Arsenal scored 31 goals in the first half and 32 in the second half. So, while their underlying numbers were down, the actual goals scored stayed the same. Arsenal were getting fewer good shots but they were doing OK to finish the shots they got.
And defensively, there wasn’t a huge drop off. They did significantly better at preventing the opposition from getting big chances, down to 14 from 21, but then were worse at saving them and conceded 7 big chances in the second half compared to 6 in the first half. And overall, Arsenal only allowed 3 more goals in the second half than they did in the first half.
Arsenal basically never “reverted to the mean”. Their percentages stayed mostly average, they dropped of in terms of creating big chances, and their defense didn’t get any better.
There was one problem area that needs to be addressed in the off season, goals conceded from distance. Arsenal and Leicester both allowed 42 shots on target from outside the box and Tottenham allowed 41. And while Leicester and Tottenham allowed just 6 goals (combined) from the 83 shots on target they faced, Arsenal allowed 11. Arsenal had the worst saves rate for long distance shots among the top teams and Cech had the worst saves rate among all keepers. Instead of getting better as the season went on, Arsenal allowed 5 goals from outside the box in the first half of the season and 6 in the second.
Cech could still win the golden gloves this season and he has been good saving shots from close range. But Arsenal conceded 9 more goals from outside the 18 yard box than Tottenham and 7 more than Leicester. Those are low percentage shots and Wenger has to be concerned that Arsenal conceded so many.
Arsenal did have a lower blocks rate (35%) than Leicester (38%) and Tottenham (43%) from shots outside the box but there is also a sense from watching the matches that Arsenal aren’t putting enough pressure on players with the ball on the edge of the box. That doesn’t show up in any stats that I collect.
Of the 18 goals Arsenal conceded in the second half of the season, 13 of them were scored in just 5 games (Liverpool 3-3, Man U 3-2, Tottenham 2-2, West Ham 3-3, and Man City 2-2). This is reminiscent of Arsenal’s struggles with big teams a few years back. Worse, in three of those contests Arsenal actually led the opposition before conceding the tying goal. That is six dropped points right there. In the other 14 games, Arsenal conceded 5 goals. And if we put the Swansea debacle in there, Arsenal conceded 3 goals in 13 contests and 15 goals in 6. That’s quite a split personality.
And finally, what happened to Tottenham?
They Spursed it!
I looked at their underlying stats and basically what happened is they did the opposite of a Leicester. They went from a team that conceded 13 goals in the first half of the season to a team that conceded 19 in the second half (remember I’m not counting own goals). And of those 19 goals. 10 came in the last 4 games. For the first 34 games, they conceded 22 goals, in the last four games, 10. HA HA HA HA HA HA HA HA.
Oh man. Despite Leicester’s incredible run and fantastic defensive display to close the season out, you know the world is normal when Arsenal finish top four, Spurs finish below us, and Everton are somewhere near 10th.
@7amkickoff
All stats today are from my personal database. All data painstakingly culled from public sources such as the 442 Stats Zone app, whoscored.com, and Squawka. Other sources include statto.com and transfermarkt.com.
Great analysis as usual. Leicester were sensational in the second half.
As always our problem was putting a full season together. We always seem to be strong in one half and not the other.
spot on. wenger’s squads have never been able to cope with competing on 4 fronts.
too many in sick bay and squad players not able to step up to give the first XI a break when needed.
They were also very lucky on their penalties conceded count. Huth generally commits a penalty worthy offence every game he plays (often two) but refs seem to ignore him in a Leicester shirt… Very odd.
Thanks for the article. I think in hindsight Leicester had lots of 1 0 wins in 2nd half of the season and their defence looked better and better. For us I think we just had very unbalanced midfield due to injuries and lack of depth. Our attacking and defending was just never the same since November. Coq and Santi missing large parts of season just got to us. Ramsey and Flamini was just not good enough and Elneny arrived 6 months too late I think. Hopefully we can address the midfield by signing Xhaka and address finishing by signing a… Read more »
The facts about Ramsey and Flamini don’t support you. The simplest is that it was when they played together in midfield (for 2 months) we got to the top of the table in January. We didn’t do that with Coqzorla or any other combination. A distinction really needs to be drawn between not looking pretty and not being effective. Not pretty but Flamini and Ramsey were certainly effective. Whether people like to admit it or not.
Would love to get your thoughts on this 10am…. With most play now geared towards scoring from the best positions (closer to the 18) and everyone seemingly moving away from long shots, and teams sitting deep into their 18 and leaving space outside the 18, we are seeing the move towards counter-attacking working well (leicester, Atletico last 3 years, Dortmund, etc.)… What would happen if a team decided to build build upon going against this trend? So they built not just a counter-attacking team happy to concede possession and get it back later, but they build a team to take… Read more »
Before I get thumbed down to Bolivia, there was an article just a few weeks ago talking about the rise of the Warriors. Bought be a Silicon Valley investor with a start up tech mentality, the assessed the state of play of basketball and noted that few teams took 3-point shots as they were low percentage and also gave away possession, so virtually all teams played toward going for 2-points (and team adjusted defensively accordingly, by sitting deeper to protect the basket and await the rebound). In short, the tore up the rule book, and decided to make a team… Read more »
The math on the basketball stuff was known by stats people well before it was implemented by the Warriors (and the Rockets and the 76ers but with vastly worse talent) and it is pretty obvious in hindsight. Say you have a player who shoots 70% close to the rim 50% from mid range and 40% from 3. The points per shot type is 1.4, 1 and 1.2. From this example the best places to shoot are at the rim or from three. If long shots were worth more it would not make sense to take many of those shots. This… Read more »
Part of why I don’t think that really translates all that well is because long shots in basketball are worth an extra point where as in soccer long shots are worth the same. The big innovation in basketball is that midrange shots are not goo shots because if you step back a few feet you might make a few less but the extra point makes it still worth more points in the end. To score a goal from outside the box you need to shoot about 33 shots on average. To score a goal from the center of the the… Read more »
Ha! Threes in basketball are far more numerous opportunities and far more likely to score. Let’s look at the top scorers from outside the box. Messi is 6/65 Insigne is 6/77 Pjanic is 6/42 Dybala is 6/52 Zlatan is 6/51 Griezmann is a badass 5/33 That is 35 goals on 320 shots. 11% conversion. That’s ridonculous. Now, let’s assume that almost all of those players were scoring from shots on the move (a lot of them were set pieces) and let’s assume that Arsenal took 600 shots (they took 578 this season) and that those players scored at the same… Read more »
So….
You’re saying it work? 😀
Way to ruin my thesis.
Not quite what you’re envisioning but Liverpool did try to build a team purely around crossing, they brought in the best talent at crossing the ball and the best striker they could find to get on the end of crosses – Andy Carroll. Damien Comolli admitted as much.
We know how that turned out…
You have to remember that the long ball nonsense all came out of a statistical analysis that showed that the fewer the number of passes before a shot, the higher the probability of scoring. That completely ignored various other factors.
If defense wins championships, what are the chances that a classic Wenger attack-first team could ever win a league title? Has this been our problem all along? I don’t know. Personally, I think we were undone by injuries this term. Looking at our squad, I’d have thought we’d have more than enough to beat Leicester to the title this season (and certainly finish more comfortably ahead of Tottenham), but the longterm injuries to Cazorla and Coquelin (breaking up a truly outstanding partnership), Wilshere, and Sanchez… We didn’t have the depth to cover for that. Or, rather, we might have, but… Read more »
Let’s be honest, even at the early part of the season, United at home aside, we haven’t really played or performed well for any long period to say “these guys look unbeatable”.
There was a short period where we looked “not as bad as the rest”, but certainly not mind blowing.
I think that’s fair.
Counterattack. I think this team, with Ozil in it, should aim at letting the opposition have the ball more, and counterattacking with speed.
Probably need Alexis and Walcott on the wings, and a speedier central striker.
@REvers, I’m really surprised at how slowly we attack even though we have some of the fastest players in the premier league. Wonder if we can get some stats for that
The old Ramsey had a different job, and in a team which played closer to a counter attacking team than any Arsenal team since we moved to the Emirates. He had a lot more license to roam back then, basically. And it worked because in the first half of that season we defended as a team.
I think that the drop in big chance creation was largely caused by Cazorla’s injury. We were clearly much more dangerous when he was in the team. The problem is that it is really difficult to find backup for someone like Cazorla in the transfer market. Players with skill sets similar to his come with high price tags and expectations of regular first-team football. Furthermore, Santi has not been an injury prone player in the past, so it seemed reasonable to expect that he would play most of the season. Combining these two factors, it is understandable that his position… Read more »
I’m sure Wilshere could play Cazorlas role in future with similar success.
I’m sure he could but this season we needed him in February and March and he wasn’t available. We will need another player with similar attributes, because Wilshere’s health is not reliable enough.
I think another big reason for the drop was our style of play changed from a team that defended and attacked together to a much more segmented and rigid team, presumably to make the Ramsey-Flamini partnership work. I’d actually like to see stats for Coqulein before and after injury, I think he’s been naff since he came back (Spurs away was a nadir) but maybe I’ve got it wrong
The thing is, in terms of results, Ramsey and Flamini did work. Also for about a month in mid-January (after Liverpool) we barely let in a goal, which is really the first (but not only) function of the central (defensive) midfield pairing. Our stats after Coq came back are particularly unimpressive, although in fairness he hasn’t played too much. But what everyone chooses to ignore is that while solid even when Coqzorla were playing we were able to step up a gear after they were replaced. Again even if only in terms of results. More than 2 months is a… Read more »
Flamini has never lost against the Baby Totts for us.
Defense and dodgy penalties.
I didn’t think our defence performed bad at all, our attackers didn’t take their chances and this invited pressure. Players like Ramsey, as much as I hate to say it, neither contributed up front nor defensively.
Apart from that’s not true on Ramsey. I’d actually say that Ramsey has been as much of a DM as either Flamini or Elneny when played in the middle. Essentially, he did a job in the middle whereas he’d normally have more license to roam like Cazorla. That may have been tactical from Wenger but it hurt Ramsey’s game
Also disagree with that we defended well. We shipped 3 to United’s U21 side, conspired to concede to 2 to Swansea at home when we needed to win, had Sp*rs on the ropes at the lane and still conceded 2 (red card for our DM), shipped 3 at Anfield, and somehow lost to the worst Chelsea team in well over a decade, TWICE (featuring to red cards for our CBs).
If it was just our attack being the issue, we would have won or drawn those games. And we would have almost certainly won the league.
The PL stats for this season are actually rather weird.
Spurs scored more goals and conceded fewer than both Leicester and ourselves and yet finished third. Man City were top scorers but finished fourth.
Normally, the team with the best goal difference wins the title, and that was Spurs. But not this year.
Like I said, strange.
Hahahahahahahahahaha
Tim: Excellent analysis! Here are the crucial points dropped for the Gunners this season: -3 at West Brom: Up 1-0, concede a lucky goal, then concede an own goal, then Santi slips on a penalty. -2 at Norwich: Up 1-0, get injuries to Alexis and Santi. No red card for a push on Alexis. Concede odd goal. -2 at Liverpool: Down 2-0, then up 3-2. Conceding to Allen as midfield/defense could not close him down. -3 at home against Swansea. Up 1-0. They score after fouling Ozil (no call) then on a BS goal by Williams. -2 at home against… Read more »
Mike Dean at Chelsea. 3 more.
Merts on Costa at the Emirates. 3 more.
Cech v West Ham opening day. 2 more.
if you cannot score shots outside 18 common sense States you cannot defend them since you don’t practice. we cant score regularly through corners and set pieces. then it’s logical you can’t defend the same.
These statistics don’t take into account the way teams play against us and the restricted space we are required to work in. Surely the conversion rate, as seen by our statistics, is reduced.
Another point is if we had our players mix it up a bit more, maybe it would stop teams from sitting back so much against us and help to create space. Also, an accurate shooter would surely create more corners or rebounds.
Overall, in my opinion, we are being a bit one dimensional.
Wenger has been playing this way for 20 years. Teams have known about Arsene’s strategy for at least 10 of those years. People are just now, this season, figuring that out? I don’t think so. What we have seen over the last 10 years is a gradual erosion of quality at Arsenal. Buying Ozil and Alexis built that back up a bit but when you sell off all the best players every season for a decade it’s hard to reboot right away. I also think fatigue played a huge part. Certainly Alexis and possibly Giroud. The frenchman looked awful for… Read more »
I’d disagree that we’ve played the same way for 20 years, there’s been a shift in emphasis around the time we moved to the Emirates. As for erosion of quality, that may be the case (though I disagree about it being gradual, there was a big step back after losing Fabregas and Nasri, RVP only managed one fully fit season etc) but couldn’t you also argue that we haven’t been playing the correct system to suit the players we have? We have some of the fastest players in the premier league so I’m shocked by how slowly we attack, especially… Read more »
Based on that argument, Denilson, Bentner, Adebayor, Eboue, Walcott, Song, Flamini, Senderos, Hleb, and so on were better players than we have now? Because a bunch of teams from the “barren” years with those players put up better displays than this season.
I feel like this is where Cazorla was missed most. We have some great offensive talents–Ozil, Sanchez, etc.–but when dealing with the restricted space, as you put it, Santi is the one player who is most capable of breaking down the opponent. That sense of creativity, of being able to see a couple passes into the future (at least it seems, sometimes) is what often allows penetration of teams that try to suffocate our attack.
I think we could simplify all this by admitting that we absolutely should have bought a striker and midfielder last summer, even at slightly ‘over the odds’ cost. Wenger completely failed by not buying a single outfield player in the summer. The only manager in the entire premiership to not do so. The 5th most valuable club in world football. That is simply staggering. Regardless of how you feel about the man, this show of arrogance, even for him, was breathtaking. What is almost as shocking, at the time, is that many of us didn’t feel too fussed about it,… Read more »
Isn’t that the point of alot of this article? We under performed when the stats would have said we should come through. If we had all key players fit, i.e Cazorla, Coq, and more consistency – we might have delivered.
Agreed a striker is needed, but then its managing team spirit and cohesion too – look at City. Two key players out at times, one in particular – and they still fluffed it
This thing about Arsene not signing somebody last summer is a bit misleading. Man City, Chelsea, Man United, Liverpool, Newcastle were also looking for and signed lots of players. I think about half a billion’s worth just from those clubs. I did see the figures showing our “rivals” had spent that much cannot remember who those rivals were. But who were the successes at those clubs: De Bruyne (at £55m). At a push Martial (at £57m). And ….? West Ham and Leicester made a couple of good signings. Possibly Stoke of all teams as well. The point is the players… Read more »
I wouldn’t say either De Bruyne or Martial were successes although I do think they’ll both ‘kick on’ next season.
I get the general theme – they wouldn’t have won the title if they conceded a few more goals – but can’t see that Leicester are a great example of defence winning the title – they conceded the same amount as us, and one more than United and Spurs. Emphasis on defence versus attack seems to come in waves, perhaps paradigm shifts. This graph shows goals conceded by the league winner and total for top 4 by season. There’s clearly structure whereby defence was king during the 2004-2009 period. You can perhaps point to Mourinho’s influence there, though arguably the… Read more »
7amkickoff has already analyzed this somewhat here (http://news.arseblog.com/2016/05/cechs-saves-from-distance-v-inside-by-the-numbers/), but Cech’s goals allowed from outside the box worry me. Part of it has to do with the number of shots from distance we give up, and their location, which is on the defense. But I also wonder if it’s a sign of Cech’s fading athleticism. His save ratio inside the box suggests that his reflexes and positional awareness are top notch. However, I don’t think I was the only person baffled to see some of those long shots sneak in at the near post. I’m not saying we should chuck Cech… Read more »
Aah statistics. With Xhaka in our team next season, he could be worth up to 70 points – albeit on the Scrabble board and at least 19 if the opposition close down the squares.
Great season of analysis “by the numbers”, I do love it all.
No matter the stats, all I know is if he does not incorporate at least FOUR tall players into the team to help at defending and converting at set pieces, the same shit that has gone on in the league for 12 years will still happen. People seem to forget that when we were winning titles we had Vierra/Sol/Adams/Gilberto even Sanderos and Kanu were all 6foot+ today if Giroud does not play we only have BFG, that is a strategic mistake. If you study SAF team, he has Bruce/Pallister and later Vidic/Rio both 6’3 plus. I do not know why… Read more »
Yeah, but you can’t put a number on poor morale, poor leadership, lack of killer instinct, muddled tactics, etc. That’s where things go wrong for Arsenal
amongst your most interesting analysis. great article