/cdn.vox-cdn.com/uploads/chorus_image/image/56010925/612978812.0.jpg)
Numerous articles have been written on the departure of Leonardo Bonucci and the arrivals of Mattia De Sciglio and Wojciech Szczesny. This piece is different from the others. I am taking an analytical approach to measuring the impact of these transfers. There are no video highlights in this piece. In a way, the stats are the collections of the video highlights.
I ask: How much will the departure of Bonucci negatively hurt Juventus? Who is best poised to replace him? Are De Sciglio and Szczesny going to improve Juventus' defense?
Bonucci
Bonucci has been lauded as the world's best center back in the last few years. His transfer to AC Milan shocked everyone. Not only did he move to a direct rival, but Juventus also only received €42 million.
How much does it hurt Juventus defensively?
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975845/table_1.jpg)
Bonucci's long pass ability is undoubtedly world class for a center back. Couple that with his defensive skills and most people agree that he is a unique defender in the world football. Last season, among five Juventus' center backs, both Bonucci and Andrea Barzagli completed about 64 percent of the long passes and ranked four and fifth highest in Serie A, respectively (among 80 center backs). Bonucci and Barzagli were also very similar defensively — both players had a lot more interceptions than tackles.
Across the league, only three other center backs resemble the play style of Bonucci:
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975849/plot_1.jpg)
Andrea Masiello (Atalanta), Raul Albiol (Napoli) and Kostas Manolas (Roma) all averaged more interceptions than Bonucci while completed long passes at an exceptional level (8th, 10th and 17th highest in Serie A, respectively). Therefore, Bonucci is quite unique.
So how good Juventus' defense was when Bonucci played?
For analytic purposes, a defense can be broken down into three parts:
- To prevent the opponent from keeping the possession.
- To prevent the opponent from generating a shot.
- To prevent a shot to convert to a goal.
Obviously, the three parts can influence each other but it is not measured here. The first part is largely related to the ability of a team to keep the possession away from the opponent while the third part is mainly affected by the performance of a keeper.
The best way to measure the performance of defense is by looking at how well a team prevents its opponent from creating a shot. I measured the performance and importance of a player to the defense of his team: I calculated the number of shot every defensive lineup conceded in Serie A last season. The number was normalized to the amount of the possession the opponent enjoyed and the strength of the opponent. I calculated this value for all the lineups when a particular defender was present or not (click here to see the details of the calculation).
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975851/example_table.jpg)
For example, the lineups featuring Medhi Benatia prevented the opponent from creating shot by about 10 percent while the lineup not featuring him prevented the opponent's shooting by about 17 percent. Therefore, his plus/minus is plus-7 percent, meaning the opponents created 7 percent more shots against the lineups with Benatia than those without him. Juventus played the opponents of similar strength whether or not Benatia played because the average number of the opponents' points are very similar. The opponent's strength was calculated by averaging the number of points they got in Serie A.
You can click here to see the methods and the full list.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975855/table_2.jpg)
The best Juventus' defensive lineups featured Daniele Rugani and Barzagli. The lineups featuring either player prevented the opponent's shooting by more than 30 percent. In their absences, opponents shot only about 9 percent less than their usual rate and their plus/minus are about plus-25 percent. However, Barzagli and, especially Rugani, played against a lot weaker opponents than their colleagues.
In this regard, Bonucci was the worst Juventus' center back. The lineups featuring him only prevented the opponent's shooting per possession by about 6 percent. His plus/minus is a staggering plus-30 percent, against opponents of very similar strength. Therefore, the team's statistics suggest that Bonucci is not the world best center back. He can't be the world best if he isn't even the best for his team or in the league.
You can say that Rugani and Barzagli were the most important center backs for Juventus based on their relative performances. However, you can also say that Giorgio Chiellini was the most critical center back for the team: although his presence only improved the defense by 3 percent, he played against a lot stronger opponent (59.9 points vs. 41.1 points).
One can argue that preventing shots does not mean preventing goals, so I looked at the number of goals scored and conceded when the Juventus' center back played (or not).
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975903/table_5.jpg)
Again, lineups with Bonucci conceded 0.81 goals per game, almost 100 percent more than the lineups not featuring him. In this measure, lineups with Bonucci performed the second worst other than the lineups with Benatia. Lineups with Rugani outperformed anyone else. The lineups with different center backs had a mild effect on the goals Juventus scored. Still, the lineups with Rugani or Chiellini outperformed other center backs.
My model already normalized the performance of the lineups by opponent's possession and strength, but it doesn't mean that, for example, Rugani would perform as well when playing against much stronger opponents. Consider an extreme analogy: A team featuring Chiellini or me may perform similarly if I am only playing against a bunch of 5 years olds while Chiellini plays against professionals. Our normalized performances may be similar, but it does not mean that I will perform like Chiellini when I play against the professionals. If Rugani is Juventus' plan to replace Bonucci, he will need not only more playing time but also to play against much stronger opponents. The positive note is that the lineups featuring him outperformed most other players by a mile in any measures.
Why did the lineups with Bonucci perform so bad?
A possible answer comes from the defensive formations that Juventus used last season. Juventus has played with a back three (three center backs) or back four (two center backs). In general, Juventus defended better with three center backs, limiting the opponents' shooting by 29 percent, as compared to only 9 percent with two center backs. Every center back performed better in a back three configurations.
Although Juventus defended better without Bonucci last season, one should not underestimate the impact of Bonucci's departure: The lineups with Bonucci decreased opponent's shooting by 6 percent, a number that is slightly above the league average (30th best out of 96 center backs). He also had the most starts in the league among Juventus's center backs (26). Replacing him can be a problem, considering that the current center backs are injury-prone. Chiellini and Barzagli are another year older and will probably play even fewer games than last season. Playing Benatia had negative impacts on the goals conceded and scored last season. Moreover, as I mentioned before, it is unclear whether Rugani could perform at such a high level against a much stronger opponent. Juventus should look into getting another center back this summer.
It is also challenging to replace Bonucci's passing ability. Only seven center backs completed 150 or more long passes last season in Serie A: Masiello, Bonucci, Bruno Alves (Cagliari), Danilo (Udinese), Gabriel Paletta (AC Milan), Nicolas Burdisso (Genoa) and Wesley Hoedt (Lazio). Defensively, lineups featuring Paletta and Alves performed worse than Bonucci last season. Danilo only completed long passes at about 43 percent accuracy. Every player other than Hoedt is older than Bonucci, and Hoedt will be pricey precisely because of his young age (23). But Mattia Caldara is joining the team next season so Juventus may only need a temporary replacement (with quality).
Looking for a solution within his team, Allegri has experimented Rolando Mandragora as a center back in the friendly match against PSG. This strategy is risky because it is not easy to turn a midfielder into a full-time defender. Just look at Antonio Conte's experiment with Luca Marrone that never worked out. This season, we may also expect the midfielders to drop back a lot more frequently than last season to collect the ball and help with the distribution. In that case, A 4-3-3 may provide more balance than the 4-2-3-1. Claudio Marchisio looks very sharp in the preseason friendlies so far. A three-man midfield featuring him and Miralem Pjanic will be able to cover the loss of Bonucci's passes.
Most people — because of Pep Guardiola? — think that Bonucci made Juventus, but the data suggest otherwise. It will be interesting to see which argument is correct next season.
De Sciglio
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975869/table_3.jpg)
De Sciglio's transfer attracts a lot of criticism and is, in a way, justified. Opponent created plus-7 percent more shots against the lineups featuring De Sciglio, a number good for 60th amongst 95 fullbacks. For comparison, the lineups featuring any Juventus' fullbacks limited the opponent's shot between minus-9 to minus-25 percent. In fact, the lineups featuring Dani Alves had the best defensive performance for Juventus last season. In this regard, De Sciglio would be a significant downgrade compared to the man he replaced.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975871/table_4.jpg)
But De Sciglio's mediocre record may be because of the woeful team defense of AC Milan. They had the fourth-worst team defense last season. In fact, De Sciglio was the most important defender for AC Milan last season, as their opponents had 14 percent fewer shots against the lineups featuring De Sciglio than the one without him. His plus/minus number is the highest among AC Milan's defenders and 13th highest among Serie A's fullbacks. He was Milan's most critical player in defense.
You can look at these pieces of evidence in the opposite ways. On the one hand, being the best defender of a bunch of bad defenders does not mean much. On the contrary, it is in fact very hard to elevate the defensive performance of a team full of bad defenders. You can use that kid/professional analogy: De Sciglio is clearly a grown-up among the kids (other Milan defenders), whether he is able to play with the big boys (Juventus) is unknown. Offensively, he is clearly not as talented as Dani Alves. He has problems placing high-ball crosses. However, his ground ball crosses are very precise and he can use both legs equally well. Only after competitive matches, we will know if he is a good purchase.
Szczesny
Szczesny was the second-best keeper last season.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8975873/plot_2.jpg)
It took 12.5 shots to score against Szczesny last season. For comparison, It took 13.1 shots to score against Gianluigi Buffon and 11.7 to score against Gianluigi Donnarumma.
Here is a reminder how good Szczesny was last season: Roma had the second-worst team defense last season. However, Roma only conceded 36 goals, second FEWEST in Serie A. Szczesny — and also Roma's dominance in possession — was a major reason why Roma conceded so few goals.
If Szczesny is able to maintain the level of play in Juventus, this transfer will represent a major coup. They spent less than €12 million on a keeper most similar to Buffon in terms of performance last season (even better than Donnarumma). Granted, Donnarumma is 10 years younger and is probably why it will cost €90 million more to buy him. But Szczesny is still only 27 years old and Juventus can have the keeper position locked up for 4/5 years after Buffon retires if his performance last season is not a fluke.