How Emery hasn’t changed the Arsenal pt. 2 (Big Chances)

Yesterday I wrote about how Arsenal’s attack is starting to shape up and gave good reasons why the Gunners should be hopeful that their two forwards, Lacazette and Aubameyang, will score a combined 40 goals this season. And now I turn to quite another matter, Arsenal’s defense. No matter how I look at the data I have to admit that the first 11 matches of this season give cause for great concern for Arsenal defensively.

Arsene Wenger’s final season at Arsenal was, by all accounts, a disaster. The players went so far as to leak to the press that they were upset at Wenger for not coaching defense. Along the way, the Gunners conceded a Wenger-high 51 goals, 46 non-penalty goals.

Looking at the underlying statistics for last season isn’t pretty either. Expected goals against last season was 49 and 44 non-penalty goals. But what concerned me the most was that Arsenal conceded a 7th worst in the League 69 Big Chances.

Big Chances are those one-v-one, open field, super close, or even open goal chances that fans expect a player to score. There were 986 goals scored last season and 571 were big chances. 58% of the goals scored last season were big chances. Big chances conceded and created is my number one metric for deciding how well a team is doing in a season.

Last year, Man City conceded just 32 big chance shots, 12 big chance goals and Ederson only saved 6 (TOTAL!) big chance shots. The main reason why I kept saying that Man U were a fake was because unlike all of the other top four teams, they conceded a ton of big chances (54) and relied on David de Gea to save them (18) or for their opponents to miss (16) to keep their defense tidy.

Arsenal this season are playing like Man U last season. They have already conceded 22 big chances in 11 matches (2 per match) and Cech and Leno have already made 11 big chance saves (1.0 per match). Just to illustrate how much Arsenal are overperforming at the moment: they have only conceded 3 big chance goals this season on 22 big chances allowed. Big chances are scored at about a 45% rate. Meaning that Arsenal’s expected goals allowed off of this type of shot is 9.9. And they have conceded just 3.

Arsenal’s big chance save rate right now is 11/14 shots on target, 79%. Keepers average 36% saves off big chances on target.

The other reason I like Big Chances so much is because I think they show how well your team is attacking and defending. It’s true that teams who play outstanding team defense don’t allow a ton of big chances. Last season, of course, Man City were the best defense in the League. But it was the creaky defenses of Liverpool (51 big chances allowed), Man U (54 big chances allowed) and Arsenal (69 big chances allowed) that needed improvement.

Liverpool have gotten that improvement, United have not, and Arsenal have actually gotten worse.

In terms of Big Chances allowed, Arsenal defense is now conceding 2 per game – incredibly, we are saving 1 per game but this is completely unsustainable. No team, as far as I know, has ever had 38 big chance saves in a season. And teams that allow 76 big chances get relegated.

Even if we compare like for like last season to this season, Arsenal “only” conceded 17 Big Chances last season in these same 11 matches. And that included 7 big chances conceded to Liverpool and Man City, who we only conceded 4 big chances to this season.

The problem for Arsenal is that they are conceding these big chances not just to the big clubs but ALL of the teams they have faced this season. It looks much more like a systemic problem if you have conceded 2 big chances in 7 of your 11 matches, which is what Arsenal are doing.

If I were looking at this team as an outsider I would probably say “Arsenal haven’t at all resolved the midfield and back four defensive problems of the Wenger era. That Arsenal’s midfield is still not up to the standards required to make a top four challenge. That Torreira is not fundamentally solidifying the Arsenal midfield and that neither of Unai’s double-pivot “DMs” are effectively shielding the back four. That the hype over Torreira is only because Arsenal are winning. And that Unai Emery is relying on an unbelievable saves rate from his keepers and a very high finishing percentage from his forwards, to keep Arsenal in the top four trophy hunt.”

But I’m not an outsider. And as an insider what I will say is that I truly hope that we start to see some better defensive performances from Arsenal. Liverpool was an improvement over the previous season and felt like a watershed moment (even if the defense was quite lucky on at least three major moments). So, it feels like there is no where to go from here but up! There are 27 more games to go this season and I think it’s fair to give Emery several seasons to turn around this Arsenal defense and clear out some of the players who simply can’t get the job done. Klopp made Liverpool better defensively with two (major) signings. We have to allow Arsenal that same latitude.

Until then, the stats show us that it’s going to be a bumpy ride.




  2. Wow, serious buzzkill, Tim. I am tralala-ing my way through this week thinking we are so much better and now you say we’re not? I’m torn between believing the stats, that say we’re not better defensively, and my eyeball test, which says we are. Is there a “next to last ball” stat that indicates the “almost chance?” LIke when Mkhi got the ball stuck at his feet when he had Auba set up for a sitter on Saturday. Sometimes a team slices open a defense repeatedly, but makes a bad pass, or a poor touch, and the chances just fizzle. I ask because it seems that while our XG hasn’t improved, we don’t seem to be giving up as many “almost chances.” It’s much harder for opponents to get into dangerous areas that allow chances. It’s been several weeks since we’ve given up a really dangerous ball over the top, it seems, even if the striker/wing can’t settle it. Maybe that’s why it FEELS like we’re a bit better. Not great, and maybe not even very good, but better. It’s all my intuition and probably biased by our winning record so far, but something feels different about how often we give up key passes. Throw me a bone here. Can’t you please let me happily tralala a bit longer?

    1. My eyeball test (and my blog posts complaining about Xhaka, Torreira, and Mustafi) has been saying that we aren’t better. Not even remotely.

    2. My observations are that fans think players and teams are better than they actually are based on results. It’s a bit of a trick of the human brain: the happy checks are remembered easily, the sad checks are often filed away. This is why last season people discounted my work on Man U’s stats, or when I said that Burnley would get relegated this year they said I had salad for brains. To a much larger extent this is also why players like Overmars, Petit (and that famous Graham back four) are remembered more fondly than they perhaps deserved. Results are king for fans. Fans are super duper stoked on Torreira because we are winning. Xhaka got praised so much in the Guardian the other day that sun literally shone out of his ass. If we had the same record as last season, fans would be braying for Guendouzi to start, to drop Xhaka, and suggesting that Torreira was an abject failure. But so far, the data tells me that he’s not even at the level that Coquelin was in 2015/16 in terms of shielding the back four. In fact, we are allowing MORE chances than ever before and we are playing with two DMS. I think there’s a strong case that he’s objectively made us weaker.

      But if you want some positivity here it is: this has to be the absolute bottom for Arsenal. Something has to break. Either Emery gets it sorted out and we stop allowing teams to shoot all over us (they concede more shots than they take) and shores up the defense and midfield to stop teams getting so many Big Chances or we start to concede a lot more goals and lose matches.

      You aren’t going to see Cech/Leno get 38 Big Chance saves this year and also see Arsenal finish top 4.

      1. Yes, if you’ve read Mlodinow’s great book, “Subliminal,” he writes extensively about how malleable memory is. We can be convinced to change our memories based on what others tell us. And even more damning, he writes something to the effect that we all believe we are scientists, who look at data and draw conclusions, while in reality, we are all lawyers, who draw conclusions and find facts to fit them. As always, the eyeball test wilts in the face of data. That’s why most of us come here. There is plenty of bandwagon punditry out there – we like a more scientific approach.
        The shots and big chances conceded should eventually catch up with us. But maybe the benefit of winning is the team believes it’s better, and believes in Emery, which makes them more willing to work hard within his system, and make the adjustments to improve.
        We’ll see how far the system can take this group of players.

        1. Speaking of bandwagon punditry, not to be too pedantic here but the sample size needed for a meaningful comparison of means that close together is somewhere around 500 per group. So I applaud Tim for gazing into the tea leaves with real endeavor but it would be nice to acknowledge that none of this stuff is dead cert, as the kids say, its more projection, rationalization and interpretation based on some very limited data in a scientific sense. But hey, there is very little data in sports that is meaningful from a scientific point of view so we are all kind of shooting from the hip. I appreciate that Tim tries. I often wish he sounded less sure of himself.

          1. Yes, 11 PL games is a terribly small sample to use for any projection, but he’s comparing to a much larger historical sample of the league, so its validity is pretty solid. Over a large enough sample, overperforming like this is not sustainable. But an entire season is only 38 games, still not a very large sample by any stretch. And it happens, as Tim has pointed out.

            In fact, here’s a heartening thought for all unapologetic optimists. Significantly outperforming XG and XGA over an entire 38 game season is an extremely tall order . But since we’ve done it over 11 games already, now our job is to do it for just 27 – statistically much easier than the same achievement over 38 games. If that’s not a reason to believe, I don’t know what is! 😉

          2. “Yes, 11 PL games is a terribly small sample to use for any projection, but he’s comparing to a much larger historical sample of the league, so its validity is pretty solid.”

            Well, strictly speaking it isn’t. The historical sample is still much too small for any meaningful comparisons, even between seasons. You can say the absolute numbers are more or less compared to this time last year and that’s fine, but my question is: should you care?

            The problem with BC as a stat is that it’s a rare event. Rare events are notoriously difficult to study. Tim may be totally right about all of this, or he may be way off base. It’s like if I tried to tell you the rate of strokes in America is increasing because we had 4,500 this time last year and so far this year we have 4,700. If you’re a health policy group, you probably say to yourself OK, let’s see where this goes and then at the end of the year let’s do a deeper dive to figure out what’s really going on. You don’t put the nation on red alert that we’re truly seeing more stroke until you’re sure it’s real. How can you be sure it’s real? You need to do a lot more math. I know I’m going to get the blinkered optimist tag for arguing this but all I’m really saying is that there is a lot less certitude about all of this than is implied by the article.

      2. Are you really telling us that 12 in a row unbeaten is rock bottom and that we’re deluded by the results into thinking that Torreira is a good player because they numbers suggest otherwise? Duuude… Far out.

        1. Are you saying that keepers will continue saving big chances, the strikers scoring at high percentiles and Arsenal being super happy?

          Its like guessing half the questions on your test paper and striking As consistently.

          1. “Are you saying that keepers will continue saving big chances, the strikers scoring at high percentiles and Arsenal being super happy?”

            I’m saying Tim paints a picture based on some numbers that’s very possibly way off base due to randomness and the way he defines his outcome of BC. I’m also saying viewer perception is not worthless.

  3. I wonder how much if the defensive instability is due to the heavy rotation recently due to the injury situation, and hiw much is because some of the defenders are “… beyond repair”. In other words, do we really need signings or just stabiluty and time for partnerships to form.

    In a slightly related point, i was thinking yesterday if one player who also has made in my opinion huge progress, but somehow nobody talks of him in the shadow of Iwobi and Torreira: Holding. The guy is so stable, so calm and composed that i really can imagine him as a long-term starter in a title-challenging team. Hopefully we can find one more like him. Curious what Emery will be able to make of Mavropanos.

      1. FWIW – I thought you made a great point about Mavro. Emery must rate him fairly highly if he sent Chambers on loan and kept Dinos. Here’s hoping he’s back for some of these Europa league games when we are not in do-or-die situations. He needs time with the team under Unai.
        Holding has really impressed me so far. He has much better touch/skill with the ball than our other center backs, and there’s a lot of room/time for him to grow and improve. And he’s got a little bit of the shithousery that Emery seems to prefer with guys like Sokratis and Lich. Gives us an edge we haven’t had back there.

  4. This is convincing and very well-argued. Results aside, we have a long way to go in terms of fine-tuning the defense/defensive midfield area.

    Ultimately there’s only so much good structure can give you – eventually we’ll need an upgrade in player quality (and athletisism!) to do the rest. It’s what both City and Liverpool did. Without Van Dijk’s repeated interventions we might have beaten Liverpool 4-1. A single elite player can really cover up a team’s flaws.

    Orbinho also tried to sound the alarm about our underlying numbers ( but ruins it a bit by being unable to properly interpret the stats he himself selected. For example, our passing numbers in the oppo half may well be down on last year. Doesn’t mean we’re ‘worse’ as Orbinho says, it just means we’re playing more vertically this season. And he later says we’re scoring more 2nd half goals ‘for no discernible reason’. But our no1 rank for distance covered in the Premier League suggests greater endurance levels and a better ability to both and attack and defend late in games.

    Good reminder that we’re pretty spoilt here with 7am’s ability to not only choose insightful stats, but properly analyse and write about the game.

    1. I’m convinced the more time people spend on twitter, the more they view extreme opinions as mainstream…

  5. Coming into the season, everyone knew this would be Arsenal’s major problem and starting off against City and Chelsea did us no favors. Defending is a team sport and cannot be coached in a single off season. It took Liverpool and City at least one season of being leaky at the back to accommodate the style of play they wanted to perfect (high press, playing out) and no small investment in defenders and goalkeepers along the way. Between Arrizabalaga, Becker and Ederson plus Stones, Laporte, and van Dijk, that’s quite an investment in the rearguard while Liverpool have also prioritized midfield with the captures of the highly rated Keita and Fabinho. Meanwhile, they’ve both stayed largely pat in forward areas, so they seem to be building their team from the back to front, figuring that the forwards are only as good as the platform and service they get from the players behind them. Brilliant, in my opinion, and it has worked for Chelsea as well after their capture of Kante and Jorginho.

    Arsenal on the other hand have been built from front to back. Our major investments have been, until this past summer, in the center forward and creative positions and so we have a squad that reflects that. The strength of this team this season will be in attack. When the players behind them cannot win the duels and control the match, the forwards will not see the ball much and Arsenal will lose games because the defense isn’t good enough to keep clean sheets while under pressure for the majority of a match. Emery’s tutelage has resulted in increased confidence and improved performance in several individuals but it’s clear that the longer term future of this squad in the back 4 positions will be in flux for some time. Perhaps only Hector Bellerin and Rob Holding will still be in Arsenal’s first choice back 4 this time next year.

    I do expect Arsenal to improve on that big chances conceded statistic and I think the “eye test” is already showing signs of improvement on that front. It’s also a truism in football that it’s better to be lucky than good, and Arsenal have been lucky thus far even if they earned a lot of that luck by grinding out tough wins from losing positions. I expect that they can be more consistent from back to front and over the duration of 90 minutes as Emery’s instructions continue to bed in and confidence builds in individuals and the collective.

    1. We conceded fewer big chances this season against City, Chelsea, and Liverpool than in the comparable matches last season. It’s not because we faced those teams that we are having problems with defense.

      1. No indeed! I think we gave away more vs. West Ham than vs. City. If we are debating the reasons for why we are conceding the big chances, I put it down to the new style of play combined with old vulnerabilitites. We can still be had with the ball over the top as in the past 10+ seasons, while the possession play at the back that Wenger abandoned last season is also contributing to transition opportunities after misplaced passes. Then there are set pieces. Lots of room for improvement!

      2. Is it possible that the style of play is significantly different against MC, Che, LVLpool vs those against more defensive teams?
        It might be that the team is taking bigger risks against “smaller/weaker” teams. If so, the approach might have some merits in that the weaker teams are expected to have weaker strikers and will convert at lower percentage rate.
        Of course, this theory falls apart if we find that the strikers of weak teams actually has a decent conversion rate…

  6. The stats aren’t pretty reading but do they show us improving over the 11 games, small sample that it is?
    Was our better defensive performance against Liverpool part of an improving trend after the baptism of fire against City and Chelsea?
    I ask as I’m still unnerved by a previous post that said Unai’s defensive stats were average to bad at all his former clubs, which would argue against hoping for significant improvement as the new system and/or new personnel bed in.
    Any comforting qualification o this possible?

    1. Yes, from a pure math standpoint these numbers mean nearly nothing. But in sports we fudge the whole type 2 error thing.

      1. OK, we get it mate. You, too, are a stats god. It’s a small sample size…check. Tim’s conclusions might be suspect, and we are not discounting that. Then again, Tim’s conclusions may be correct.
        Let’s wait for 10 years and get that sample size to something more statistically significant…or not.
        Perhaps we should also consider confirmation bias, selection bias, or the power of stochastic influence?
        Or, maybe…just maybe we can reflect on the point I think Tim was trying make – enjoy the ride, but beware the hype of 12 straight undefeated results because, in year-over-year numbers, our defense isn’t much improved.
        Or not…

      2. Stats god? 🙈

        “Tim’s conclusions might be suspect, and we are not discounting that. Then again, Tim’s conclusions may be correct.”

        Yes, that’s my whole point. I’m not sure everyone is as circumspect as you are about this.

  7. The statistics make for truly ugly reading which match with the eyeball tests. We still very much suck at defending.

    Cech has had two left feet in playing out from the back. Leno is better in that regard but has lost focus at key moments (what was that diving punch?)

    We can’t control the ball well enough, we can’t stop the opposition from getting into the area and in dangerous spaces.

    That being said, the Liverpool game was one in which we actually we were actually decent passing out from the back. We ARE capable of the cultured passing we’ve been known for in the past, starting with our GK. Against Liverpool Leno played some nice little passes over their press into midfield.

    So Xhaka and Torriera in the double pivot now? They were effective out of possession as well and were a major annoyance to the front three of Salah, Sane and Ferminho.

  8. An amazing article. Totally in depth. You are not the first to mention how our xG ratings have yet to improve the both offensively or defensively. @orbinho really did that. And you are right when u say like Man utd last season we are riding our luck. But unlike Man utd it’s Emery first season. The bad defence can only rub on him at least a year or two, the same way a good defence did on Wenger for three weeks years.
    However although our stats haven’t really improved remarkably, and our football still looks the same, the real Arsenal’s improvement have been intangible. Mentality, Unity, Agressiveness, in-fighting, Spirited, fight-until the end, Commitment, Hardwork. The stats won’t show you this. The last 5 years of Wenger’s team inconsistently displayed this values. Watching the gunners last 5 games there is something about their football that inspires confidence in me that they will be okay. Maybe it’s the fact that I got no expectations from them as this is surely a trial season and thereby I watch my games with no real pressure. But at a point in October, I began feeling more confident. Maybe we rode our luck in our first few matches after against Watford and Everton. However in the last 5 matches luck had nothing to do with it. More players played at their natural position and we expressed ourselves better. Rather we were unlucky against Palace. Emery’s PL inexperience showed because he would have had a better approach to the game. I think this is something he would suffer this season as he gets to learn about PL pitches and stadia.
    I am really impressed with the improvement since the beginning of the season. I like the way we use the whole pitch to play. I like the confidence with which our back 4 plays out from the back. I love the directness the attack. I like the fight from the midfield (Against Palace we bullied em). But there is way more work to do as the stats have proved. We need to consolidate on the Liverpool experience.
    I like reading the 7amkickoff at 7am. Keep it up Tim. Hopefully you always want a Nigerian to share his thought from an African point of view from here in Nigeria on your blog.

  9. 70% of the Earth is covered by water. The rest is covered by Torreira.
    – An amusing tweet

    And here we have Tim saying that Torreira may not be giving the back 4 great levels of protection? I mean the tweet is only a slight exaggeration, no?

    As a general view that there’s much improvement that needs to be made by Arsenal, especially in defense, I totally agree. But I agree based on the eye test. I like stats because they show trends, but I also do not instinctively understand stats. Are all big chances created equal? Obviously not. They don’t account for a slightly overhit pass, a bouncing ball, a defender fast closing in rushing the shot. (Or do they?) Wenger used to say football has a billion different techniques as opposed to say golf or tennis. This is where stats ‘fail’.

    Arsenal are failing to keep the opposition big chances down. The stats are failing to explain/describe how Arsenal are getting the results they have.

    Personally, I felt the lack of availability of our experienced CBs caused a lot of defensive issues, while Mustafi was having a stinker a game. Mustafi has said he gets more specific instructions from Emery and is liking the effect it has on him. Holding has grown. Sokratis added some knowhow. And I think it has helped even when the defense fails.. Likewise adding Torreira will have an effect. He’s new to the league and the team. He’ll get better at reading the game. We’ll improve our defensive stats.

  10. Excellent analysis, Time. As I said before, you spoil us with this stuff. However, can I go with just my eye for a bit?

    My analytical eye (which I trust based on using it for 25 years; and not based on any statistical info) tells me that Arsenal, having been wide open like a Dutch sluice gate in the first half dozen games or so, have tightened up. And it’s due to 2 things…. one, Torreira settling into the team (rather than coming on as sub), and (2) Leno taking over in goal. Yes, we gave Liverpool chances but they are a high quality attacking side . I’d say that limiting them to 1 goal is an achievement, albeit done with a bit of luck.

    I think (and I’m guessing here) the stats will show that we gave up more big chances in the first half dozen games than in the second half dozen ones, Palace and Liverpool excepted. In fact we went through a period of a couple of comparatively easily achieved clean sheets.

    Leno transmits calm, clears and distributes much more accurately and has shown himself to be a decent shot stopper. I really wish that Tim loved Torreira as much as I do, because I haven’t seen such a tough, combative and effective Arsenal midfielder in a long, long time. He ran the show against Liverpool. I think he’s a boss, and is one his way to becoming one of the best in his position in the league.

    That said, Tim’s right in saying that we shouldn’t get ahead of ourselves on Xhaka and Torreira… the great synergy they had was only for 1 game and (as I said about Mesut) we need to see more consistency before declaring it great. But they showed in that game that their styles can be complementary and effective. Xhaka, who isn’t my favourite player, took a giant step forward as an Arsenal player on Saturday. He reminded me of learning to drive. It all looks so hard despite all the time put in, and then one day it just clicks. He himself echoed what I was thinking… it’s the best he’s ever played for Arsenal. And to be fair, he played well when out of position covering for Nacho.

    Related to the calming influence of Leno has been the play of Rob Holding. I don’t know enough tactically to offer a good analysis on complementary CB styles, but it seems to me that he and Mustafi are more of match than Mustafi and Sokratis. Want to throw that one to Josh for his thoughts. But as any rate, Holding has been good. He’s a very reliable outlet for playing out of the back, is assured in the tackle, reads well, and finds himself rarely haring to get back into position as the often scatty Mustafi is.

    So Im seeing green shoots. Is anyone else?

  11. I can’t disagree with the stats (even Statsbomb’s proprietary xG model doesn’t like our performances) but I wouldn’t suggest that we haven’t improved. It seems the issue is that we’re doing things very differently to how we used to go about defending and therefore we’re not very good at defending in an entirely different way.
    I wonder if what we don’t see in the stats would be better represented in a good non-shot xG model? In as much as, Fergie’s last great team had a very unusual attack that didn’t produce all that many shots because of the way their key-pass selection was coached.
    Relating that to the difference between old and new Arsenal, I wonder if we were conceding even greater volumes of potential shots that simply weren’t being taken, as a result of the ineptness of our opponents? Hence why we regularly got absolutely battered by top teams even when they weren’t playing particularly well.

  12. Maybe it’s all planned and the manager just wants to trick the opponent into thinking we’re in trouble….

  13. I missed the start of the season on account of my work. But in the games I have seen so far, I have not yet fully understood the hype around torreira. He is probably just about above average, but definitely not the awesome tiny lynchpin of the arsenal midfield as the Guardian says he is. I hope I am proven wrong in the future though.
    As for Xhaka, I dont think he’s improved that much cause he is not that good a footballer to begin with. We would be safer without his services. I say safer because he is a liability in 1on1 tackles. And his long passes are normal but not spectacular or caviar.
    Let’s try and make a profit on him when he is at his peak and get a proper Pirlo please.

  14. I’m glad that you bring those stats up. It shows that a lot of fans have seen the Emery era with rose-tinted glasses. It also confirms that it is still Wenger’s team in the sense that there are only 2 new players in the starting lineup (Leno & Torreira). I think there are 2 major factors for Arsenal’s good start to the season: 1. Luck (i.e. the discrepancy between big chances allowed and goals conceded); 2. Chemistry. Something similar happened at Leicester the year they won the PL title. Leicester over-performed and the change of manager (Ranieri replacing Pearson) had an impact in the dressing room. Now that Wenger is gone, there’s obviously less negativity around the club. Arsenal players also needed to hear a different voice.

  15. I would tend to agree with the conclusions even on an eyeball test. There are some positives but the much lamented negatives of old do not really seem to have been dealt with. Top 4 might still be possible not because we have improved by leaps and bounds but because Spurs and Man Utd look ripe to be overtaken

Comments are closed.

Related articles