Salutations, stalwart Seattle Sounders supporters! I commend your commitment to this club we love.
Last week we saw some highs and lows; in fact, many of them occurred over the course of Sunday’s loss to Philly. To be honest, last week’s periods of excellence and... let’s say, misadventure, have been kind of a microcosm of the season to date. At our best we look great, but at other times, quite vulnerable.
But is it unpredictable, or are there patterns to our successes and failures in the run of play? Perhaps it would be useful, or at least interesting, to look into that.
Welcome back to Sounder Data, the series where I use data, statistics and visualizations to explore interesting questions about the Sounders and MLS. This week I’m exploring where we tend to score and concede goals.
For an example, consider a goal we conceded against RSL two weeks ago:
Gustav Svensson (#4) misses a pass in the middle of the field at 51:01. Joao Plata (#10) eventually picks up the ball and moves it laterally to Kyle Beckerman (#5) at 51:05. Beckerman kicks it to Jefferson Savarino (#7) two seconds later. After a short dribble, Savarino finds Albert Rusnák (#11) at 51:11, goes on a run and gets the ball back one second later, setting himself up for the goal at 51:13.
A real soccer analyst (not describing myself here) might describe this goal as starting in zone 11, passing through zones 15 and 18, and ending in zone 17 right in front of the goal. Never heard of these zones? I hadn’t either until a little while ago, but don’t worry, it’s just a broad grid overlaid on the field:
(Some of you might have seen these in a different order, where 4 is next to 3 and 6 is next to 1, but I think that’s totally bizarre and refuse to think that way. Anyway, it doesn’t really matter.)
I think we can usefully apply this framework to understand the Sounders’ strengths and vulnerabilities. So, just to see what would happen, I did! I took Opta data like the first figure, determined the zones that each attacking sequence passed through, and aggregated every goal we’ve conceded or scored over the entire season, excluding penalties and corners.
The first result to explore is the zone in which we’ve scored each of our goals. This one’s a no-brainer:
You can see that among our 30 open-play goals, 23 have come from zone 17. This shouldn’t be too surprising to anyone, since it’s obviously easier to score from right in from of the goal. We also occasionally score from each of the adjacent zones, 14, 16 and 18.
The same goes for the goals we’ve conceded:
Of the 28 open-play goals we’ve conceded, 22 have come from zone 17. Interestingly, the goals we’ve conceded elsewhere have all been deep shots (zone 13, 14 or 15), never at a tight angle from zone 16 or 18.
On Savarino’s goal above, the final shot was squarely in zone 17. However, the shooting position is hardly the most important part of the play. To analyze our vulnerabilities, we really need to know the zones where the defense broke down, where the play went, and where the ball was just before the goal. For Savarino’s goal, the build-up play was via zones 15 and 18. To know our strengths, we likewise want to know the ball movement before the goal.
Let’s consider the zone before the final zone for all of the goals. First, for every goal we’ve scored this season (besides penalties and corner kicks):
There are some more interesting patterns here. For one, you can see just how often we score from crosses compared to play through the middle. Zones 16 and 18, the flanks of the goal, are tied with seven build-up plays apiece apiece going through them immediately before a goal. On the other hand, zone 14 is kind of lacking considering its proximity to the goal (some teams, I should note, score almost all of their goals via zone 14). The other notable pattern is on the left wing. Although the numbers are small, an excess of goals have originated from zone 10 compared to zones 11 and 12, or any of the zones behind it.
In fact, zone 10 is also among the hot spots for the second-to-last zone before our goals:
Although the variance starts to get higher when we look two zones before the goal, more of them have passed through zone 10 than any of the other midfield zones. I call your attention to zone 10 because this is the area where Joevin Jones, one of our assist leaders, likes to spend his time (more on that in a minute). Our other assist leader, Nico Lodeiro, does his business in zones 14 and 15; hence the hot spots there. In short, this is a map of our set-up guys.
But let’s now look at the other end of the field. Here’s the zone before each goal we’ve conceded so far this season:
This figure should be alarming. Over a third of our conceded open-play goals have come from our opponents’ zone 18, i.e., our left side. That’s seven more than any other zone. This again is Joevin Jones’ (although more recently Nouhou’s) territory. I’m not saying he’s to blame for the goals per se, but there’s definitely a pattern here to worry about.
Of course, this is a simplification of the complexities of game dynamics, and has a number of limitations. For one, I’m not taking into account who was on the field when these goals were scored; just where on the field they were scored from. However, we seem to have a generally consistent strategy this year regardless of who’s on the field, so these patterns could be agnostic to lineup and still represent our vulnerabilities. Another issue with looking at strengths and vulnerabilities this way is that the center of the field is bound to originate fewer goals. That’s almost inevitable since there’s simply more traffic to move through. So some of our key strengths could reside there (like, for example, Christian Roldan), without showing up on the graphs. Third, game state (winning, tied or losing) and pattern of play (counterattack, build-up, etc.) matter a lot and don’t factor into this analysis. And finally, this is definitely a small sample size thing. Statistical significance tests fail to turn up anything, with one major notable exception: goals conceded from the left corner.
Mainly though, the simplified nature of the above is a good basis for further analysis. How, for example, have we not scored a goal this year that originated from zone 12 (as the second-to-last zone)? Is there a consistent pattern to the buildup of goals that we concede from our opponent’s zone 18?
One way to get a little deeper insight is to look at entire sequences of goals we’ve conceded, not just one step removed, two steps removed, etc. Not surprisingly, the most common sequence is 15-to-18-to-17 (like the Savarino example above), with four goals. In fact, every other goal-scoring sequence we’ve conceded has happened only once, with the exception of 14-18-17, which has happened to us twice. Again, what this is saying is that teams have had the most success - actually their only repeatable success - playing down our left side. I consider that to be meaningful, albeit limited, evidence of a vulnerability.
Perhaps my next analysis should include expected goals. Maybe a better way to assess strengths and vulnerabilities altogether is to focus on turnovers (or passing/dribbling success rates) rather than goals at all, using heatmaps like the ones I’ve written about before. I’ll do some digging to find out.
But even with those limitations, I think it can be useful to look into these types of figures. If there’s one figure in this article that displays a strength or vulnerability, it’s that last one. Our chief vulnerability is on the left. What I argue above is that our strength, or at least one of our strengths, is the same side. In other words, in this up-and-down season we’re having, we live and die by Joevin Jones.