Defensive Improvement Explained ….or is it?

As the title states I set out to use a statistical analysis of the last 18 months to really prove why Arsenal seem to be playing better defense over the last 9.  I put some time into gathering statistics thinking there was a pot of gold at the end of this dizzying rainbow of numbers.

What explains our improvement?  How do we prove any of this?  Are we allowing fewer chances/shots?  Are we keeping the opposition out of their attacking 3rd?  Are we tackling and intercepting better?  I’ll look into these things and more, using data from all Arsenal Premier League matches from the start of last season thru the end of 2013 (stat sources).

Blue data points are game by game numbers, the orange line is the trend line.

Goals Allowed

Lets cut right to it and start with Goals Allowed – the number all defenses are measured by.  Take a look, its not that impressive at first – a rather flat line, no real improvement:


But I cannot get over that damned Man City anomaly.  A piss poor day that was.  Sloppy putrid shit that it was.  Had we not passed the ball to them in front of our back four so many times, what may this look like?  Lets forget the day ever happened – say we had such a fantastic post-match bender we simply cannot recall the day…


Ahhhh, thats better.  Amazing what a day makes.

What sticks out is the few times we’ve given up more than one goal recently.  Compare that to last season where you see many games where we gave up two or more.  Since mid-March we have only given up more than one goal twice – and lost both games.  That is a telling statistic.


OK, great, but why – whats next?  How do we explain?  Are we giving up a bit more possession, allowing teams to come to us while we’re in better position to defend?  Maybe.  Our possession trend does suggest a small change in that respect.  Isn’t possession good though?  As long as you don’t get carried away, yes. I’ll touch on that later.possession

Midfield Defensive Contribution

So that must mean our midfielders and forwards must be making a greater defensive contribution (tackles and interceptions).  Keeping pressure off of our back line.  …Or perhaps not, that trend is as flat as my grammar school girlfriend was (trust me, these years later, I’d have her back). So this isn’t it.midfield-contrib

Attacking 3rd Allowed

Are we giving them more possession, but less dangerous possession?  Is the percentage of their passes in their attacking 3rd going down?  Again, preventing pressure on our back line.  …Ugh, no – another flat-line. This isn’t it either.attacking-3rd

Chances and Shots Allowed

Ok, ok – it must come down to Chances and Shots Allowed right?  The fewer of those opportunities the less likely we are to give up a goal.  So lets look at…… oh dear.  Really… they are on an upward trend?  What in the hell is going on here? This really isn’t the reason.


Less possession, no increase in defensive activity, no change in allowing the opposition in our defensive 3rd, and an actual slight increase in chances and shots allowed. How is it we’re giving up fewer goals?  All of these things suggest we’d be giving up more. I was pulling my hair out!  And I don’t have a lot to spare!

Then I remembered.  There are three kinds of lies.  Lies, damned lies, …and statistics.  Chuck em all in the fucking bin.

But I’m a believer in data.  I believe there’s gold in them thar hills.  And I recalled one of the articles I read when I started this merry goose chase.  So I went to reference it and saw he’d even updated it.

Here is @colinttrainor and his article on Arsenal shot statistics – Arsenal Shooting Report.  What he does is much more advanced than what I’ve done here, albeit just for shots.  He gathers a lot more information on shots – x-y coordinates, was it with a foot or a head, etc.  He has an algorithm that evaluates shots – because we know not all shots are created equal.  I found his earlier Arsenal article like this when I originally got this hair-brained idea and was doing a bit of research.

I thought it was great information – that we’re giving up less dangerous shots (prime positions) than most of the league.  But I also wondered two things – is this a change from last season?  And if so, what led to the change?  I have tried and failed to explain this.  Or explain it using the statistics I gathered anyway.

The only possible explanation is that we may, may, be giving up less dangerous shots than before.  Statistics may tell us that, but they are not telling us how or why.  In the end we’re left to our own devices.  So to make this less than a total loss, what do I think it is? Two things…

Fewer mistakes

Maybe that little less possession keeps our defenders from wandering up the pitch too far too often.  When we do you get a result like the home leg against Dortmund – both defenders get caught up the pitch and we got punished.  Also, Koscielny being a bit more positionally disciplined than Vermaelen may have an affect as well.  The Verminator had a tendency to go all David Luiz on us.

Fewer possession mistakes in front of our back line as well, perhaps.  Turnover statistics were flat as well, but I’m suggesting much like shots – where it happens is key.  We saw what a lapse of concentration in that respect can do to us against Man $ity.  Thankfully it hasn’t happened that often.


Is Flamini’s presence, attitude, and general chattiness a difference?  Those are three characteristics you cannot measure with statistics.  He played the embarrassing City game so he’s not always a difference maker, but is he in general?  Perhaps.  Mertesacker is talking more too.  This can keep players from going on walkabout as well.  Can also keep players from going walkabout before acknowledging the fans – doubt Ozil does that again.  Much love for the BFG!


So I wanted to say “hey, look – there it is, the reason for all this!”. Not conjecture, not opinion – but right there in indisputable fact. But I came up short – you get my conjecture, my opinion. Suffice it to say, this is not the article I’d hoped to write. Statistics may have failed me this time, but I will not give up.  At the very least it does expose some assumptions as incorrect.

Whatever the reason, long may it continue!




* The bulk of the statistics were gather from individual match reports from Four Four Two’s StatsZone.  Statistics are also available on and




Leave a Reply Text

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.