My statistical prediction model of EPL results did something a little unexpected this week. With some caveats around van Persie and whether or not he was fit, its odds gave a high likelihood of Newcastle beating Manchester United at Old Trafford.
We all know how the game turned out and that Manchester United – by their usual standards – have had a very poor start to the season. For this post I want to see if we can find some pointers as to what’s going on. There must be systematic issues, rather than just random chance, or a statistical model wouldn’t have seen that loss coming, so can we find some of them? In particular, I want to look at player-by-player passing rates and ball retention because to my eyes on Saturday, Manchester United looked to be giving the ball away far more than they usually would. Passing also underpins my prediction model and a big drop in pass completion and shot rates will see a big drop in expected win percentages.
Here are the season-on-season pass completion rates, shown as changes vs. 2012/13, for those personnel who have played in both 2012/13 and 2013/14. Differences are expressed as percentage points, so a drop from 80% accuracy to 70% accuracy, would be -10%.
Thickness of the bars is number of passes per minute on the pitch. With the exception of Jones, Hernandez and Buttner (who has only played 90 minutes this season), nobody has improved significantly and most have seen a fall in accuracy. Rafael, at the bottom of the table has seen the largest drop, but this is very possibly through injury and having only played in four games. During those four games though, he was giving the ball away more often than once every six minutes; fifteen times per game.
If your passing accuracy drops, the balance of who receives the ball on your team is likely to change and we can see that this has happened. If we look at changes since 2012/13 in the number of touches per minute on the pitch, Manchester United’s strikers plummet to the bottom of the table. The ball just isn’t reaching those players with the frequency that it was last season.
Robin van Persie and Javier Hernandez are isolated and each seeing the ball around 14% less than they did last season, controlling for the amount of time that they’ve been on the pitch.
Now let’s take time on the pitch out of the picture. Do Manchester United’s players still shoot when they do have the ball? Here’s shots per touch, again shown as changes since last season.
Anderson, Evra and Smalling have been shooting more often (from range? We don’t have that data) when they get the ball. Robin van Persie has seen a small increase of 4.1% too, so when he does get the ball, he’s still getting a shot away. Unfortunately, we can’t analyse whether those shots are from as good areas as last season using EPL Index’s data, but van Persie is still shooting when he sees the ball.
Shot frequency for everybody else has fallen and for most, by 25% or more. The defenders at the bottom of the table have fewer shots overall and might be less of a worry than the midfielders and strikers, but those defender shooting stats are likely to be a symptom of less commitment of the team forwards. Defenders are no longer getting into positions to take the odd pot-shot.
So far, there probably aren’t too many surprises here, apart from possibly the magnitude of some of the falls in performance. We can add a final layer to this analysis though and look at the combination of who sees the ball and who gives it away. Where are the weak links in the chain? If you’re seeing the ball a lot and misplacing passes a lot, you’re a much more serious issue to your team’s build up, than if you’re a striker who sees the ball infrequently and mostly shoots rather than passes.
Looking at possession and passing as a grid, here are Manchester United in 2012/13. The vertical axis is touches per minute and horizontal is passing accuracy. The top right hand corner is where you want your midfielders (in particular) to be; seeing a lot of the ball – expressed as many touches per minute – and passing it accurately. In the bottom left, you’re not seeing the ball so often, but when you do get it your passes often go astray. In the bottom right, you’re also less involved in play but passing accurately.
Top-left is where you really don’t want to be. Here, players see the ball often and they pass inaccurately, so they lose the ball a lot. These are the players who will frequently cause a move to break down; the ones who concede a lot of possession.
The line positions on this chart are slightly arbitrary, but effectively split Manchester United’s players into quadrants. We’re interested in how they perform relative to each other and to last season, so this will work for illustrative purposes.
In 2012, the majority of players were clustered in the ‘green’ zone; involved in play (plenty of touches), and retaining the ball with a high pass success rate. Only the strikers (and Nani) appeared well outside of this zone.
Now look at what’s happened this season…
Four players – Rooney, Smalling, Nani and Rafael – jump into the ‘involved and wasteful’ top-left zone, which was empty last season. Rooney’s been receiving rave reviews, but giving the ball away more frequently. Rafael’s and Nani’s changes are dramatic.
Giggs does also now appear in this quadrant but it’s close and only a relatively small change in pass completion rate has got him there.
For me, these charts illustrate some of the numbers behind United’s performance drop. Passing has broken down more often and so shooting frequency has dramatically dropped. But is there anything here that suggests what could be done about it?
An analysis like this can’t find answers on its own, particularly with the relatively aggregate data that we have, but it can definitely point the way to finding some solutions. These numbers raise a few key questions, that could lead towards finding an answer to such a dramatic swing in performance. Pick off the big issues one at a time and you might start to fix the problem. I’ll just pose these and leave them here, to hopefully provoke a few thoughts…
Rafael has had injury problems, which probably explains his passing figures and he’s also only played four games in the EPL this season, so he can’t have been responsible for a lot of the team’s under-performance.
Nani though, looks like a problem. He’s playing more minutes than last season, seeing the ball a lot more when he’s on the pitch and he’s more likely to give it away when he receives it. That’s not good. To an extent, Smalling’s figures show the same movement – more involved in play and with weaker passing than last season.
Rooney may be looking good but… He’s giving the ball away more often and shooting a little less frequently. I’m wondering if Fergie’s tactical discipline has broken down somewhat here and we’re seeing Rooney do more all over the pitch, but less effectively, for the team as a whole.
Robin van Persie and Javier Hernandez are seeing the ball much less frequently and it would be lovely to have the more detailed data that would let us work out why. Their supply line must have been cut, but beyond pointing towards players whose performances have dropped versus last season, we can’t see exactly where the chain has broken.
Are all of these obvious problems? Maybe. But I feel strongly that analysis like this can help to focus the mind on solutions. When you’re trying to solve a jigsaw, you don’t attack it all at once, you look for the corners. In Manchester United’s current performances, a few players do stick out as potentially key pieces of the puzzle.