Are there sweet spots for direct free kicks?

I joined Stratagem Technologies last Monday and instantly became absorbed by the depth of the data available to me. For my first task, I was asked to find the “sweet spots” of direct free kicks by investigating whether there are positions on the pitch where significantly more are converted.

Personally, I love the excitement a direct free kick produces, especially when there’s a high quality set piece taker waiting to take it. I went to watch Barcelona host Sevilla at Camp Nou in February this year and remember the excitement when Messi was standing over the ball waiting to take a free kick; we were all just waiting for the inevitable!

With the same level of excitement running through me, I started to explore the StrataBet databases trying to find all direct free kicks, both scored and not scored, between the 1st of July and 3rd of October 2016. As I dived into the world of NoSQL and Python programming, I couldn’t help but hypothesise where the sweet spot for free kicks could be. Is it right in front of goal? Just offset from the centre of goal? Is it closer to the goal or just a bit further away? StrataData helped me find the answers, thankfully.

My search covered 1,488 unique matches across various European leagues, like the English Premier League and Championship, German Bundesliga 1 and 2, French Ligue 1, Dutch Eredivisie, Portuguese Primeira Liga and the Norwegian Tippeligaen to name just a few. I also incorporated international club leagues like the United States MLS and the Japanese J1 League, as well as European club competitions like the Champions League and Europa League. Finally there were also a few games from an international competition for good measure, with the last six matches of Euro 2016 included.

To find the sweet spots, I set my pitch coordinates relative to both free kicks scored and not scored and went about exploring the data (additionally trying to find the guy who chanced his arm from near enough the halfway line!)

I decided that if I was going to find the sweet spots, I was going to have to set my own boundaries, and these are shown by the grey rectangles in the figure below. Zones 2 and 3 are where I found most goal clusters, and so to help understand the impact of distance and angle from goal I decided to create smaller rectangles, in an attempt to capture the granularity of the dispersion of the data points. The yellow dots represent free kicks that were not scored, whilst the blue dots represent free kicks that were scored, and the yellow dots are darker in positions where more free kicks were taken. If you ignore the grey rectangles, you can probably see clusters of blue dots yourself:


In the StrataBet database, there were a total of 1,433 direct free kicks that weren’t scored, whilst 132 found the net. This gives us an overall conversion rate of 8%. So lets explore where on the pitch these were converted, keeping in mind that rectangles with exceptionally low attempt numbers will be overlooked.

Firstly, it seems clear that free kicks taken from the sides of the penalty area (rectangles A1 and G1) require incredible skill (and some luck) to find the net, with both the right and left segments resulting in a conversion rate of just 4%.

Then we have zone 2, which is a section of the pitch that at first may appear to be too close to the goal for someone to get the ball up and down quickly enough to score, unless there is an angle wide enough to exploit. This is highlighted by the conversation rates on the far right and left rectangles (B2 and F2), which are 13% and 8% respectively, compared to those rectangles closer to the centre (C2 and E2), which are 0% and 6% respectively.

Zone 3 further highlights the parabolic relationship between distance and angle from goal. Here, the sweet spots are in the rectangles closer to the centre (C3, D3 and E3), which show a conversion rate of 14%, 13% and 7% respectively. These rates appear to follow the curve of “the D”, indicating that the distance away from goal is important relative to the angle of the free kick from the goal:


Given this relationship, I also found that conversation rates are not necessarily weighted equally across both sides of the pitch. For instance, this sample shows that free kicks from the right are converted more than free kicks from the left. Take rectangle C3 on the right, which has a conversion rate of 14%, and it’s left sided equivalent E3, which at 7% is exactly half of this.

This pattern is also evident across zone 2, where rectangle B2 on the right has a higher conversation rate (13%) than it’s equivalent left sided rectangle F2 (8%). This seems interesting enough to require a deeper analysis and I will seek to explore it in more detail in a future blog post, so stay tuned on that.

In terms of zone 4, I decided to open up the rectangles here due to the spread of data points. The greatest conversion rate is 7%, from direct free kicks taken in line with the goal (D4). This is comparable with the other rectangles in this sector, and conversion rates across D are generally high (D2 and D3 each have 13%).

To finish, here’s a table outlining the attempts, goals and conversion rate for each rectangle on the pitch:


In conclusion, it seems that direct free kicks that are scored have a parabolic relationship between distance and angle from goal, with rectangle C4 producing the greatest conversion rate. Since the 1st of July, direct free kicks from the right seem to produce more goals than direct free kicks from the left, whilst we also find a consistently high conversion rate with direct free kicks taken in line with the goal (zone D).

There are a lot of interesting things here from our perspective as a company. The most valuable outcome of this investigation is that we can improve our training of data collection analysts on the most effective ways to categorise direct free kicks, depending on their location.

In terms of the future, please stay tuned for an analysis on right and left sided direct free kicks, as well as another follow-up piece investigating how free kick conversions differ by league, by team and by player.

Until next time!

Sagar Jilka (@DrSagarJilka)

P.S. If you’re wondering who chanced his arm from near the halfway line, it was Luigi Bruins for Excelsior against Groningen on the 13th of August.

Click here to read part 2 of this post.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s