Advanced Passing Visualizations: An Analysis of Logan Costa's Passing

Last time, I gave you a sneak peek of a pass sonar; today, I'll offer a comprehensive overview of everything you'll find in the most important post of the month, which will be published on the 30th.

As mentioned earlier, I chose Logan Costa as the subject of this analysis. The Cape Verdean central defender — born in Paris — plays for Toulouse and stands at 1.89 m according to Fbref. Last season, he played a total of 3,279 minutes, marking his first full season as a key player in his career. Not yet 24 years old, Bologna has been interested in adding him to their squad for a few weeks now as I said previously, but it the end he’s gone to Villarreal. Through his numbers, I will illustrate the visualizations I've created. Without further ado, let's get started.

1. Last time, I hinted that the pass sonar was created by another analyst. Indeed, it's an improved version of the classic pass sonar created by that guy (@TheComeOnMan):

This version subdivides each angular column into further blocks, offering greater granularity in analyzing a player's tendencies to pass in certain directions and their completion rates through color usage.

These colors can be modified to indicate average expected passes (xP) or other metrics at will. However, there are some limitations to the visualization: the rings represent passes of 10.5 meters in length, progressing up to 105 meters (10 rings). This means that the first ring covers passes from 0 to 10.5 meters, the second from 10.5 to 21 meters, and so on. However, this limits the possibility of seeing the rings fill horizontally since the field is only 68 meters wide. Despite this, it is currently the best version I have managed to create of this visualization.

As for Logan Costa, we can see that his completion percentage drops significantly starting from the second ring, indicating a passing ability that is not exceptional. This is confirmed by the fact that he completes only 83.3% of his passes.

2. Where Logan Costa Directs His Progressive Passes:

The visualization is quite intuitive.

We see all progressive passes (defined as passes that bring the ball at least 17.5% closer to the goal, are vertical, longer than 5 meters, and do not start in the opponent's area) tracked to give a general idea. However, the main element of the visualization is the blocks where these passes end, with percentages for each block exceeding 1.5%. About 25% of Logan Costa's passes are classified as progressive, an interesting figure because it tells us that he is not overly conservative.

However, he completes about 90 fewer progressive passes than expected, which can indicate several things, but if we assume the model is perfect, it tells us that Logan performs significantly below expectations.

3. This Visualization is Similar to the Previous One but Represents Buildup Passes.

Though not exactly. In this case, there is considerable room for improvement: currently, I represent all passes originating in the first third of the field (the defensive third), but a true buildup would be better represented by considering only those that start a sequence or according to a different definition. Nonetheless, Logan Costa seems to perform well in this specific skill.

4. Here We Have Clusters Grouping Passes Based on Certain Characteristics:

I created 16 clusters and decided to show 4, the ones with the most attempted passes, displaying the values of ATOMIC-VAEP (a possession value model) using colors.

As in the other cases, I established numerical references to evaluate performance: for instance, we see that in the first cluster, Logan Costa completes 5.9% fewer passes than expected, and so on. In three of the four main clusters, Logan shows a negative balance, which is a red flag; however, the negative balance is not extremely damaging, ranging from 1.3% to 10.8% worse, at the worst.

This is something to consider if we are scouting a player who needs to facilitate buildup play for a coach like Italiano, who wants control of the ball.

5. Similar to the Previous Visualization, This One Focuses on Passes Where Logan Performs Poorly:

There's not much to add: they are essentially the same three previous clusters, plus one related to very long passes. If there is something positive, it is that in the other 12 clusters, Logan shows a better or similar balance to expectations.

6. The Last Visualization is, in My Opinion, the Most Interesting:

Here we see the passes Logan Costa made following a won tackle or an interception, simplifying, when he recovered the ball.

We can see that these passes occur in different areas of the field, with most in the defensive and central third on the right side, but also with some high recoveries near the opponent's goal. In this visualization, we can see that Logan completed about seven more passes than expected in these situations. Moreover, the color of the passes reflects not only the ATOMIC-VAEP value of the pass but also that of the tackle or interception, which are summed. So, a won tackle with a high value, followed by an equally valid pass, results in a very high-value action, highlighted by an intense red color.

In summary, Logan Costa appears to be a decent passer rather than an exceptional one.

None of the visualizations show extraordinary numbers, but in terms of buildup play and passes after recovery, he shows more interesting figures. Overall, he completes 2.56 passes above expectations for every 100 passes, or — put in a less complex way— he performs passes 2.56% better than expected, which is only the 47th best figure in Ligue 1 23/24 among central defenders with at least 1,200 minutes played (out of a total of 57). This ties into the fact that the player adopts a non-conservative style; his pass completion rate is indeed 83.3%, and as mentioned, it's already an improvement over expectations.

This suggests that probably in a less risky system, he might have better numbers in this regard, considering that last year Toulouse controlled possession for more than 55% in only 10 matches out of a total of 45 played.

However, it should be noted that his bold style doesn't add much value; in fact, Logan ranks 33rd among the same 57 central defenders for ATOMIC-VAEP per thousand passes, with a value of 0.964. This confirms the feeling that Logan still needs to grow and mature in terms of ball playing ability, which is consistent with his role, where the prime age is usually between 26 and 31 years old, and the fact that he didn’t play a lot of minutes in first team so far.

In conclusion, Logan Costa is a decent passer for Ligue 1 and a great subject for demonstrating these visualizations.

See you soon.

If you enjoyed The Cutback, I’d appreciate it if you would forward it to a friend. If you’re that friend, welcome! You can subscribe to The Cutback by clicking below.