Before his time at Borussia Dortmund came to an abrupt end, Lucien Favre’s teams were well known for scoring loads of goals, and far more than predictive models would expect. Somewhat lost in the disappointment of second place last season was a record setting 84-goal Bundesliga total for the Black & Yellows. Before Haaland’s injury, the team this year was on track to break that record again. Taken as a whole, Favre’s Dortmund teams were on track to surpass even the goal scoring exploits of the Jurgen Klopp golden years of 2011 - 2014. But where Klopp was famous for aggressive defense and direct attacking football, Favre’s teams chose patience at both ends of the pitch, which often frustrated the fans, and sometimes even the players.
Dortmund under Favre consistently led the Bundesliga in passes completed and overall pass completion percentage. Dortmund had more short passes and fewer long passes than Bayern, and generally had fewer switches, crosses and through balls, and they were rarely flagged for offsides. The overall profile was that Dortmund progressed the ball cautiously, looking for a breakthrough, and played the ball back out to try again if they couldn’t find a gap in the defense. How did such a cautious approach lead to so many goals, and what changed?
For most of the past decade, Favre has been famous to a certain type of fan for coaching teams that consistently score goals without taking a lot of shots and wreck everyone’s expected goals (xG) models. In 2019/20 Dortmund only took an average number of shots in the Bundesliga, tied with Hoffenheim for 8th in the league. According to Fbref.com Dortmund scored 24.8 goals more than xG would have predicted, leading all teams in the big 5 leagues by a wide margin. While Dortmund didn’t take as many shots as other top teams, Dortmund’s shots were converted at an incredible rate. Last season, just like the season before, Dortmund led the league in percentage of shots on target, goals per shot and goals per shot on target. Looking at Understat, Dortmund, as a team, shot less, missed less, had their shots blocked less, and saved less, hit the post less and scored more goals than any team other than Bayern Munich. Their shots just went in the net more than other teams.
Marco Reus, Paco Alcacer, Jadon Sancho and Erling Haaland all developed reputations as marksmen under Favre but the trend spread to the entire team. The same pattern had previously emerged at Nice and Borussia Mönchengladbach, but with different players. Did Favre have a secret insight about shooting accuracy that he imparted to his players, or as some would have it, did he practice dark magic?
Writers for Statsbomb have accused Favre of being some kind of xG warlock. No doubt, he was put on this earth to destroy their carefully crafted statistical models. The inner workings of the most popular xG models (StatsBomb, Understat, and Opta) are not uniform, and beyond a certain level of public access the technical details can seem opaque. The basic idea is to compare each shot taken to all similar historical shots in the data, based on location on the pitch and several other factors. By determining what percentage of those similar shots resulted in goals, you can place a predictive value on the current shot separate from the randomness of blocks, deflections and saves.
In this way, analysts focus on the quality of the shooting chances produced, which has been shown to be far more predictive of future performances than simply looking at goals scored. Using expected goals to predict future performance is one of the key elements of 538’s predictive modeling, which has never really rated Favre’s Dortmund teams very highly before this year. Analysts looking at xG have often pointed to Favre as an outlier, whose teams consistently overperformed. At the same time, they have questioned whether Dortmund could really keep beating xG by such incredible margins.
But this year, rather than questioning whether Dortmund’s scoring prowess could continue, analysts have been left wondering why Favre got sacked when his team led the league in xG differential. On the day Favre was fired, 538’s model rated Dortmund higher than it had at any time in the two previous seasons. Dortmund has only lost the xG battle in two matches all season. The first was the home draw against Lazio, who benefited from a very questionable penalty call, which significantly inflated Lazio’s xG. The second was the Stuttgart game, in which the team seemed to completely quit on Favre. Despite sitting in 5th place, having allowed 16 goals in the last seven league matches, Dortmund still leads the league in xGDiff. While Bayern and Leverkusen have performed 10 or more more goals above expected, Dortmund has scored less and conceded more than expected goals would predict, reversing the trend of the previous two years.
Dortmund’s attack showed signs of change before Favre was fired, seemingly taking a more direct approach, either out of necessity or to suit Haaland. This season Dortmund has taken more shots, with a greater xG than ever before, but without Favre’s trademark efficiency, which has resulted in the team scoring slightly less than its xG for the first time since Favre joined BVB. The November game against Bayern, in which Dortmund repeatedly sent long balls over the top for Haaland, demonstrated a clear change in approach, matching the highest total of through balls for a single game under Favre. But even as Dortmund’s attack seemed to morph to fit Haaland’s incredible talents, the rest of the team lost its shooting touch.
Jadon Sancho has been the poster boy for the team’s problems in front of goal. He is shooting slightly more often per 90, and his xG per shot is only slightly lower. However, from 24 shots in the league this season, only 5 have been on target, and none have gone in. This marks a drastic drop-off compared to last year, when Sancho scored 1 out of every 3 shots. However, the offensive woes are not limited to Sancho. Julian Brandt has yet to score, Marco Reus’s three goals are still 0.7 fewer than expected goals would predict, and Thorgan Hazard has missed most of the season with various injury problems.
The end result is an overreliance on Haaland’s production. As a squad, Dortmund’s percentage of shots on target, goals per shot and goals per shot on target have plummeted, even as Haaland has put up incredible numbers. Dortmund this season are arguably missing a dozen goals that they would have scored if they had continued to score at the same rate as previous seasons. And once Haaland picked up an injury, the dip in goals scored was precipitous.
Haaland’s return won’t be in time to save Favre’s job and it doesn’t answer why the defense started underperforming a few games before Haaland’s injury. The good news is this isn’t the first time Dortmund has gone through a shooting slump. Something similar happened in Klopp’s last season, when Dortmund’s underlying statistics indicated the team was still much better than the disastrous results they were getting on the field. And as Michael Caley wrote then, the strong expected goals differential suggests that Dortmund is likely to turn it around.
What do you think led to the sudden collapse in Dortmund’s offensive production? Was it as simple as Haaland’s injury or was it always bound to happen? Either way, Favre’s spell faded and it will now be up to others to chart a new course.