FRI, 2:30PM ET
MAR
NAN
FRI, 2:30PM ET
SOC
ELC
FRI, 2:30PM ET
FRE
STU
SAT, 7:45AM ET
WBA
ARS
SAT, 10AM ET
MUFC
HULL
SAT, 10AM ET
SWA
CRY

Where To Go When Our Crude Tools Can’t Capture Carlo’s Trequartista

3747581208 6b049b2914 o 300x214 Where To Go When Our Crude Tools Cant Capture Carlos Trequartista

Nicolas Anelka, the third of two forwards in Carlo Ancelotti's 4-4-2 (Photo credit: jaymeydad/Flickr)

Numeric annotation of formations was on my mind before I started firing off emails late last year. Even before I got his response, I’d seen Michael Cox’s thoughts on the practice, the Zonal Marking owner giving his views to Tom Williams. When the AFP writer asked Cox why we annotated attacking fullbacks as part of the defense when their average position is often in midfield, Cox called himself a traditionalist. ZM depicts the base, defending formations, he said, though there were exceptions for sides that press intently.

When I interviewed Michael a few days later, we redressed the discussion.

“I’m a bit of a traditionalist in the way formations are depicted,” Cox said. “I don’t really like this kind of 2-4-3-1 (idea). I don’t like two fullbacks in midfield, even though it’s slightly more accurate in terms of their average position on the field.”

The reason why we note four defenders when describing formations as 4-Y-Z is because some players, at their core, are defenders, be it by tactic, position, or both. That’s what three band notation is all about. At some point, you need to communicate how many defenders, midfielders and forwards are being played. The annotation doesn’t try to describe roles, and it only poorly hints at shape. It’s only goal is to give a quick and cursory description of formation.

As we seek more and better ways to describe formations, three band notation is proving insufficient. Whereas two years ago, when Joachim Löw switched the German national team’s approach, 4-5-1 was an acceptable way of describing the side’s set-up, now nobody would prefer that description over 4-2-3-1, a description that highlights the pivot’s players. In three band notation, those players are relatively lost.

But another complication of three band notation appeared this weekend in the reporting on Chelsea-Liverpool. While the problem categorizing defenders was evident in labeling Liverpool’s formation a 3-6-1, our comfort with the average, high positions of defenders Martin Kelly and Glen Johnson (the wingbacks) excused that. And Chelsea’s three levels between defense and strikers being crunched into a 4-4-2 has become so conventional as to be expected, with the idea of calling it (the more accurate) 4-1-2-1-2 spine shivering to the point of nausea.

No, the new problem hearkens back to the defender-versus-midfielder issue, only this time, the problem is farther up the pitch, one that asks: What’s the functional difference is between a midfielder and a forward? And, how does that manifest?

Consider these three players, with these Guardian heat maps showing the positions from which each player attempted their passes during their weekend match:

Player 1 Player 2 Player 3
Screen shot 2011 02 08 at 1.20.38 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista Screen shot 2011 02 08 at 1.23.10 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista Screen shot 2011 02 08 at 1.33.40 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista

Player 2 has the greatest range of actions. Most of Player 3′s are concentrated in the opponent’s half. Player 1′s somewhere in between but still providing the profile of somebody who does very little in his own half of the pitch. That’s a pattern you normally see from a forward or (perhaps) a midfielder whose side is dominant in possession. Player 1′s team did control 59 percent of his match’s possession. Player 2′s had 52 percent, while Player 3′s saw 57 percent of the play.

Two of these players were described as forwards (supporting strikers) in most post hoc analysis – the match reports where we’re talking if teams played a 4-4-2, 4-3-3, or 4-2-3-1. A third was described as an attacking midfielder, some even breaking out the trequartista card.

Can you guess which one was thought the more midfielder-y? Consider it a game, one with a to-be-decided prize, though perhaps more information is needed.

While Guardian chalkboards are the most ubiquitous tool for this find of inquiry, but they’re not the only ones. Both ESPN and Fox Sports’s web sites have tools which allow you to look at average positions and heat maps. Consider first Fox’s average positions, where I’ve circled the players we’re trying to evaluate:

Player 1 Player 2 Player 3
Screen shot 2011 02 08 at 8.18.13 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista Screen shot 2011 02 08 at 8.19.36 PM1 Where To Go When Our Crude Tools Cant Capture Carlos Trequartista Screen shot 2011 02 08 at 8.16.48 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista
In Player 1′s picture, the team is attacking from bottom-to-top. In the other two screenshots, sides are attacking from top-to-bottom. Starters and substitutes show.

Cute pictures, right? It’s almost hard to take them seriously in the context of an overly analytical piece. Regardless, they provide some depth to the Guardian chalkboards. Player 1 had, by far, the more advanced positioning. Players 2 and 3, have near identical average positions. Again: Two of these were thought of as forwards. One has not.

One more tool, just to put this ailing horse down. Here’s what you’ll find at ESPN’s Soccernet version of a heat map, which seems a combination of the Guardian’s (event-based) chalkboards and Fox’s average position:

Player 1 Screen shot 2011 02 08 at 8.57.52 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista
Player 2 Screen shot 2011 02 08 at 8.57.30 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista
Player 3 Screen shot 2011 02 08 at 8.56.52 PM Where To Go When Our Crude Tools Cant Capture Carlos Trequartista
Player 1 is attacking right to left. Player 2, left to right, as is Player 3.

Figured out which one is thought the attacking midfielder? I hope so, because it’s time for the reveal:

After looking at all three of these measures, Player 2 would seem the most likely to have played as an attacking midfielder, though that player is Carlos Tévez. Player 1 seems to have the most forward-esque positions and behaviors. That’s Wayne Rooney. Somewhere in between is the player thought the attacking midfielder: Nicolas Anelka.

Before last week, Anelka would have been the least likely of that trio to be painted a midfielder. Both Rooney and Tévez have shown the ability to play a more withdrawn role, where as Anelka profiles as more of an outright forward, if often a wide one. But against Sunderland, in the middle of last week, Anelka was deployed at a level behind the strikers, and when he reprized that role on Sunday against Liverpool, he was naturally thought the tip of a midfield diamond normally associated with Carlo Ancelotti’s 4-4-2 variant. Thus the experts – such as Jonathan Wilson and Michael Cox – described it as 4-4-2, because … well, that’s what that system is. That’s a 4-4-2, even if one of the players is a forward, had the positioning of a forward, and in the events the Guardian and ESPN tracked, was essentially a forward.

And Wilson and Cox are right, right in a way that’s starting to illustrate the limits of numeric annotation. They’re the the same limits that have the 4-Y-Z label for systems that see two of the defenders act more as wingers than fullbacks. The notation has evolved beyond being a way to describe how many defenders, midfielders and forwards are deployed. It’s becoming language, nomenclature: short-hand that you either understand or you don’t.

“If I say to you a 4-2-3-1,” Cox explains, “you know what it means, and I know what it means, and if it needs more explanation, I think it’s better to elaborate on it afterward rather than confusing people with a new formation.”

Thus you have the three forward system Chelsea used Sunday described as a 4-4-2. It’s less confusing that way.

At least, it’s less confusing to me, and it’s less confusing to Michael. And for those with the patience to make it past nine screenshots and their corresponding descriptions, it may also be less confusing, even if the discussion is slowly becoming less accessible to those trying to catch-up. After all, how does numeric annotation make sense if the numbers become more meaningless?

It’s a slippery slope of which Jonathan Wilson is keenly aware.

“I think there’s a couple of (places) where the modern discussion falls down slightly,” Wilson told me, late last year, “and one of them – possibly I’m sort of guilty of as well – is that you can get too obsessed with the numbers. You can be too obsessed with 4-4-2 or 4-2-3-1. I think we should always remember, that’s a basis. That’s sort of a very crude tool that gives us a general way of talking about something.

“We shouldn’t get too carried away in thinking there’s some great magic in these numbers. They’re very, very crude tools.”

And as with any tool, crude or otherwise, we should be careful not to misapply its use. After all, crude tools tend to break.

Keeping attacking fullbacks in the first group of a 4-Y-Z is one thing. You can always explain that decision as Cox did (describing base defensive positions) or I did (at some point, defenders are defenders). Describing a three forward set-up as 4-4-2 is different. At that point, the annotation is just a proper noun, like calling Ford Motor calling one of its models a Fiesta, though the compact car is nothing like a party.

I’ve decided to go the other way, to use a different term to describe the same formation. I know the most effective way to communicate Chelsea’s formation is to call it a 4-4-2, but that doesn’t feel right. To me, the numeric annotation should describe the quantities of defenders, midfielders and forwards, even if the rest of this discussion’s evolving beyond it. To my eyes, Chelsea played a 4-3-3 on Sunday, and I’ll take it upon myself to establish the appropriate context for my description.

Unfortunately for me (and illustrating the wisdom of Wilson and Cox), I usually need 1900 words, an undo amount of defiance, and enough pictures to fill a PowerPoint before I’ve established my context. And as a result, I probably won’t be getting any email offers from Sean Ingle, which may make my next assertion even more foolish: Perhaps it’s time we consider the world beyond numeric annotation, even if that world includes even more words.

As four and five band notation are more commonly used, we see a reflection of people’s desire for more information. The 4-5-1 description is insufficient. People want to know if it’s a single (4-1-4-1) or double (4-2-3-1) pivot, and for those who have come to augment the numbers with symbols (4<4>2, shudder), there’s a desire for even more depth.

At some point, we may have to admit it: Simplistic numeric notation does not meet the demands of an ever more curious readership, a readership intelligent enough to deal with more wordy, more accurate descriptions. It’s the same readership that has made for the rousing success of a just-over-one year old Zonal Marking. It’s the same readership that’s made an Eastern Europe specialist into a best-selling tactics historian.

It’s the same group that will understand if you eschew 4-4-2, 4-3-1-2, 4-1-2-1-2, or 4<4>2 for a better (if still imperfect) description: On Sunday, Chelsea used a central defensive pair, attacking fullbacks, a holding midfielder behind an offset central midfield pair, with a trequartista supporting a striking tandem.

That may be too much jargon, but it’s more accurate. It also moves us away from a crude tool we’re starting to misuse, a numeric system that we want to do more than it was originally designed to do.

And ultimately, that undo reliance deceives. Tying ourselves too closely to that numerology, according to Wilson, takes us away from the realities managers face, a world that’s rarely defined by hyphen-delimited numbers.

“It’s almost become counter productive to think of players having a position,” he said, recalling a thought exercise he’d engaged regarding Steven Gerrard. “Of course you have to think of them having a position. That’s how you characterize them. That’s how you have a vague sense of what they do. But I think the best coaches, they don’t think of Steven Gerrard as being a left-sided midfielder or a player that plays off the front man or a holding midfielder. They think of him as being a bundle of attributes, and they think how can I best use that in my team.”

“I think that it has to go that way around. You can’t start from the basis of having the numbers and saying how do I fit the players into this? You got to have the players first and then say how do these players best fit together?

“Obviously the terminology – 4-2-3-1, 4-4-2 – has been around for a hundred years and it’s going to stay around, and it is useful, but I think we have to be aware there are limitations to that terminology, and I suspect in time we will develop, as they have in Italian, more sophisticated terms for different players or different roles.”


Quotes from Jonathan Wilson and Michael Cox were taken from larger interviews, discussions which will be distributed in audio form later this month.

For full versions of the Guardian chalkboards used, you can visit Rooney, and Anelka, though the Tévez one will not save, thus preventing me from linking to it. To recreate, select 2010-11, Manchester City, Sat 05.02.2011 – West Bromwich Albion (H), C Tévez, Passes, Show heatmap.

A hat-tip to Graham MacAree, writer at SBNation.com/Soccer and manager of Chelsea blog We Ain’t Got No History, for helping me flush out these thoughts, a process that was described by onlookers (who clearly don’t know me well) as “the nerdiest conversation ever.” Graham far prefers using ’4<4>2,’ with good reason.

Richard Farley (me) is the host of the EPL Talk and Major League Soccer Talk podcasts and is an infrequent author at each site, preferring to save up all his words for one, huge missive. You can email him or talk to him on Twitter or send him snail mail (everybody likes checking their mail). He’s also the soccer editor at SBNation.com. Feel free to contact him to talk football or share recipes.

This entry was posted in General, Leagues: EPL. Bookmark the permalink.

24 Responses to Where To Go When Our Crude Tools Can’t Capture Carlo’s Trequartista

  1. Hmmm says:

    Great piece Richard, very interesting read. I’ve never had any problems with the banded notation (3 or 4 band varieties, any more seems tedious) and have noted its obvious shortcomings but never really went further as paired with a lineup it communicated a basic formation. I think the basic point, that the game is too complex for a simplistic numerical formula to describe is pretty basic. With Trequartista in the title I should mention that Spanish seems to have a more developed vocabulary for describing these complexities then English (i.e. traditional winger, holding midfielder etc.)

    • Dave C says:

      I think Trequartista is actually an Italian term (although I might be wrong, and the Spanish word is probably similar). But you’re right that other languages seem to have some more colourful and accurate terminology for positions, a point that I think ZM has discussed occasionally. I particularly like “Lateral” to describe left-backs and right-backs. Personally I’ve always felt a little weird calling them “Full-backs” when they are not the furthest back player on the field. And it causes confusion talking to a lot of American soccer fans, who tend to call center-backs “Full-backs”.

      • Richard Farley says:

        Trequartista is an Italian term. If my “I really don’t know” Italian is right, it literally translates as three-quarters. That implies a location rather than function, but I think it’s safe to say in this parlance it’s used more often to refer to the latter. Or, at least, a combination of place and function.

        I wish my mind was better with such things, but there seem to be terms in all languages that refer to more specific roles that we have labels for in English. I’ve heard Russian and Japanese ones and … hmmm, I guess I don’t dive into French much. I’m sure there are some there.

        One thing that comes up often, when talking about these things, is the lack of tactical variation, at a given time, within the English game. No idea how true that is, relative to other cultures. However, without an array or things to describe, there’s no need for an array of terms?

        • Dave C says:

          If my “I really don’t know” Italian is right, it literally translates as three-quarters.
          My “I really’ don’t know” level of Italian agrees with your translation. We obviously are both graduates of the Starbucks School of Language! I’ve alwats interpreted it to mean that you are three quarters along the spectrum between midfield and forward/striker.

        • Richard Farley says:

          We clearly need more coffee choices and given the synergy of java and football, I hope somebody reads this and starts opening a Brazil-themed chain of coffee houses.

    • Richard Farley says:

      Thanks, Hmmm.

      This will come up again in the comments, but I have a large bias against four band, let alone five band, notation, something that’s ill-served me in the context of Chelsea-Liverpool. Aside from 4-2-3-1, I can’t think of another time I’ve used a four-band annotation regularly.

      That should change, because as Tom notes in the next comment thread, the most straight forward way to communicate Chelsea’s set-up without going to five bands or abandoning numerals all-together is 4-3-1-2.

      But I definitely agree that the game is too complex for simplistic descriptions. Those simple terms are fine, though, as long as we don’t try to have them describe the complex things.

  2. Tom says:

    A thought-provoking read, Richard, and thanks for linking to my interview with Michael. Surely the simplest way of referring to Chelsea’s formation against Liverpool would be to call it a 4-3-1-2?

    • Red20 says:

      I was wondering this as well. If he wasn’t quite an out and out forward while certainly not being a full mid- why not give him that different distinction. Even if you were to draw out a rudimentary 4-3-1-2 on paper for someone, I would think that would better encompass his actual role, than if you were to draw out a 4-3-3 or a 4-4-2.

      Either way, fantastic article. Would love to see more stuff like this on here.

      • Dave C says:

        Yeah I agree with that too – I think 4-3-1-2 is a better way of describing Chelsea’s recent shape. It’s still not perfect, but I think it is much closer to the truth than simply calling it a 4-4-2.

      • Richard Farley says:

        Thanks, Red20. I wish I had time to write more of EPLTalk, as it’s a great site (which is why I love doing the podcast for it). I hope to contribute more in the future, though this piece was actually born of a strange motivation.

        I mention in the piece that I’m questioning numeric annotation. I actually don’t like it very much, at all. It’s a square wheel. When I do use it, I try to stick to three bands, for historic reasons I allude to in the post. I’m not sure these biases served me well before this piece.

        Having written this, I’m left wondering if I need to adopt defiant stances, sit on them, and then wait for a glimmer of relevant context before pouncing. If yes, that’s fine. I may have just defined what blogging is. If not, how does this discussion I’m having with myself (and, hopefully now, others) move forward.

        Or, maybe I just caught up in the Torres-to-Chelsea thing and overwrote the story.

        One of the two.

    • Richard Farley says:

      Tom: It’s awesome to get a comment from you. I really enjoy Football Further and following you on Twitter.

      I completely agree on 4-3-1-2, and in my stubborn insistence on trying to keep everything three band when not 4-2-3-1, I’ve come to think my writing didn’t help anybody on Sunday. Crusades are better waged in posts like these than match reports.

      But another reason I shied away from that description was its emphasizing Anelka more than the opposite end of the diamond, Nicolas getting a “1″ where Mikel did not. However, now I think that reasoning doesn’t hold up. As you said, surely that’s a better option then either the five or three band versions.

  3. dawleylad says:

    A brilliant post – thanks!

    Dawleylad
    http://www.keelbyunited.co.uk – Want to know about local football?

  4. Drew M. says:

    More articles like this one, please.

  5. David G says:

    Like Drew M said

    “more articles like this one, please”!

    • Richard Farley says:

      Thanks, dawleylad, Drew M, David G. You guys made my Wednesday.

      • Drew M. says:

        For a follow up, how about a ramble on what it means to be a Number 9 or 7 or 10 or whatever. Not just the history of numbering, but what sort of approaches a stereotypical 9 would bring, rather than, say, a 7.

        Then you can look at current players and see what number they best fit.

        • Richard Farley says:

          That’s a good idea, Drew. My interest may lie in describing the traditional conception of the roles in light of formation changes that have continued to invert the pyramid (perhaps up until recently). However, I think such a post could serve the same purpose.

          We’ll start with the No. 1 … :)

  6. Dave C says:

    I agree with everyone above – great article. Ridiculously long, but worth it!

    Just a couple of technical questions about the various screen shots –

    1. The average position graphics (the second set of pics) seem to show 14 players on each side. Presumably this is because it includes the starting XI and the 3 subs. Is there any way of showing JUST the starting XI (or combining the players subbed-off with their replacements so that we still only have 11 on the field). I think those graphics could be the most informative, since they provide context of a player’s position relative to the others, but it would be helpful if there were only 11 on the field so we could identify some kind of basic formation.

    2. On the 3rd graphic (the heat maps), both player 1 and player 2 have a red circle near their own right corner flag, and player 3 has an orange circle in that same location. Is that some kind of glitch? I assume it must be, because the first set of graphics show 0% for each player in that defensive corner. I can’t imagine why those three players would see significant action in their own defensive right corner.

    • Richard Farley says:

      Agree on the ridiculously long, Dave. I’ve been reading too much SwissRamble.blogspot.com. Then again, if you’re as good as he is, you can write all you want.

      1. Unfortunately, there’s not. Those are the Fox graphics you allude to. If there’s a way to isolate one player or just the starters and somebody knows how to do it, that’d be very helpful for me.

      2. I’m not actually sure. I went through the data on the site in search of an explanation for Rooney’s, but one wasn’t readily available. I do notice, when looking at the ESPN heat and location mapping, there there seem to be periodic data problems that can create some weird reporting, particularly when looking at average positions early in the matches. Perhaps those red marks are examples of “unclean” data? Just speculation.

      • Dave C says:

        Thanks for the reply. I hadn’t heard of Swiss Ramble, but now I’m sure to add that to my list of “things I do instead of having a life”.

        The reason I asked question #1 was that I think the position of the rest of the team provides a lot of context. Two players on different teams might physically occupy similar areas of the pitch, but if one is the farthest forward player on his team (eg if his team play a counter-attacking game with everyone dropping back, or if they play a 4-6-0 formation, or he’s a “false 9″), he will obviously have a different role to someone who might occupy that same area but with one or more guys ahead of him.

        For question #2, I’m sure it must be a data glitch on Fox’s part. Especially since it seems to recurr for all three players.

        • Richard Farley says:

          I agree on the position of the team, but I think, more generally, we have to consider the dynamics of the game we’re talking about. I think you can see that in comparing Rooney’s positioning to that of Tévez and Anelka. Through experience we know that Rooney and Tévez’s roles can play as quite similar, all things being equal, but because Manchester City was in control of their match while United chased for a half, Tévez’s data depicts a #10 while Rooney seems a #9.

          I suppose that’s why I used all three of these data sets in addition to providing the possession numbers.

  7. tonyspeed says:

    all these terminologies are antiquated anyway. They were created way before the invention of wing-backs or even total football or even the holding mid-fielder. So we are trying to shove new wine into old wineskins. These inaccuracies are bound to come up. We really want one notation that describes formation and a second that describes intent.

  8. Earl Reed says:

    I’m a little late to the game with this, but this was an excellent article Richard. I’ve definitely felt that that there’s a lot to deplore about the traditional numerical system since I’ve started putting articles and diagrams out there. I think that is a huge reason for the success of Zonal Marking, and to a lesser extent what I’ve been producing, because a pictorial says more to the uninitiated than a grouping of numbers and hyphens.

    Take this week’s Carling Cup Match:

    http://epltalk.com/arsenal-1-2-birmingham-city-carling-cup-final-review-29649

    You can see the bands in the depiction, but they would be a snapshot compared to what happens in a game. The time it takes to scrutinize the game via DVR and charts is often overkill, but to say Birmingham’s formation was 4-5-1 is not accurate. Bowyer (and to a lesser extent Gardner) pushed ahead at times to become a second striker, both defensively and as Zigic tried to win aerials. It wasn’t static at all, but very much fluid. And Arsenal’s backs were at a level with Song and Wilshere much of the time, so literally one could have called it 2-4-3-1.

    (Let me clarify this: I often verify my points by pullling from the Guardian/ESPN/FOX Soccer triumvirate of precise data presentation, but this info isn’t available for Cup matches, and through both the FA and Carling Cups I have become even more anally-retentive about my depictions.)

    I have typically reported defensive formations in my charts. When teams begin to support an attack, all hell breaks loose. Not literally, obviously the players are trained in offensive tactics and have knowledge of rotations, positioning, etc. But to the untrained eye, things get a little chaotic. Denoting such plays, swaps, and shifts in a diagram that doesn’t encapsulate the time dimension is fruitless IMO.

    It’s the defensive positioning that is most important in my opinion. This is because it informs about the obstacles that a team looking to advance a ball from the goalkeeper to the final third will encounter. Even then, it’s not perfect, like in the Carling Cup game, “4-5-1″ doesn’t instruct that that 5 midfielders were exceptionally narrow, looking to choke the Arsenal center-pitch attack. But a diagram can.

    In a world of acronyms, I really think that conventions like the D-M-F nomenclature will persist, simply because it’s easy to say and anyone who cares tends to know what it means. It doesn’t mean it should be the end of the story though.

    • Richard Farley says:

      I don’t want to ruin a future piece, but I’m becoming more and more convinced the DMF and A-B-C, A-B-C-D systems are outmoded. If they’re surviving because they’re easy, the they’ll soon be the batting average of football.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>