Skip to content

Match Analysis Using Shot Quality A Review of the Melbourne v Hawthorn Game

Posted in ExpScore, Game Analysis, Scoring Map, and Shot Quality

Last week I introduced a model which can be used to calculate the true quality of any scoring shot taken during a match. This basically quantifies the way we already think about footy. A shot, close to goal, on a good angle and with very little pressure1 is a great chance. In tight games these chances are few and far between, if you want to win a match you need to take your chances.

Knowing the quality of every shot taken by both teams gives us a wonderful tool to analyse and review matches. Was the game particularly one sided? Who created the better chances? Which shots were the most crucial? Did one team throw away the win with poor finishing?

Today I’m going to give you a glimpse at a couple of ways Shot Quality can be used to tell the story of a game. And what better way to introduce these new tools then to review and analyse a match from the weekend just gone, Melbourne’s stirring win over Hawthorn at the MCG.


ShotPlot MELHAW201620

I’ve dubbed the graphic above the “ShotPlot” of the match. In it you can see every scoring shot made by both teams throughout the match, Melbourne in blue, Hawthorn in gold. A green outline indicates a goal was kicked with that shot. The size of each circle represents the quality of the shot. Bigger circles are better quality shots. So, for example, Christian Petracca picking up a turnover and waltzing into an open square…

…is considered a much better chance than Jack Gunston’s under pressure snap across the body from in the pocket.

What we can see from the ShotPlot is that both teams restricted their opponents from creating many high quality chances throughout the game, with territory inside the corridor and close to goal tightly fought. The big difference we can see between the teams is their conversion rates from 40m+ out.

Melbourne kicked 8 goals from the 10 chances they had from further than 40m out. The Hawks on the other hand created 16 of these chances but only converted 6 of them into goals.

Despite the 29 point final margin to the Dees, both teams have a Shot Quality Production of 97 points.2 This is roughly the average score we would expect an AFL level team to kick, given the quality of shots they created if we could somehow retake the shots over and over again many times. For example, we would expect Petracca to kick that goal from that position about 96 times out of 100, so the Shot Quality Production for that shot is 5.8 points. The Shot Quality Production for Jack Gunston’s kick was only 2.9 points.

What this tells us is that Melbourne did very well to capitalise and convert the chances that they created whereas Hawthorn were rather wasteful up front.

If both teams had kicked closer to the long-term league average, we could have had a much much tighter game. In fact, if we were to simulate each of the 49 shots taken over the 4 quarters many times, we see that slightly more often than not, Hawthorn actually win this match. This is primarily due to the fact they created one extra shot compared to Melbourne. This extra chance of a goal slightly outweighs the fact that the Dees created on average higher quality chances.

Win Prob MELHAW201620

So now we know what kind of shots each team created, what’s missing is when the chance occurred and what effect this had on the game.


 Score Worm MELHAW201620

The ScoreWorm shows the progression of the margin over the course of the game. This is likely a familiar concept to footy fans. The AFL publish their own worms and there are many others available including these great interactive ones from fellow footy analyst InsightLane. The worm I have here however contains one major difference.

Along with the thick black line that shows fluctuations in the game margin, I have also included a red line that shows the ongoing Shot Quality Production difference between the two teams. By comparing these two lines we can see the different moments of the game where a team managed to kick particularly straight or particularly poorly and how these moments may have effected the result of the match.

Take a look at the ScoreWorm from the 5th minute of the 2nd quarter through to about the 25th minute. In this period of time, Hawthorn had 6 shots on goal but only managed to convert 1 of them. Most of these misses were fairly routine shots from experienced players such as these two set shot misses from Burgoyne and Rioli.

In the same period of time, Melbourne only took 3 shots but Weiderman, Brayshaw and Watts each converted these shots to majors.

Later on in the match Melbourne kicked 5 goals straight in order to storm home and turn a 1 point deficit into a healthy win. Once again you can see some separation between the black and red lines here meaning the Dees kicked above expectation by finishing all 5.

If Hawthorn had converted a bit better earlier on, the Melbourne wouldn’t have been able to kick away like they did3. In fact, looking at the red Shot Quality Production line, you can see that on another day, Max Gawn’s shot after the siren could well have been a kick to decide the game.

None of this is to say Melbourne did not deserve the win, they played well, prevented Hawthorn from scoring easy goals and managed to convert their own chances very well. 29 points is probably not a fair representation of their dominance, but they certainly hung in their and gave themselves every chance of winning against the top team on the ladder.

Wrap up

These graphics are certainly not the only ways that scoring shot data can help us visualise and better understand a match. Although I am excited by this and think this kind of analysis opens up a whole new world of possibilities when it comes to discussing footy.

I hope to spend more time digging into SQP and I plan to write regular analysis/reviews if they seem to be well received. If you have any ideas or comments4, let me know here or follow me on Twitter.


  1. Think a set shot 20m out, right in front
  2. For old readers of the blog, I used to call this “ExpScore”. I’m transitioning into referring to it as SQP which was suggested by Nick Welch.
  3. Also, the Dees probably wouldn’t have had quite as much free space in their forward line late in the game if the Hawks were ahead and defending a lead like their Shot Quality Production suggested
  4. Or find any inconsistencies or mistakes in the data, it’s not perfect.


  1. I’d love to see this run for a large portion of the games throughout the year. Obviously in some games, i.e. big wins against Essendon, kicking accuracy isn’t a big factor, but I’d love to see how many other games were decided by kicking accuracy.

    One of the things that is so interesting about AFL is that it’s so difficult to consistently pick the winners. In any given week there will be 2-3 upsets but it’s very hard to accurately predict the upsets. I suspect the reason for this is at least partly to do with kicking accuracy. For example, Hawthorn effectively lost because they kicked worse than the average and Melbourne kicked better. This can almost never be predicted unless one team is obviously better/worse than the average.

    Love your work, keep it up.

    August 8, 2016
    • Thanks for the comment, I’m glad you’re enjoying the posts. I’ll try and look back when I get some time. There’s so much I want to do right now that time is really the biggest pressure.

      I agree completely. I think a lot of results come down to more or less “luck” on the day. I wrote an article about the randomness in goal kicking accuracy a few weeks ago. I have tried not to mention the “luck” factor too much here as I really think the idea of Shot Quality can be accessible to all but the concept of probability distributions is probably a bit tough for most.

      If you’re interested in the “luck” conversation in AFL, I strongly recommend Darren O’Shaugnessy’s latest paper.

      August 8, 2016
      • That’s a really interesting article and I also enjoyed your earlier one on goal kicking accuracy when I read it a few weeks ago. Keep it up!

        August 10, 2016
  2. Jeremy

    This is a really terrific analysis, and I think it can open up a whole world of new discussion about kick quality – as well as yet another way to measure team performance. Please keep it up!

    One question; where are you sourcing your shot location data & is it public? First time I’ve seen someone use it (outside of Champion Data)

    August 8, 2016
    • Thank you. I hope that this type of analysis does become more of the norm and we move away from just listing endless stats which nobody can really apply to the outcome of the match. Footy is about storytelling. If you have the right data, you can enhance that story.

      Unfortunately, these data aren’t public. I think great things could be done with scoring shot data in the public space without really affecting CD’s ability to sell their far more detailed stuff to clubs, but that’s a conversation for another day.

      August 8, 2016

Leave a Reply

Your email address will not be published. Required fields are marked *