I decided to look at how Arsenal's possession % related to results in the 2013-2014 season. Why did I do this? Because. Because, I am cool. And because charts.
To do this, I went through each match at whoscored and recorded the possession % and the result. I decided to create different buckets of poss %, and see how Arsenal performed in those matches. I divided the buckets into arbitrary 5% segments, so there is a 40-45% bucket, a 45-50% bucket, and so on. All games above 60% had their own bucket (this was the largest segment), and same for all below 40%, which was actually just two matches, the two against Bayern Munich, in which Arsenal's collective poss. % was 27%. You can screw off, Arjen.
I included all Premier League matches, FA Cup matches, and Champions League matches including the qualifiers against Fenerbahce. I did not include the League Cup matches because that can also screw off.
Within each bucket, I calculated two things:
1) the percentage of total possible points won, so, if in two matches, Arsenal won and drew, the Gunners would collect 4 points out of 6, for a % of 66%. (I counted the FA Cup semifinal against Wigan as a draw and the final as a win).
2) the goal difference per game, so if Arsenal, in those two matches, won 2-0 and drew 1-1, their goal difference/game would be 2/2 = 1.00.
So, the charts. The first one shows percentage of points earned:
The red line at roughly 70% (70.37%) indicates Arsenal's season-long % of points attained. For reference, Bayern Munich obtained about 82.5% of possible points.
Perhaps the first thing that sticks out is the high rate of success in matches with 40-45% possession, though on a small sample size. Matches in that grouping include the 1-0 results over PBFC, the fortunate FA Cup victory over Liverpool, and the ugly home draw vs Everton. It's hard to know if the results in this bucket indicate a fruitful counterattacking strategy or mere good luck, or, maybe, Spud incompetence.
Perhaps more surprising is the poor results in the 55-60 bucket. Tellingly, included in that grouping are the 1-5 at Anfield, the 0-3 at Goodison, and the 1-2 loss to Dortmund, all matches in which Arsenal were undone on the counterattack or in quick transitions. Given the above-average performance when Arsenal control over 60% possession, the idea here might be that in matches against similarly talented opposition, Arsenal is vulnerable when they keep possession in the opposition half.
The underperformance in the 45-50 bucket is partly because 4 of the 7 matches in that bucket were against Southampton and Man City, both difficult opponents. Also in that grouping is the 0-6 against Chelsea, which almost fits more in an >50 bucket, due to the capitulation to counterattacks, but Arsenal were down a man and probably conceded defeat in that one.
As for the 60+ bucket, you can see that Arsenal did well on a large sample size. The only loss in that group was the weird 1-3 result to Aston Villa to kick of the Prem season. The only other dropped points in that segment were to Swansea in the league (ugh, that one), when Mourinho played for a 0-0 draw, and to Wigan in the FA Cup. Many of Arsenal's relatively easy league victories fell in this bucket.
Now, to get more clarity about how Arsenal performed in each bucket, take a look at a chart with goal differential per game added:
In some of the buckets, the goal differential per game was less than 0, so to make the graph less awkward, I shifted the gd/g up by 3 for each value. The black line at 3 represents a gd/g of 0. The yellow line at 3.81 represents Arsenal's average gd/g of 0.81, or a goal differential of +44 in 54 matches.
The main thing to notice here is that Arsenal did especially poor in matches in which they controlled only 45-50% possession, particularly in the cumulative 3-9 battering at Man City and Chelsea. There were ugly losses in the 55-60 grouping and unfortunate results like the 1-0 loss to Stoke, but Arsenal also had some strong performances in such matches, one example being the 3-0 win at Hull late in the league season.
Also of note is that Arsenal may have gotten results in the 40-45 bucket, but they were all narrow wins, not the authoritative, dominant variety as seen in the 60+ group. This should not be surprising.
The next step is to compare possession stats home and away, as different strategies may apply when home and away. Is it better to control less possession on the road? Which possession mix is best for Arsenal at the Emirates?
First, a look at how many matches of each possession segment are played home and away:
As you can see, the majority of the 60+% matches are played at home. You might also notice more away matches in the 55-60 segment. This is partly because Arsenal's poss% downshifts by about 5% away to lower-table fodder, but interestingly, this season, Arsenal controlled more possession away versus several top-half sides, including Liverpool, Everton, Stoke, and United. Arsenal won 0 points in those four matches. The prevalence of away matches in the 50-55 bucket is also due to a downshift, where Arsenal controlled more possession in the equivalent home fixture, a pattern that held for all three CL group away matches.
So, how did Arsenal do home and away?
What I first notice is the perfect away record in the 40-45 and 60+ groups. The two away wins in the 40-45 group were at Tottenham and Swansea, both matches in which Arsenal was able to soak up pressure and score their chances when presented.
As you would figure, the perfect record in 60+% away matches is due to Arsenal playing lesser opposition and controlling those matches. For all the poor performance against better squads, Arsenal took care of business in 2013-2014 against the likes of Crystal Palace and Sunderland.
Also, notably, the poor performance in the 45-50 and 55-60 buckets is down to away form. Think Anfield, Goodison, Etihad, and Stamford Bridge. But you already knew that. Bringing down the home number in the 55-60 group is the loss to Dortmund.
Those two matches Arsenal won at home with 50-55% possession? The FA Cup victory over Everton and the league victory over Liverpool. Notably, in the memorably bad losses to both, Arsenal controlled more possession. It's difficult to know what to make of that. Did Arsenal have more possession in the losses because they had to chase goals? Or was Arsenal tactically and strategically naive? There is also the matter that a 3-5% difference in possession isn't huge, and could be down to random variation.
This next chart I made shows home and away gd/g. I find it a little noisy but it should reinforce some of what is discussed above:
Also, I made a chart for Arsenal's performance against the top 8 English teams in the League and FA Cup, and All CL matches except for against Marseille. Everything of note in that chart is pretty much covered above, so I haven't included it, but if anybody who reads this is interested, make note in the comments.
It's hard to conclude much from the charts provided, partly down to sample size, partly down to the arbitrary size of the buckets, and partly because soccer stats are hard and it is difficult to know what to make of possession% - for it to be truly meaningful, more granular metrics are needed. I think there a few interesting nuggets here, however.