xG data journey – the raise of M. Gladbach

After getting all this expected goals data, it’s of course most obvious to take a look at the insights such data can produce and in which way xG can be interpreted. I have decided to take a look at the current development of Borussia Moenchengladbach in the Bundesliga . Even if RB Leipzig took over now the first place, the development of Gladbach in comparison to the last seasons is impressive. And now I just want to know: Does xG data reveals the secret of Marco Rose?

xG vs goals

Goals decide a match. Goals are the only result, which matters and the division table tells the truth. But that’s no longer the case. With the introduction of advanced metrics like Expected Goals (xG), we are able look behind the result. Expected goals provide the information, what would happen on an average day.
Looking at this match of Freiburg against Wolfsburg reveals a lucky or very effective team, which won 1:0 but just created chances for 0.5 goals. In comparison Wolfsburg created chances for 1.5 goals, but did not score once.

Freiburg vs. Wolfsburg (07.12.2019)

Looking at the xG data for a single game, might get interesting, when you want to know, whether a team was just unlucky. But that doesn’t help analyse the performance of a team. Therefor you need to consider the xG / goals ratio over a longer period of time.
Following picture shows the attack and defense performance of Gladbach for goals for / against and expected goals for / against. I used a 7 games moving average to smooth the variance of single games. For a even bigger long term analysis this number can of course be increased.

Until Marco Rose joined, we can see a typical xG / goals ratio of a constantly performing team on the attacking side. The goal line fluctuates over and under the xG line. During some periods the team overperforms or underperforms in respect to their xG data. But in general the average number of scored goals always trends to the xG line. That’s the so called regression to the mean, as expected goals represent the expected average.
But since Marco Rose joined, we can see a clear difference. The average number of expected goals increased from 1.5 to over 2.5. On top the average number of goals scored is even smaller than the xG average. So you could expect Gladbach to get even more efficient in attacking.
The defense side provides not so many insights. Currently the team has an overperforming defense. The concede less goals than expected based on the xG data. But similar phases already happened e.g from December 2018 to Februar 2019. The increasing number of goals conceded last games already indicate a regression to the mean for the defense.

xG vs shots

The next logical question is, how Marco Rose achieved the increase of average expected goals per game. There are basically just two options: increasing the number of scoring opportunities or increasing the quality of scoring opportunities.

Marco Rose went the second way. While the number of shots stays relativ constant around 15 shots per game, he achieved a way higher xG value per shot. The probability to score with each shot was dramatically increased from 10% per shot to about 18% per shot.

The shot quality Gladbach allowed their opponents provides not many insides and just confirms the interpretation of the xG / goal ration timeline. Gladbach is still in a phase of providing shot opportunities to their opponents of worse quality. But such phases already happened in the past and can be interpreted as the normal variance of a team over time.

Conclusion

We now determined, that the good results of Gladbach are caused by a improved offense, which is able to score more goals as they increased the quality of their scoring chances. If you now want to go a step deeper and determine, how they increased the scoring opportunity quality, you would need more data. An increased probability of scoring can be reached by respecting all the single features, which are used for a expected goals model: shot position, free scoring opportunities, low pressure during shots. All this is provided by position data.

Marco Rose – the best coach?

Seeing the current result and the improvement in the offense, could we assume, that Marco Rose is the best coach of Borussia Moenchengladbach in the last years? Considering shot quality, we should not forget, that some years ago, there was a coach, master in limiting the shot quality of opponents: Lucien Favre.

Based on the data available for me, Lucien Favre was the best defense coach. On average a opponent had only a chance of less than 8% with each shot. So they were able to reach the 3rd place in the season 2014/15. In comparison Marco Rose is currently the best offense coach. The team has the probability of about 17%, when getting the opportunity. Will this lead to an even better end position? We will see.


If you have further questions, feel free to leave a comment or contact me @Mo_Nbg

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s