Friday, January 13, 2017

Defensive stats reveal style not performance

In this piece we’ll conclude the look at individual defensive stats that we started last week….

Clearances
Clearances are a spectacular stat to which there’s usually less than meets the eye. Although you can make some interesting comparisons between full-backs or holding midfielders, it’s really only significant for central defenders. Worse, the stat is severely skewed, because defenders who play for teams that have to defend a lot tend to rack up large numbers. The league’s King of Clearances is James Collins, who’s invariably near the top of the list. Last year’s leaders per 90 minutes were Lamine Koné at 9.4, followed by Collins, Ramiro Funes Mori, Younes Kaboul, Ashley Williams, Ryan Bennett, Steve Cook, Sebastien Bassong, Phil Jagielka and John O’Shea. Two from Norwich City, Sunderland and Everton. Note that although it can help, you don’t have to be great in the air to make the list. Funes Mori, Bennett and Koné were average in aerial duels, Williams, Kaboul, Jagielka below average. Aerial monster Virgil van Dijk was a couple of places short of the top ten, as he is this year as well.

Incidentally, Koné’s 9.4 was quite low for the league leader; usually it’s over 10, and sometimes up around 12-13. This year’s current leader is Cook, at 9.8, followed by Papy Djilobodji, Jan Vertonghen, Winston Reid, Williams, Ben Mee, Jagielka, Calum Chambers, Nicolas Otamendi and Will Keane. Two each for Burnley, Everton, and for Sunderland if we go to 11th place and Koné again.

In one way, the more clearances you have, the less talented you are, or at least the less continental. That’s because clearances are defined (explicitly for Opta, implicitly for Whoscored.com) as getting the ball away from your own goal without any intended recipient. So to overstate it a bit, clearances are for hoofers. A check shows that teams in La Liga average 11.4 to 24.1 clearances per game, whereas Premier League teams average 20.8 to 31.1. Last year, when Funes Mori and Jagielka were in the top ten, their teammate John Stones was next-to-last among central defenders. In case you hadn’t heard, Stones likes to play it out from the back. This year Cesar Azpilicueta and Gary Cahill are 1-2 at the bottom, __with Stones fourth from last. Oddly, Wes Morgan is quite low, only three places above Stones, and both he and Robert Huth have dropped from their championship season. Neither is what you’d consider to be a ball-playing defender. That’s worth a closer look when you watch Leicester play.

Along __with such season-by-season comparisons, clearances tend to be most interesting at individual match level. They can tell you how a defender stayed strong under siege, like Nathan Aké’s 15 clearances as Bournemouth held off Stoke 1-0 this season. But again, you can have your share of clearances and still get beaten for the winning goal, as Aké did this year when he had eight clearances against Southampton.

The highest single-game clearance numbers actually have little to do with the result. Because defensive headers from long balls into your half of the pitch usually count as clearances, the big numbers come when you’re facing an aerial bombardment. Thus Otamendi notched 19 clearances against Burnley this year, and OMG here-he-is-again-call-his-agent Vertonghen managed a ridiculous 21 against the same side. And if you’re very lucky, you don’t have to go far up the pitch at all. In the immortal match when David Moyes’ Manchester United delivered 82 crosses against Fulham, very tall person Dan Burn sent back 22 of them, 20 from inside the penalty area.

Blocks
We’ve already taken a brief look at blocks, in the first piece in this series. As noted, a high number of blocks usually indicates a defender and a defence that sits back. James Collins has been a regular here too, as was Richard Dunne in his time, and more recently Ryan Shawcross and Gary Cahill. with the advent of the three-man back line, we can expect players in the wide positions to have fewer blocks than usual, and ones in the middle to have more. For example, Cahill’s blocks are at an all-time low this season, Michael Dawson’s at an all-time high, although Dawson’s are also influenced by the way Hull have played. The central defenders at the bottom of the list this year are Shkodran Mustafi by a wide margin at 0.2/90, Azpilicueta (influenced a little by his time at FB), Otamendi, Koné, Stones, Jonny Evans (a little FB there too), Joel Matip, Cahill, Jose Fonte. The surprise is Koné, particularly since Djilobodji is so high. We’ll come back to this contrast in a while.

But next is a stat which you rarely hear about, but which is one of my favorites, and measures a precise and valuable skill: crosses blocked. Your leaders for last year (per 90 minutes, as always): Christian Fuchs at 1.8, followed by Aké, Brendan Galloway, Paul Dummett, Cédric Soares, Neil Taylor, Danny Simpson, Angel Rangel, Aly Cissokho and Martin Olsson. That both Fuchs and Simpson make the list is very much due to Leicester’s overall defensive set-up, but it also seems likely they were instructed to block where possible. (They’re both in the top five this year so far.) Swansea has both Taylor and Rangel on the list, and Everton would have had both Galloway and Oviedo if the latter had played two more minutes. So coaching must play a part here. At the bottom was Daryl Janmaat (and I’m sure Newcastle fans would have a lot to say about that, since Dummett was in the top ten), then Bacary Sagna and Gael Clichy for Man City. I’ll let you check out this year’s list on Whoscored.com.

Fouls
Fouls are a relatively clear and simple stat which can tell you a lot about a player’s style. But if you’ve never seen lists of the top foulers, you might be in for a surprise. Here it is for 2016-7, per 90 as always: Josh King at 2.4, followed by Gareth Barry, Nordin Amrabat, Christian Benteke, Fernando Llorente, Marten de Roon, Paul Pogba, Callum Wilson, Victor Wanyama and Zlatan Ibrahimovic. Five out of the ten are strikers. That’s a bit of a shock until you realize that they’re always battling for balls both on the ground and in the air.

But the foul numbers that tell us the most are about midfielders and defenders. Strikers pretty much have to foul a lot, midfielders and defenders not so much, depending on their job. Gareth Barry second on the list is telling – he just doesn’t have the pace to keep up anymore, and frequently fouls because there’s no alternative. Amrabat is a surprise; he’s neither a defensive midfielder nor a striker, although he does play some defence as a wing-back. But the high ranking suggests indiscipline on his part. Pogba and De Roon are aggressive central midfielders, and we know about Wanyama. By the way, a couple of places down is Roberto Firmino, who’s on the list because he’s such a strong presser.

A central midfielder may have to foul now and then for tactical and intimidation purposes, but when a back line player fouls, it’s almost always a failure of some kind. The top fouler among central defenders is Jordi Amat, who we’ve met before as an aggressive tackler and shaky defender. The top fouler among full-backs is Allan Nyom, who has a rough-and-ready style, although I thought he was a better defender than that. In fact, his fouls are up from 1.3 to 1.8 this year, a pretty large jump. Is Tony Pulis a factor?

As suggested, though, fouls come with the territory if you’re aggressive. So if we want to measure a defender’s precision, we can divide fouls by tackles + interceptions: the higher the number the clumsier you are, and the lower the number the more surgical. I haven’t seen this stat anywhere, so I’ve done a few calculations for central defenders this year, and the highest number belongs to…Robert Huth. Damien Delaney is second. Good stat, then. Two players who foul a lot but make up for it with high totals in the other categories are Shkodran Mustafi and Winston Reid.

Who’s the most precise central defender? Very much to my surprise, it’s Curtis Davies. He’s been great one day, poor the next, but he commits very few fouls. John Stones follows, and perhaps it’s significant that they both often play in three-man defences, where the fouls may be spread out a bit. Among centre-halves in a back four, the most surgical (as Daniel Storey probably could have told you) is Steve Cook.

So what does this all add up to? Anomalies in general may tell us worthwhile things. Tackles among holding midfielders and full-backs seem to a degree to track skill; tackle percentages can be useful; interceptions for holding midfielders have some application; maybe cross-blocking can measure performance as well as tactics; fouls can tell you about defensive precision. And before I forget, aerial duel percentage, which we covered in the opening article, is a worthwhile measure of central defenders. But that aside, when we’re looking at centre-halves, those crucial members of the XI, we really don’t have stats to measure overall competence.

In fact – and here’s the punchline – it’s been shown that individual defensive stats for centre-halves, even when adjusted for possession,  have in fact a very low correlation with defensive success, measured by shots conceded. So we really can’t use them to measure performance.

But, as we’ve suggested, they can be used to measure style. We’ve already seen the difference between Koscielny and Mertesacker in interceptions. Let’s compare the two for a range of numbers per 90 minutes for the year 2013-14, when both were relatively healthy and in their prime:

Koscielny
Aerial duels: 3.2
Aerial success %: 53.5
Tackles: 1.8
Tackle success %: 83.1
Interceptions: 2.9
Clearances: 7.7
Blocks: 0.8
Fouls: 1.0

Mertesacker
Aerial duels: 4.1
Aerial success %: 68.3
Tackles: 1.2
Tackle success %: 74.5
Interceptions: 1.7
Clearances: 7.2
Blocks: 0.5
Fouls: 0.3

It’s clear that Koscielny is the more aggressive defender, with notably higher tackles and interceptions, more fouls, and the better tackling percentage. Koscielny also seems more willing to throw himself in front of a shot. Mertesacker is more passive, better in the air, significantly more precise, and asked to make more aerial challenges.

Now let’s move to this season’s contrast between the Sunderland pairing of Lamine Koné and Papy Djilobodji.

Koné
Aerial duels: 4.2
Aerial success %: 65.7
Tackles: 0.9
Tackle success %: 75
Interceptions: 1.8
Clearances: 7.4
Blocks: 0.5
Fouls: 0.5

Djilobodji
Aerial duels: 5.0
Aerial success %: 68.7
Tackles: 1.4
Tackle success %: 65.6
Interceptions: 3.5
Clearances: 9.5
Blocks: 1.3
Fouls: 0.5

Remarkably, Djilobodji leads in almost every category. He’s clearly the more aggressive, with a solid lead in tackles (although Koné is the better tackler), and a considerable lead in interceptions and blocks. Despite his aggression, he’s significantly more precise than Koné. He’s also more likely to go for the aerial challenge, and is a little better in the air. He has more clearances, too, but that may be more style than competence. Koné more often passes short, 76.0% of the time vs. Djilobodji’s 65.8%, which may indicate Koné is more comfortable with the ball at his feet, and is thus likely to pass out of the back more. Again, something to watch.

Either way, remember that Djilobodji’s lead in almost every category doesn’t mean he’s overall a better defender than Koné. He’s had some good games lately, but Burnley a few weeks ago was nightmarish. The numbers just won’t tell you how defenders mark, or how often they get beaten. Still, the good news is that using stats to measure central defensive style has reached a high level of sophistication. We can only hope someday to do the same with performance.

Next time: Passing.

Peter Goldstein