Hello! Welcome to the last section of this miniseries detailing the different radars being used in this project. Last post, we explored the midfielder radar, covering a well-rounded set of metrics to capture the variety of midfielder styles. This time we’re a little more specialised around the art of goal scoring. Using metric taken from fbref and adapting them to suit the aims of this project. Let’s get started!

The Attacker Radar

Attackers, or forwards, have the primary aim of scoring and creating goals - whilst defenders may be seen as preventing defeats, attackers often are the match winners. This radar will also involve 12 metrics covering different styles of forwards.

Metric 1: Shooting

To reflect its importance to forwards, this shooting category involves 5 metrics. The obvious one is goals per 90 - need I say more? A less typical metric used by the average fan is something called non-penalty expected goals (npxG) per shot, per 90. Excluding penalty kicks which have a vastly inflated xG score (around 0.8), the number of non-penalty expected goals is divided by the number of shots taken by an attacker. This shows how many goals we would expect the attacker to have scored based on the chances they had and controls for the number of shots an attacker takes. An attacker may play for a very good team and thus have a high npxG, but this could just be due to them having more opportunities to shoot rather than being better at shooting as such.

In a similar manner, the next metric taken the number of goals scored per 90 and subtracts the non-penalty expected goals per 90. This will give us either a positive value, suggesting the player outscores their xG score, it could be 0 in which case the player is converting chances as expected or it could be negative in which case the player is under-performing in terms of chance conversion. This metric is not ideal (no metric is!) as again factors like shot volume, team tactics or play style can bias their goals-minus-xG performance.

Shot on target percentage is the fourth metric, showcasing the attacker’s ability to shoot accurately - albeit regardless of how well they convert those shots. But to account for this limitation somewhat is the last metric in this category - goals per shot on target per 90. Having lots of shots on target is great, but should correlate with goals - this metric helps to show if this desired relationship is occurring for an attacker.

Metric 2: Creativity

As with the previous two radars, assists per 90, expected assists per 90 and shot-creating actions (SCA) per 90 all capture an attacker’s creative output. Creativity is probably most important for attackers (and attacking midfielders) and their reputation can heavily rely upon metrics like assists. However, creativity is not important for all types of attackers. Some attackers simply contribute with goal scoring itself and rely upon the creativity of teammate for their and their team’s success.

Metric 3: Dribbling

To reflect a particular type of attacking player who looks to take on the opposition directly, dribbling can give insight into their ability at doing this. Take-ons attempted per 90 tell us how many attempts a player makes at dribbling past an opponent. This metric alone can give a good indication of what kind of attacker the player is - a high take-on rate suggests a player that is direct and likes to dribble rather than favouring a pass-first approach. However, this does not say much about how successful this player’s dribbling is. This is where the take on success percentage comes into play, allowing us to compare dribblers.

Metric 4: Hold-up play

An alternative style of attacker might be someone who exploits their physical prowess to help their team progress the ball into the final third or even to attempt shots on goal using their strength, aerial reach or speed. The percentage of aerials won gives an indication of how well the attacker fairs at battling with opposition players to win the ball in the air. The other metric in this category should be interpreted in reverse to the previous - rather than a high percentile for aerials won being “good”, performing well on this metric corresponds to a low percentile score. This last metric is mis-controls per 90. To successfully use their physicality to progress the ball by holding off opposition players, the attacker must also be able to actually keep control of the ball whilst under pressure from an opponent.

Conclusion

And just like that, this “preparing our radars” miniseries comes to a close. We’ve looked at 4 different kinds of radars, ending with an exploration of the metrics involved in assessing attacking players. Hopefully this helps to make the metrics being used in these percentile radars a bit clearer as well as making you aware of some of the things to consider when interpreting them. The next post will be the last in this project where I will go over the final dashboard and showcase it to you, the reader. Thank you for following along!