Thread by @rozlivek99

⚽️| ANALYSIS OF THROW-INS in the Czech Football League

📈| How several attributes influence the throw-in performance?
💣| Is switching an efficient weapon?
🎯| Throw-ins aiming into 🆚 outside of the penalty box
📊| Team Analysis (2021/2022)

💈👇 (1/20)

#throwins #vizzes

There are essential terms related to the analysis of throw-ins.

💈👇 (2/20)

To intoduce, there are a few probabilities of each throw-in type. We have an unbalanced dataset because we have 90% of complete throw-ins and 10% incomplete throw-ins. The mean xG generated is about 0.003 xG.

💈👇 (3/20)

As shown, the defensive and middle throwing sonars illustrate a preference for more progressive throw-ins in order to minimize the risk of losing possession closer to their own goal.

💈👇 (4/20)

The peak of the completion rate (CP) is between 9 and 12 mts. In the mid & att zone, xGD is increasing as the distance increases, on the other hand, possession retention (PR) is dropping as the distance increases.

💈👇 (5/20)

There is no significant difference between each zone. Throw-ins directed toward the opp’s goal have lower probability of being completed or retained. The little rising of xGD is valid for the att zone. The peak of xGD is the bin with att lateral throw-ins.

💈👇 (6/20)

Valid for all zones, the quicker throw-in, the greater the possibility of the completion (Cmp) or PR. Big jumps across bins for PR tells us that the timing of a throw-in has bigger impact on PR than Cmp. The mean xGD of att throw-ins are discussed (in detail) later.

💈👇 (7/20)

The clusters, discovered in the def zone, demonstrate one dominant pattern when we selected throw-ins when the possession is kept +10s following the throw-in. This pattern involves two-pass switching. Thus, let’s discuss switching in terms of xG...

💈👇 (8/20)

In the def zone, switches often improve the probability of scoring while decreasing the danger of allowing a goal. In att zone, the mean values of xG and xGA from throw-ins with a switch significantly drop within any given time.

💈👇 (9/20)

When the 1st pass of a two-pass switch results in the loss of the ball, there is a greater likelihood of conceding, implying that if the situations allow for a one-pass switch, it seems to be a safe choice in contrast to loss the ball in the center of the pitch.

💈👇 (10/20)

There is a relatively high expectancy of being accurate with throw-ins aimed towards the penalty box's half-spaces and near-post throw-ins. PR within 7 s seems to be less likely to be held the closer the opponent goal, but PR is in a similar manner.

💈👇 (11/20)

It is more efficient to play throw-ins aiming into the box, since they have a higher probability of scoring and not conceding a G, as determined by the xGD and GD statistics. The bin with a time between 20 and 25 s after previous action has the greatest xG and xGD.

💈👇 (12/20)

Teams, that are losing, seem to be more effective with throw-ins in all areas except the xG, which is more closely connected with overall team performance.

💈👇 (13/20)

Losing teams tend to throw at a less risky angle and look for positions that are more advantageous for ball possession. The exception of att throw-ins, where the losing team attempts to use more long throw-ins aimed into the penalty box to create more chances.

💈👇 (14/20)

The table shows the best strategy of taking a throw-in in the certain zone. E.g., the ideal option how to take the att throw-in to create chance is to play a long, slow, and lateral throw-in, which aiming into the penalty box without utilising a switch.

💈👇 (15/20)

To begin with the team analysis in the season of 21/22, the triangular plot shows the clusters that corresponds to each team’s domination of their playing throw-ins. For example, Slavia Praha tend to create chances from retaining possession.

💈👇 (16/20)

🔴| Defensive and Middle Throw-ins:

Sparta Praha are the kings of playing defensive and middle throw-ins in terms of completing, possession retaining, and xG. Č. Budějovice often utilize switching after their throw-ins.

💈👇 (17/20)

🔵| Attacking Throw-ins:

Slavia Praha are the best of playing attacking throw-ins in terms of xG, they also dominate in possession retention.
Slovácko and CEB often use throw-ins aiming inside of the box, although their xGD values are significantly different.

💈👇 (18/20)

The xThrow model was constructed based on the examined features. Unfortunately, the performance of predicting incomplete throw-ins is low. Despite this, the charts help to comprehend the impact of each feature for predicting the completion of a throw-in.

💈👇 (19/20)

To conclude, each match has an average of 54 throw-ins (46% out of all set pieces types), demonstrating the high occurrence and the potential importance of throw-ins in terms of team success. Thus, we must pay attention to the details, such as the perfect grip.

💈👇 (20/20)

The data analysis has been part of my thesis. BIG THANKS to @Wyscout for providing me with the data. Some of vizzes were inspired by @etmckinley, @elliott_stapley, @pwawrzynow. The photo with the grip is from the king of the throw-ins @ThomasThrowin. Any feedback is appreciated.

⚽️🇨🇿| When I looked at the posted charts describing the team analysis of throw-ins, I realized I had posted the plots with errors. I apologize for that, and I hope you’ll accept my apology. So, for 🇨🇿 followers in particular, here is the updated team analysis:

💈👇 (1/4)

⚽️🇨🇿| Triangular throw-in style plot:

💈👇 (2/4)

🔴| Defensive and middle throw-ins:

💈👇 (3/4)

🔵| Attacking throw-ins:

💈👇 (4/4)