Dortmund Defensive Analysis

Over the course of the season, one of the biggest points against Dortmund that people have made is their inability to keep a clean sheet. And, well, that’s totally justifiable. Over the course of the season, Dortmund have only managed to keep a total of five clean sheets with two of them coming against non-Bundesliga sides in the DFB Pokal. The main question that I’ll be looking at today is this: “Are they actually that bad?”

Methodology

  • Focused purely on goals conceded, rather than big chances. This absolutely loses some important data points, but for the sake of time, we’re sticking to the goals.
  • Creation of a scoring system that ranks goals based on four categories: Defensive error, tactical failure, shot quality, and luck factor.
  • Scored each goal on a scale of 0–1, using decimals to denote a partial truth, such as a tactical breakdown that is influenced partially by player failure.
  • Divide each category’s score by the total amount of possible points (which was equal to all goals conceded) to find out which percentage of goals fall under each category.
Spreadsheet used can be found here, if you’re curious!

When creating this concept, I ran into a few problems that I think are important to mention. The first is obvious; football isn’t this binary. I wholeheartedly agree, but I think this is still a useful exercise for one main reason. By doing this, we have the opportunity to look at the best ways to remove balls from the back of the net, while not needing to make overall structural changes, or we can see where large structural problems are systemically failing.

Boy, that’s a bit of a packed thesis. Let’s clarify: when a team is struggling on defense, the coach’s job is to figure out how to stop that. To do this, a coach needs a clear picture of the ways that goals are being conceded, without the muddy water that can be found in a wholistic approach of a full 90 minute match. When I considered the tactical aspect, I only focused on larger issues rather than small changes like a pressing trigger within each match. For example, two of Bayern’s goals were scored because of how narrow BVB was. This is an intentional choice made by the coaching staff, and it poses it’s own set of questions, like is this risk worth it? Including something like luck allows us to examine whether or not goals are being scored purely because of sloppy scenarios, similar to the defensive mistakes category. The shot quality category makes for a high number because it simply considers whether or not an opposition is taking a high quality shot. Goals tend to come from the best shooting situations, so this number will naturally be higher than the rest. This is where the 0–1 score comes into play.

  • Was there a mistake?
  • Was there a tactical problem found?
  • Was it a high quality chance?
  • Were there rebounds or other abnormal deflections?

After grading on these scales, I created a total “problem score” for each goal. This essentially gives me a score out of four which considers how ‘problematic’ each goal should be for a coaching staff. A perfect score would include four problems:

  • A mistake made by a player
  • A tactical problem
  • A high quality chance
  • A high quality run of play

Conversely, would be the lowest possible score. These goals are essentially unavoidable, and one of the few times that you can write off a goal as a ‘fluke’ or purely random. They would include the opposite four. Zero mistakes, no tactical issues, and deflections/bounces on the shot and run of play.

The quality of shot box is a bit of an anomaly in my scoring here because it is more true than not, so I will be looking at that separately. The three other categories I will be calling COGSO, which stands for ‘causes of goal scoring opportunities’ as they should explain every potential goal scored. (If you can think of some possibility I’m missing, let me know please, I’m curious.)

Two other weird anomalies in my data set: three goals scored by Ajax after a BVB red card. I’m including them, but it’s a bit harsh. The second is set pieces. Dortmund have conceded three of them, two corners and one free kick, and I simply do not know enough to grade them. I will be excluding those entirely. Goals scored directly off a set piece itself (Grifo FK and pens) are included.

So, given this formula, let’s apply it to BVB’s goals conceded and see how problematic their defense actually is.

Defensive Errors:

Hands down one of the most frustrating part of watching Dortmund this season has been errors made in possession that caused goals. This has come in a variety of ways, penalties given up, own goals, and passing mistakes galore. At least four penalties have been conceded this year, and at least two of them have caused BVB to drop points. (Bochum and Bayern are those two matches, and I’m saying at least because I feel like I’m missing one.)

How do they land in this category? Well, unsurprisingly, it is the worst of the COGSO opportunities. Our post-set piece score here is a whopping 18.25/33 potential points. This is good for 55.3% of all goals scored by opposition, and is pretty damning.

This category is the main driving force behind me doing this project, because I was curious. Is any of the defensive stability actually Rose’s fault, or is it player execution? In a shocking turn of events, it’s not. When you look at your goals and over half are directly related to silly mistakes, it becomes hard to consider tactical solutions.

Here’s a great example. In the clip below, Hummels has the opportunity to play a ball wide and open his hips, but he keeps his head down and drives it straight back where he received from. This body shape gives Muller free reign to pressure and he wins the ball with comedic ease.

Tactical Failure:

This is hands-down the most difficult category to grade because it relies on a prior knowledge of Dortmund’s matches, and their larger goals in mind when in and out of possession. Fortunately, there’s an awesome Twitter account (@jwzfootball) who wrote over 20 pages on the topic, which I used as the groundwork for this section. If you’re interested, check out that article here.

Back to COGSO, we’re going to focus on larger goals in mind in this section. One of the things that I discussed a lot is Dortmund intentionality with staying in compact spaces, and attempting to force an opposition to play in a tighter space of the field. After winning possession within these pressing traps, the midfield will play more expansive and find those transition moments where Haaland(and others) is(are) able to thrive. Given that this is largely the goal in mind with their game model, I focused purely on those ideas, rather than anything more granular. If you’re doing analysis for a club it is important to include this, but I will not be doing it because I don’t work for BVB. I don’t know every single thing they’re looking for and would prefer accuracy.

SO. How did they score here? With a score of 11.5/33, just under 35% of goals came off of some sort of tactical decision within the game model. Honestly, I think this is pretty good. For the most part, the ways that these goals were scored was due to some sort of expansive play, or the opposition escaping the press, which is an intentional, attacking approach taken in Rose’s game model. Goals will be conceded like this, and it is simply a part of counter-pressing and general pressing.

Here’s a couple examples, two different scenarios. First we return to Bayern, where a long switch to Davies unlocks a vast amount of space. I have two problems here: and the first is ball pressure. I’m not happy with the fact that a ball like this was played so easily. The second problem is the fact that Raphael Guerreiro is 5'7. Grow a couple inches dude.

Next example is here against Wolfsburg, where another instance of expansive play results in a high quality chance. Similar issue, complete lack of ball pressure results in a diagonal, but also: so many people on one half of the pitch. It is absolutely vital as a team that you secure your narrow positioning by either being in a low enough position where shuttling and defending the box is possible, or you are putting enough pressure on the ball and angling your press so that these balls are impossible.

Another fun example is against Ajax, who had BVB stretched before even playing the switch of play. Their style of play is essentially anti-Roseball, and it is entirely unsurprising that an expansive pressing team demolished a narrow one. Here’s your example of how: expansive passing creates the opportunity for overloads before supporting defenders can arrive.

One final tactical category that I want to touch on here: counters. Due to how many numbers Dortmund commit forward, there are times where the defensive line is left exposed. This happened vs. Freiburg for their second goal, and while it was somewhat a mistake, they were in that position due to a tactical decision. This is why all three categories are important.

Luck Factor:

The final section of the COGSO ratings is what I have called the ‘luck factor’ and it looks at things like weird rebounds, bounces, or anything like that. I had a tougher time with the math here because I don’t believe that these things should count against you. So to do this, I graded all “lucky” goals as a zero. This means the goal does NOT count against them, whereas good play from the opposition will, counting as a one or somewhere between.

For example, the Coman chance above was considered a zero in my scoring because he would not have even received the ball without a weird bounce off a clearance. I’ll include another example of a zero here. Weird shot that lands at Diaby’s feet again, then he manages to needle it through roughly half of the population of the North Rhine-Westphalia. Shit happens sometimes.

This leaves you with the 1’s and 0.5’s, which are essentially “good” goals, for lack of a better way to phrase it. They involve minimal luck, and thus raise the ‘problem score’ higher. Chances like these are the one’s we would like to avoid seeing at all. We scored the same as the tactical section here, finishing with a 34.84% otherwise known as 11.5/33. This isn’t awful, it essentially means that 1/3 of goals can be chalked up to dumb luck or high quality finishing.

Shot Quality:

The final piece of our puzzle today is grading each chance on shot quality, particularly based on how difficult the shot is. To put it simply, a zero would be a screamer, while a one would equate to a shot from inside the box. There weren’t many that didn’t score full points, but this goal from Pedro Porro was scored by me a zero.

In this category, 81.81% (27/33) of all goals scored some amount of points. This doesn’t really alarm me much, as mentioned earlier, this category is drastically higher because most goals come from good shooting opportunities. This category is used to consider ways that we can improve the side, because chances that scored above a zero are important to remove. At least seven of these chances also coincided with some sort of defensive error, while another 3–4 were caused by partial errors or weren’t fantastic shots. Those scored in the decimals. This is important information for the next section!

So… what do we do about it?

This is the part I’m the most excited for: how can we take the data collected and the observations made, and turn them into some sort of actionable improvements within the side?

Let’s start by looking at defensive errors, because I think these are the most avoidable chances. 26 defensive errors were committed out of the 33 goals we’ve looked at, and 24 of them resulted in some better-than-bad opportunity to shoot. While largely devoid of larger context on each individual scenario, we can make a conclusion here. This is unacceptable.

We then have to consider, from a game model perspective, what is causing a) so many errors and b) whether or not those errors are worth building around. Our first question has a couple of answers, because of course not all 26 errors fall under one category. My primary focus here is that Dortmund players lack an escape rope when put under opposition pressure. Losing Haaland for a fairly large number of games in the past couple months means you have lost the target man. This is the most obvious solution to pressure, lump the ball forward and get rid of it.

My second solution is slightly more complex, and it focuses on Dortmund’s build up structure of a 2–3–5/2–3–2–3. This shape means the defenders are more isolated when possession is lost/an error is made because the center of the pitch is mostly comprised of only the 6. The other two members of that ‘3’ are the fullbacks who are usually high and wide, rather than tucking in on the weak side. To help fix this, I would make one of two changes, either invert the opposite fullback and play the opposite winger slightly deeper, or I would drop the 6 into the defensive line and create a 3–2–5 similar to City. This would defend the center of the pitch better, but would harm Meunier’s strengths. Probably best to wait until Morey is back and can cut into the end line.

Besides for this: players literally just need to stop making mistakes. Hummels is playing in slow motion, and it’s genuinely just unacceptable. I don’t know what else you can say except that. I wish I had tracked by player because Hummels was the mistake player for at least half of them. He’s having a tragic season :(

Moving past the error creation section, we need to consider our two final sections of COGSO: tactical problems and luck. Both of these sections are difficult for me to even discuss because they are difficult to change. Dortmund’s two main issues tactically have already been discussed, but they are intentionally, and that is ok. Both the build-up structure, and the defensive shape (how narrow it is) are decisions made by the coaches to achieve their goals, and in large part, they’ve been successful, but I’d be lying if I didn’t say improvements can be made in its execution.

My only real worthwhile comment tactically is that at times, the passive press enabled by Kobel can sometimes be their own downfall. I’m not entirely sure that a passive shape can exist while simultaneously being so narrow and achieving their goals against big sides (who have beaten BVB at almost every turn this year). It works well against the lower sides, but it might be wise to increase the tempo of the press (like against Koln) or use a lower defensive block (like against Sporting @home). We’ve seen the success of both simultaneously with Freiburg.

As for the luck category; that is all it is. Every time a tactician is hired by a club they say something like “my philosophy is to eliminate luck” and everyone goes crazy (looking at you United…) but really, that is what football is about. Those moments of ‘luck’ are where you can hold your hands up and say “I was beat!” and that is ok. Football works in mysterious ways sometimes and you rack up 2.4 xG to your opponents 1.0 (with a pen) and you still don’t score. C’est football or something idk.

Conclusion:

Believe it or not, Dortmund don’t suck. Well, they do, but it’s mostly their own fault. Rose and co. have actually done well, but eliminating silly mistakes will be this team’s next step. If they never make that leap, they will never be successful unfortunately, simple as that. Just comes down to making some meaningful changes for the young man between the sticks, who has been immaculate this season.

@bvbimagesbew on Twitter

I’d like to do this test with other teams over time and see whether or not my conclusions hold water in the grand scheme of things, but I think this was a pretty decent way to consider a team’s overall success in their season. If you have any suggestions on my methodology, which I am well aware is all-too-binary, I’d genuinely be curious on what you think is a better way to do this.

Thanks for reading!

— JZ

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store