Everton’s brilliant start to the 2016-17 campaign hit a slight hiccup midweek, as a fairly strong team fell 2-0 to Norwich in the EFL Cup.
The goods news is that in this weekend’s match against Bournemouth, the Toffee’s have an opportunity to bounce back straight away. Eddie Howe’s side sit 15 places below Everton in the league table and are off to a slow start this season, with a draw against Crystal Palace and rather fortuitous win against West Brom the source of their only points thus far.
The Cherries have won plaudits over the last few years for their brand of attacking football. In 2014-15 they won the Championship with the most goals, the most shots, the most possession, and the best pass accuracy in the league. Unlike many promoted sides who resort to more defensive tactics or borderline anti-football to survive in the Premier League, Howe mostly stuck to his guns last season and finished the season in the top 10 in both possession and pass accuracy, which is pretty impressive for a club with their budget.
Perhaps even more impressive though were some of the defensive numbers they posted in their maiden season. Their shot suppression was outstanding, as only Man City, Liverpool, Man United, and Spurs allowed fewer. They ended up with a dismal record against the top four—losing six and drawing two—but positive football and a seemingly solid defense meant that they were comfortably out of the relegation zone with time to spare.
This season however things haven’t gone so well. Michael Caley’s model has them at league worst in expected goals allowed and second worst behind only Burnley in expected goals for. Adjusting for schedule, they’ve been dead worst in the league in chance differential:
While it seems like a fairly dramatic drop-off from how things were going last year, there are some underlying issues with Bournemouth coming to light now that were somewhat masked over the course of last season.
Bournemouth have a shot quality problem on both sides of the ball. I mentioned they were 5th in preventing shots last year; they were also 11th in expected goals against and 18th(!) in plain old goals against. There are two things happening here: first, the Cherries had an absolutely miserable save percentage, at around 57%. Save percentage is a number that’s generally shown to be not repeatable year after year (and thus subject to regression to the mean), but Mohamed Mohamed fired out a warning last May:
...similar to Aston Villa, when this trend repeats itself for 3-4 years, the noisy nature of save percentage can start sussing out. Case in point, every season that Artur Boruc has been the starting goalie his team has had a very low save%, and he was on Southampton who were good-very good defensively. Paul Riley’s expected goal model has rated Boruc poorly ever since his debut Premier League debut four years ago, and his shot stopping rating of 0.78 this season is 2nd worst for keepers that have faced at least 75 shots on target. If you want to see the benefits of high shot stopping rates look at West Ham this season or Swansea last season, teams that have pumped up 4-5 spots up on their true talent level cause of favorable variance. I know it’s not exactly the most strident analytical take, but anyone who plays some form of organized football must know the feeling when you need just one save to get you out of trouble and the keeper can’t get it. That’s the type of hell Bournemouth have been stuck in this season...
Things in that department have improved slightly thus far this season (66.7%), but that brings me to the second thing happening with Bournemouth: the chances that they concede are of high quality. Since expected goals are a shot-based measure of chance quality, we can divide expected goals by shots to give a measure of the average quality of a shot in a given sample. Last year, Bournemouth were dead last in the league in this metric on the defensive side.
Putting it together, Bournemouth didn’t concede very many shots, but the ones they did were in high quality situations, and furthermore they were really bad at keeping them out of the net. This is how a team concedes the 5th fewest shots but 18th most goals in the league.
On the other side of the ball, Howe’s men were 11th in shots but 16th in xG, with a shot quality ranking 18th in the league. Their conversion rate was a much closer to league average than their save percentage so they still scored a decent amount of goals, but the shot quality is still a worrying trend and it begs the question as to whether there might be a concerted (and possibly misguided effort) here to emphasize quantity over quality.
Bournemouth struggles this year
It’s fair to ask at this point what any of this has to do with Saturday’s game against Everton (“Enough with the numbers NERD, I thought this was a match preview” I can hear you saying). The main point is that the problems I highlighted above have carried over for the Cherries this season. They are again dead last in shot quality against, and 16th in shot quality for. The big problem now is that they’re not getting the volume to mask it on either side of the ball, hence the paltry xG numbers I posted above. Right now Eddie Howe’s team are a far cry from the side that dominated the Championship two seasons ago.
Part of their problem is simply talent and budget. If you want to play possession football in the Premier League year after year, you either need to be unbelievably shrewd in the transfer market, have a tremendous academy, or both. Southampton are the model for how to push on sustainably, whereas Swansea are perhaps a cautionary tale in stagnation. Plunking down £15,000,000 for Jordon Ibe this summer was not a particularly encouraging move in this regard for Bournemouth.
A related issue is something I mentioned with Middlesbrough last week—Bournemouth lack creativity in the center of midfield. As such they can be stagnant in possession and over-reliant on wide play. Part of this is a stylistic choice: instead of a true number 10, Howe likes to play a second striker in a 4-4-2/4-4-1-1.
Against Man City last weekend they attempted to deploy Jack Wilshere at the top of a midfield which was somewhat of a disaster:
Sam Gregory has done some great work on a non-shots based xG model called PEP (Pass ExpG Projections, nothing to do with City’s new manager). Part of that was looking at the probabilities of passes into the box becoming shots within 5 seconds. He produced the below chart using 2014-15 Premier League data:
This illustrates the effectiveness of passes from the zone 14 area, or the central area just outside of the box.
Below is a similar pass chart, but from Bournemouth’s game against Man City. This chart shows only forward passes on the ground into the final third. Blue represents Bournemouth, attacking to the left.
Note the huge hole in the exact areas with the highest PEP in the previous chart. This is the result of lacking creativity, vision, and/or technical ability in central attacking areas: very little penetration from the most dangerous areas. Against Crystal Palace, a much worse side than City, there was a similar trend. Again, blue represents Bournemouth, who are attacking the left.
This sort of thing used to be Bournemouth’s strength back in the Championship, but as you’d expect they’ve found themselves up against some stingier defenses in the top division.
So how do they go about their attack? Usually by pushing the ball up the wings and using the movement of either Joshua King, Callum Wilson, or both to open up space.
Against City, Howe knew he wouldn’t be able to play a possession game and as such there was a clear plan to get balls up to King in the channels:
Everton should expect to see something like this on Saturday and the back four will have to be very aware of their spacing so as to let King and Wilson’s pace wreak havoc. Occasionally the two can work in tandem to good effect as well, as in this chance against Palace:
Unfortunately for Bournemouth that was more of an isolated incident, as for the most part opponents have done a good job of shuttering Bournemouth’s buildup to the flanks where it is easier to contain. Everton will want to do the same.
One of the most worrying things about Bournemouth’s defensive numbers is the opponents they’ve played. Most egregiously both Palace and West Brom exceeded 1.5 expected goals against the Cherries, which is not good. In particular, Bournemouth have struggled against crosses in these matches. Courtesy of the FourFourTwo StatsZone, here are the chances in open play that West Ham, Palace, and Albion created against Howe’s side:
Man City, on the other hand, just absolutely picked Bournemouth apart with counterpressing, through balls, and all of the other glorious things that they do.
Overall the signs are worrying from a Bournemouth perspective, as they’ve been undone by multiple different approaches from a few different types of sides, and they haven’t had the offensive firepower to counteract it.
Matchup with Everton
Given that Everton have a bit of a number 10 problem themselves, I’d expect Koeman to do something similar to what he did against Middlesbrough and push Idrissa Gueye into more advanced positions to supplement the attack. The game will also suit Yannick Bolasie and/or Gerard Deulofeu as both players love nothing more than getting down the flanks and putting in crosses. There’s enough evidence already to suggest this season that Bournemouth struggle with this approach, and Romelu Lukaku will be buoyed by the fact that his compatriot Christian Benteke found quite a bit of the ball in and around the box when he faced Bournemouth, evidenced by his touch map (attacking right):
It’s been noted here that Everton’s soft schedule is a serious opportunity to put themselves in a solid early position, and Saturday’s match is another piece of that. I think Koeman will expect to create plenty of chances here and will be fully intending to walk away from Dean Court with all three points.