Free-kicks: Where should teams shoot from?

Welcome back to Cambio de Juego, but to something a little different. Last season I produced data-focused articles on Liga MX and Mexican football three-to-four times a week, analyzing every jornada, key tactical concepts and stand-out players. During the summer I dug into the transfer window, and assessed how some teams should look to rebuild their squad.

Now, alongside some pieces on this season, I want to delve deeper into specific aspects of football and what we can learn from data analysis. Where possible, the studies will focus on Liga MX, however, given the lack of public event data for the league, I will often have to use different data sources, like in this article.

To begin, I want to look at attacking free-kicks. It’s a bit of a simplification, but there are two main options for a team when they have an attacking free-kick; shoot or pass. So, depending on the position of the free-kick, what’s the best decision? Where should teams shoot from?

Summary:

• The sample size may be too small, but by using StatsBomb’s xG model, it appears that, on average, teams could be more efficient from free-kicks by shooting less, from various zones.

• Individual quality of direct free-kick taking can have a huge impact on goal probabilities and understanding a player’s dead ball ability should be part of the set-piece planning process, especially as ability can differ per zone.

• This analysis didn’t take into account that teams can increase their indirect set-piece efficiency through intelligent routines or by having strong aerial threat, or that direct free-kick productiveness can be improved by blocking the goalkeeper’s sightline.

I used StatsBomb’s open data sets to tackle this question, and their industry-leading xG model. I took all of their publicly available data from competitions post-2000, including; the “big 5” European leagues in 2015-16, men’s and women’s international competitions, the Indian Super League, WSL and NWSL, and Lionel Messi’s career.

The method of analysis, using all of the competitions outlined, looks at xG created from free-kicks and the 5 actions following the free-kick (only passes or shots), to take into account the impact of rebounds and subsequent corners. The free-kicks were divided into direct and indirect (passed/crossed). Any plays with more than 1 shot were adjusted to avoid over-rating the impact of rebounds and remove the possibility of the xG exceeding 1.

Short taken free-kicks were filtered from the study, which is another, but less popular option. Therefore, only free-kicks played into the box were taken into consideration, and the analysis was limited to just final third free-kicks.

First of all, let’s take a look at average decision making from the analysed free-kicks.

Shooting tendencies from free-kicks are pretty clear. From within the width of the penalty area, the x coordinate 88 onwards (using StatsBomb’s coordinates), the majority of free-kicks are direct shooting attempts. However, there are less shots from the right-side, probably due to the lack of left-footed players. The majority of wide free-kicks are crossed, with shots slightly preferred from the central channel (width of the 6-yard-box) between x coordinates 80 and 88.

For reference, the maximum x coordinate in StatsBomb event data is 120, and the edge of the box is 102.

But, where should teams be shooting from?

The heat map displays the average difference, per zone, in xG created from indirect free-kicks and direct free-kicks. More zones are blue, as passing has historically been a better decision than shooting in most areas (at least, within the data set available). Due to a limited sample size, some zones haven’t been included in the graphic.

This analysis suggests that teams may be wise to attempt indirect free-kicks more often, especially from central zones (coordinates 80 to 96) where shots are preferred. In one zone, where more than 90% of free-kicks are direct shots, an indirect free-kick creates, on average, 0.04 xG more. It’s not a huge difference, but something to consider.

However, the results may show an issue in a lack of large sample size, and there’s difference between the left and right sides of the pitch. On average, it appears more efficient to pass/cross from the left, but on the right side there’s very little difference between xG created from direct and indirect free-kicks. This could be related to impact of left-footers attempting free-kicks, but could also be the result of a few outlier events in a small sample size, and the impact of the Lionel Messi data set...

What if you have a great direct free-kick taker?

Thanks to StatsBomb releasing the Lionel Messi data set, we can assess how an excellent free-kick taker, rather than an overall average, impacts goal probability. Whilst few players historically have shown Messi’s deal-ball ability, this analysis gives an idea of how a team’s most efficient decision can be impacted by individual factors.

The heat map below compares Messi’s goal probability (percentage of direct free-kicks which he’s scored, by zone) with the average xG of indirect free-kicks (based on the previous graphic). Using post-shot xG would be another option, however that metric isn’t available in the free event data.

Clearly, the ability of a team’s free-kick taker has a large impact on the effectiveness of free-kick decisions. Messi has outperformed his xG from central zones and particularly from towards the right-half-space. Within these zones, Messi’s direct free-kicks have been much more effective than the average creation from an indirect free-kick.

An indirect free-kick would normally be more effective from this zone…

However, footedness and angles are still important, and Messi perhaps attempted too many direct free-kicks from the left side during his career, where a pass/cross seems a better choice.

Understanding the free-kick quality of individuals, per zone, is an important part of preparing set-pieces and deciding which type of free-kick is the most effective. Training data could be collected and used in such decision making, to increase the sample size. Also, objective analysis could be useful to convince players, if a coaching staff decide to attempt more indirect free-kicks than what’s commonly seen in football.

Thank you for reading, if you’d like to contact me regarding any questions, comments or to consult my services, you can contact me on:

• Bluesky - https://bsky.app/profile/tomh36.bsky.social

• Linkedin - https://www.linkedin.com/in/thomas-harrison-a682a2175/

• Twitter/X - https://x.com/tomh_36

También, se puede leer este artículo en Español- https://cambiodejuego.beehiiv.com/