Chelsea's Free Kick Problem - Statistics And Analysis

One of the most common themes that has popped up over Chelsea's last three Premier League matches has been the Blues' leaking of goals via free kicks. In three games, Chelsea have conceded three free kick goals, and that's taken our defence from looking pretty good to merely adequate - eight goals conceded in seven games is only a little under league average, and the club has made its living off of being very hard to break down.

So, what's up with Andre Villas-Boas' set piece defence? Is there an actual problem, or have the last three matches been a fluke? We have some statistics at hand that can start to answer the 'fluke' question, although if there is a problem, any diagnosis will have to come from studying the videos. With Villas-Boas supposedly switching to a zonal marking system, there may be some issues with tracking runs and the like. Let's jump in and see if we can spot anything.

Statistically Speaking

It's a commonly cited fact that Manchester United have faced the most shots on goal out of any Premier League team so far this year. With Norwich City subjecting them to a hilariously impotent onslaught at Old Trafford this weekend, that little tidbit is still true, although Bolton Wanderers are fast closing the gap. What isn't talked about as much is the team that resides on the other end of the spectrum: Chelsea.

That's right. So far this year, according to the statistics at whoscored.com, the Blues have faced just ten shots a match. That's significantly lower than the average of 14.7 and only seriously rivalled by Manchester City's eleven. United and Bolton? They're all the way up at nineteen. That hasn't stopped the Blues from racking up goals conceded, though - three more than each Manchester team (and, bizarrely, Aston Villa, who feature Alan Hutton and Stephen Warnock at fullback) and on more than Fulham, whose primary defensive technique is to bore the other team to death*.

*It worked really really really well against Queens Park Rangers!

So despite giving up far fewer shots than average, the Blues aren't giving up many fewer goals. Why is that? There are a couple of explanations here. The first is simple - it's simply random variation. More shots than expected are going in due to nothing more than bad luck. That's a possibility that's overlooked far too often in sports analysis, but when you're looking at seven-game sample sizes, it's not something to simply ignore.

Of course, relying on luck as a crutch is pretty stupid, because it's lazy and leaves no room for improvement. The second possibility is that Chelsea are systematically allowing easier shots than they ought to, whether that's a function of bad shot-stopping from the goalkeepers (hah!) or better positioning by the opposition on the shots themselves. What's a good way of ending up with high-percentage shots for opposing strikers? Bad marking on set pieces.

The graph below would suggest that Chelsea are giving up an inordinate amount of goals on set pieces. It doesn't say that they're bad on set piece defending, just that they have been worse at it so far than they've been at defending in open play:

Immediately we can see a bit of a statistical problem here - Chelsea have only conceded eight goals, three of them from free kicks. If one of those free kicks doesn't find the net, their set piece goals to goals ratio goes from 38% to 29%, and things look far less dire. Such are the issues with tiny sample sizes (for reference, whoscored has our 09/10 and 10/11 numbers at 46.9%* and 24.2% respectively).

*Ed note: Wib wob?

Even if we take these numbers at face value, they don't really tell the full story. Chelsea haven't played everyone in the league yet - what if they've been facing a disproportionate number of teams that do well from free kicks and corners? I mean, we don't make too much fun of Bolton for losing all the time because they've had to face an impossible schedule to start the season - what if Chelsea have faced the equivalent in terms of an aerial bombardment?

Of course, we have to strip Chelsea's games out of the opposition's numbers in order to stop from recursing, but that's easy enough, and doing that gives you the following set piece goals scored per game for Chelsea's opponents:

Summing them up would give you the average number of set piece goals that the average might expect to concede against said opponents over a seven game span. Which is three, the same number that the Blues have conceded. Oh.

Is that a far comparison? Yes and no. It's not particularly granular - what if Chelsea given up fewer scoring set plays than average? How would we account for defence of specific plays rather than looking at entire games? One possible solution is to look at how many shots a team concedes from set piece situations and how many goals result:

This is another one of those things that's hard to figure out. Not shown on that chart, because I'm stupid and too lazy to change it now, is that Chelsea are on the very low end of shots given up from these situations - across the Premier League shots against from set pieces ranges from 13 to 28, and Chelsea are at 15. That leads to an impossible set of questions - are Chelsea giving up easier shots, or getting unlucky with the shots that do get taken? Is the job of a defence to restrict the shots that result from set pieces or to ensure that they're low percentage ones? How can we track goals per 'scorable' set piece anyway?

The short answer is that we can't. Better statistics would help, of course, but we've reached the end of our rope without being able to determine whether this set piece issue is significant or not. To the tapes!

Video Analysis

A word of warning before we continue - just as statistics are dangerous ground if you don't know what you're doing, highlight tapes are very easy to get tripped up on. Nobody saves videos of the free kicks or corners that were easily cleared, so the three goals depicted below are obviously subject to a massive selection bias. You knew all that, of course...

Here's Chris Smalling's goal for Manchester United at Old Trafford. It's not really a pretty sight.

What went wrong here? It's all well and good to say that Chelsea stepped up to get United offside, which is the defence Andre Villas-Boas used, but a) the offside trap on free kicks is a ludicrously risky strategy that you'd have to be mental to use in the first few minutes of a match at Old Trafford and b) from repeated viewing it seems at though Smalling and Wayne Rooney simply make the mistake of their own accord rather than Chelsea doing much to goad them into being offside.

Anyway, the offside excuse is an extremely dangerous one to use, because as we all know offsides aren't always flagged correctly, and so even if your team is perfect at catching the opposition off, you're going to concede shots and goals without the linesman waving his flag. That's the whole human element thing, and while that can be annoying, I sort of like the disincentive to relying on the offside trap for all of your defensive needs.

The fact of the matter is this: Chelsea blew their coverage here. There were three moving red shirts (Smalling, Phil Jones and Johnny Evans) with Nani holding back, and there were four Chelsea players able to track them. Only John Terry did, which meant Fernando Torres, Frank Lampard and Jose Bosingwa essentially stood around looking lost. You could say (again) that they were playing for offsides, but their colleagues at the near post tracked Rooney and Chicharito just fine, and you play to the whistle anyway.

I don't think the offside excuse holds water, to be frank. This is just a matter of three players failing to do their jobs, which probably means a failure on Andre Villas-Boas' part - that many players screwing up at once almost has to be tactical.

Better news in terms of tactics for the next goal, which came in the 85th minute at home against Swansea, by which point Chelsea were already up by three goals. It was hardly a high pressure situation, even though the Blues were down to ten men following Fernando Torres' amusing yellow card.

Here we see zonal marking working its magic. Players are tracked correctly, decoy runs to non-threatening areas are ignored, spare defenders are preserved to clear knockdowns. It's all perfect.

And then Ashley Williams scores anyway. Whoops.

Fortunately, this is the sort of error that isn't indicative of a flawed system. Unfortunately, it's the sort of error that's completely indicative of a flawed player.

Williams' goal was entirely due to the failure of right back Jose Bosingwa to stay goalside of the defender, and even if he had it's doubtful whether Bosingwa would have been effectively able to challenge the Welshman in the air. We've seen the Portuguese have issues with left-right diagonals in open play throughout the season - he just doesn't seem to track them right - and here was that weakness displayed again, this time in set piece form. This will be a problem until Bosingwa is either moved away from marking responsibilities in free kick defence (especially for a zone as important as the far post) or replaced by Branislav Ivanovic, who is far better in the air.

Last but not least, we have the goal scored by Bolton's Dedryck Boyata shortly after halftime in Chelsea's 5-1 win at the Reebok. It's not good news here for Villas-Boas' system either.

First of all... this line is not good. From Petr Cech's right to left, we've got Raul Meireles, Frank Lampard, Ramires, David Luiz, Kevin Davies, two Bolton players sandwiching John Terry, Zat Knight, Didier Drogba, Ashley Cole, and Gary Cahill. Boyata is kind of chilling out near the edge of the area.

Obviously, you can point to the collision between Drogba and Terry as a reason Boyata was able to blow through the gap in the Chelsea lines to head past Petr Cech (a saveable one, in retrospect), but to do so is to ignore the fact that Chelsea were outnumbered six to three at the far post, which is lunacy. And one of the three was Ashley Cole. Relying on him to clear your set pieces is sort of like asking Natalie Portman about molecular biochemistry - no matter how awesome other aspects of Ms. Portman may be, she's not going to be your go-to resource when you want to explain RNA polymerase to the layperson.

So, yeah, this was pretty embarrassing. Nobody ever picks up Boyata, and even if anyone had even noticed his wiggly little run into the penalty area everyone was blocked off anyway. Perhaps Chelsea might try putting their defenders where the opposition is next time? You can't even explain players relaxing thanks to being 4-0 up and home free for this one, because everyone was completely out of position from the beginning.

It's not all back, though. Raul Meireles picked up a man running into his zone perfectly without the rest of the line being disrupted. Zonal marking, ladies and gents!

Summary

While we didn't find anything obvious in the statistics that might reflect a new-found Chelsea weakness in terms of free kicks, the analysis of the goals the Blues have conceded hardly paint a pretty picture. If one wanted to go after Chelsea on a set piece, the far post seems to be routinely under-defended in the free kicks shown above, and should be targetted with extreme prejudice until they've stopped doing things like leaving Bosingwa one-on-one with a centre back or letting six players (no, I'm not going to get over that) vie against three blue shirts.

Now, that doesn't mean Chelsea's defending set pieces is bad - as I mentioned, there's a big selection bias in just watching the goals we've conceded - but it certainly has issues, and the diagrams highlighted above share a common thread in terms of that weak area at the far post. It's unclear whether the issue is Villas-Boas' zonal marking system or not, but watching United, Swansea and Bolton all hurt Chelsea on free kicks means that the manager's going to have to look hard at what's going on here.

My suggestion, for what it's worth, is to be less rigid with the zonal system at the start of the play. This should prevent the opposition from overloading one side while still keeping the beneficial aspects of zonal marking in place - the only tradeoffs would be in how effective low-leverage zones are covered, and I think I can live with our defending being more lax in unimportant areas in exchange for covering the far post properly.