The Art of Goalkicking

A Quantitative Analysis - Part 1

Goal kicking is one of the most important skills a forward can possess and can often be the critical difference between winning and losing. In fact, since the first official game in 1897, the more accurate team has prevailed 64.7% of the time. Any semi-serious fan can probably recall an important match where their team squandered several game-winning (and very gettable) opportunities en-route to a painful defeat. Given the importance of goalkicking, and the stark difference between success and failure (six points vs. one or zero points), one would expect that professional footballers, in particular forwards, would be capable of exceptional accuracy. Oddly enough, this appears not to be true. Whilst professional footballers are certainly capable of kicking further than the average person, they do not obviously appear more accurate. Youtube is replete with videos like this, showing countless examples of shocking misses by modern players. Media pundits often lament current goal kicking standards, claiming it is the one area of the game that hasn't improved over time. Some even claim that goal-kicking has gotten worse over time, with increased fatigue cited as a possible reason. Take this clip for instance, where Eddie McGuire dramatically laments the recent plague of poor goalkicking that is overtaking the league. After repeatedly interrupting his much more qualified co-hosts Ross Lyon and Matthew Lloyd, he condescendingly goes on to suggest that modern inaccuracy may be due to lack of practice. What evidence does he provide to support this assertion? Well, he heard that former player and current commentator Brian Taylor kicked nine from ten at the park with his kids recently! Leaving aside the blatant cherry-picking of bad kicks for goal, which presumably could've been done for any era of the game, the segment raises an interesting and testable question: Has goalkicking accuracy gotten worse over time?

Before answering, I'll first lay out of the limitations of the data available: I'll be using scoring shot data, i.e. goals and behinds, as recorded by the ever-reliable AFL Tables website. This includes rushed behinds, which aren't even kicks, and excludes shots for goal that were out on the full or didn't make the distance. Why? More detailed scoring shot data for the AFL simply isn't available before 2014 and this recent window isn't large enough to test the claim that goal-kicking hasn't improved. So we'll have to proceed with imperfect data, but over a large enough sample, the quirks of the data should average themselves out and the underlying trends (if there are any) should become apparent.

Decade	Games	Average Goals	Average Behinds	Accuracy
1890	201	5.1	8.1	38.8%
1900	756	6.7	9.5	41.5%
1910	757	7.9	11.0	41.6%
1920	935	10.0	12.1	45.2%
1930	1120	12.3	13.4	47.9%
1940	1102	12.2	13.5	47.3%
1950	1126	10.5	12.4	45.9%
1960	1145	10.9	12.2	47.1%
1970	1378	13.7	13.8	49.9%
1980	1446	14.8	13.9	51.6%
1990	1761	14.1	12.7	52.6%
2000	1850	13.9	11.9	53.8%
2010	2037	12.8	11.4	52.8%
2020	414	10.5	9.5	52.5%

In this instance, the data belies the common belief. Goalkicking accuracy has actually improved steadily over time from 38% in the 1890's to 52.5% today and this conclusion has been written about elsewhere . Accuracy actually peaked in the 2000's at 53.8% (which coincides with the opening of the indoor stadium at Docklands), perhaps adding some weight to the fatigue theory, but the decline to 2020's is only 1.3 percentage points - hardly the shocking affair it is made out to be. So we can confidently call BS on Eddie McGuire's sensationalistic rant.

The adjacent claim implicit in Eddie's rant is that the specialist forwards of yesteryear (in this instance Brian Taylor) were better kicks for goal than modern forwards. Certainly spending the whole game in the vicinity of the goal square meant they got more opportunities, but defensive tactics were also different - defenders typically played one-on-one, and loose players in defence were rare. This claim seems more plausible to me, and piques my curiosity about quantifying goalkicking ability. I have previously attempted to answer who is the greatest player of all time, using RAPM - an approach common in basketball analysis. Whilst interesting, that piece was ultimately unsatisfying as the results didn't really accord with common sense for reasons I discussed in the post. Here, I'll attempt to identify who the best kick for goal of all time is, and hopefully this problem proves easier to solve.

Again, I'll start with a short note on the data. Behinds weren't credited to individual players until 1965, so we will be unable to assess the sharp-shooting of some of the great forwards, such as Gordan Coventry, who kicked 1299 goals in the 1920's and 1930's and Jack Titus who kicked 970 goals over a similar period. Other careers, such as that of Doug Wade (1057 goals, 1961-75)) are partially covered. But with the exception of a handful of games, we have complete coverage from 1965 onwards of how many goals and behinds each player kicked and can therefore pass judgement on their kicking ability. So, who is the best kick?

Well, it turns out this is not necessarily a straightforward question to answer. For example, Darcy Fort, the Brisbane Lions backup ruckman, has maintained a perfect record in front of goal (as of R6 2022), although he's only had 11 shots in his short career to date. Nevertheless that puts him in at the top of the pile as no player, past or present has maintained perfect accuracy over a larger sample. While he's probably a better kick than average, it's also likely that his spectacular accuracy will revert to something more normal as time goes on. AFL Tables ranks sharp-shooters on accuracy and uses a minimum threshold of 50 scoring shots, to avoid the problem of small samples. According to this criteria, little-known journeyman Michael Murphy is the most accurate kick of all time, registering 63.19 across a short 29 game career between 1988 -1993. This still feels very unsatisfying though. The 50 shot minimum is arbitrary and seems too low. More importantly, looking at pure accuracy doesn't account for sample size. It's much easier to maintain a high percentage over smaller samples, as Darcy Fort has proven. Personally I find second-placed Tory Dickson's 74.8% accuracy over 242 shots more impressive than Murphy's 76.8% accuracy over 82 shots. Peter Jones (71.5% from 397 shots) and Tony Lockett (69.7% from 1950 shots) are also strong contenders for the title of best kick.

The binomial probability distribution is a useful tool at this point, because it allows us to quantify kicking skill in a more objective and rigorous way. If we start with the naïve (and wrong) assumption that all goal-kickers are average, we can reframe the question to: "How likely is it that an average kicker could attain this record?" In other words, could Darcy Fort just be lucky, and how likely exactly would that be? The answer to this question will give us a probability, known in statistics as a p-value, which can be used to rank goalkicking skill across players with varying sample sizes. If Darcy Fort is in fact an average kick for goal (average accuracy since 1897 is 56.2%), the chance he would successfully convert his first eleven shots at goal is 0.18% or about 1 in 563. That is to say, very unlikely, but not impossible. Given that thousands of players have played the game since 1965, it wouldn't be surprising if one or more of them had been the beneficiary of such luck. What about Tory Dickson, with 74.8% accuracy over 242 shots? This is much stronger statistical evidence of kicking skill. The cumulative binomial probability distribution tells us there is only a 1 in 633 million that Dickson's record is lucky. We can extend this analysis to every player who has played since 1965, and rank the players by the improbability of their record, assuming uniformly average ability. According to this method, the most accurate kicks for goal of all time are:

Player	Goals	Behinds	Accuracy	P-value
Tony Lockett	1360	590	69.7%	1.03E-34
Matthew Lloyd	926	424	68.6%	8.37E-21
Jason Dunstall	1254	641	66.2%	6.08E-19
Peter Hudson	714	330	68.4%	5.39E-16
Peter McKenna	874	470	65.0%	2.96E-11
Luke Breust	451	205	68.6%	3.24E-11
Michael Roach	607	306	66.5%	1.55E-10
Daniel Bradshaw	524	255	67.3%	1.92E-10
Peter Jones	281	113	71.3%	4.81E-10
Tory Dickson	181	61	74.7%	1.58E-9
Saverio Rocca	748	411	64.5%	5.03E-9
Tony Modra	588	308	65.6%	5.96E-9
Eddie Betts	640	342	65.1%	6.79E-9
Fraser Gehrig	549	285	65.8%	9.65E-9
Josh Kennedy	695	382	64.5%	1.74E-8
Barry Hall	746	421	63.9%	5.24E-8
Tom Hawkins	681	381	64.1%	9.75E-8
Jack Darling	454	234	66.0%	1.09E-7
Paul Salmon	561	303	64.9%	1.14E-7

There we go! This list looks very good. Certainly cleaner results than the RAPM analysis yielded. Tony Lockett ranks as the best kick for goal (technically speaking, the least likely to be average) maintaining a nearly 70% accuracy rate over almost 2000 shots! Note: If you're wondering what the strange-looking numbers are in the p-value column, they are scientific notation, which is a convenient way to write very small and very large numbers. The negative number following the "E" tells you have many zeros after the decimal point the number has. Suffice it to say, the chance of Tony Lockett's record being the product of pure chance, is infinitesimally small. Less than one chance in trillions. A interesting aspect about Lockett's goalkicking is how much value it provided his team. Given that he converted at 69.7%, a full 13.5 percentage points better than average and he averaged 6.9 shots a game, his accuracy was responsible for approximately four additional points of value. This may not sound like much, but when you consider that even the best players in the league today are often not worth much more than a goal to their team (a fact which can be discerned by observing betting odds movement when they are unexpectedly announced out of their team), four points of value from pure sharpshooting alone is remarkable!

Matthew Lloyd, the one person in the Eddie McGuire clip who talks sensibly about goal-kicking, ranks second best kicking 926 goals at 68.6%. Lloyd did benefit from playing 76 of his 270 games at Docklands, which is indoors and protected from the elements. If anything, this confirms Lockett's greatness. Luke Breust, currently sixth, is the highest-ranked active player and this makes a lot of sense to anyone who watched him in 2014 when he recorded an impressive 57.12 during a premiership year for Hawthorn.

Noted sharp-shooter, and member of the 70%-club, Tory Dickson sits tenth and would've sat much higher if he could've maintained his staggering 74.7% accuracy over a larger sample. Like Lloyd, he was fortunate to play frequently at Docklands, but unlike Lloyd, and unfortunately for Tory, he wasn't able to generate enough shots per game to be considered a great forward. He averaged just 2.1 shots per game compared to the other great sharp-shooters, such as Lockett 6.9, Lloyd 5.0, Dunstall 7.0. Nevertheless, Tory has been lauded for his goalkicking accuracy, and rightly so.

The current West Coast forward-line duo of Josh Kennedy and Jack Darling both make the top-20. Coupled with the fact that they also generate a good number of shots per game, at least by modern standards, (Kennedy 3.8 and Darling 2.9), they are in my opinion an underappreciated combination and responsible in large part for West Coast's recent success, 2022 notwithstanding. More on this later...

You might be curious about the other end of the scale. Who are the worst kicks for goal? Whilst the best kicks for goal make a lot of sense, the worst kicks will likely contain surprises. If you are really bad, you will struggle to hold your place in the team and will thus be unable to accumulate the sample size required to statistically prove your inadequacy. Coaches, fans and commentators don't need to see 500 shanked kicked before realising you're a dud, so there will be an unusual survivorship dynamic at play here, which will impact the results.

Player	Career	Goals	Behinds	Accuracy	P-Value
Alan Martello	1970-83	165	250	39.8%	1.19E-11
Geoff Ablett	1973-85	153	234	39.5%	3.15E-11
Stuart Magee	1962-75	117	178	39.7%	8.06E-9
Greg Wells	1969-82	270	331	44.9%	1.75E-8
Warwick Irwin	1970-83	243	304	44.4%	1.99E-08
Glenn Elliott	1969-79	76	126	37.6%	8.16E-8
Alan Johnson	1982-90	95	146	39.4%	1.17E-7
Kevin Bartlett	1965-83	769	777	49.7%	1.77E-7
Des Meagher	1966-1976	96	144	40.0%	3.21E-7
Robert Neal	1974-88	51	93	35.4%	4.02E-7

Most of the worst ten sharpshooters are names unfamiliar to me, which is most likely due to the fact that they all played in the 60's, 70's and 80's. However two of them jump off the page. Geoff Ablett, played 229 games as a speedy winger for Hawthorn during a time when the Hawks were the most dominant club in Victoria. Most notably however he is the older brother of Hall-of-Game inductee Gary Ablett Snr, who booted 1031 goals at a very respectable, but not elite conversion rate of 60%. The other familiar name is of course Kevin Bartlett, Richmond superstar who played 403 games for the Tigers and kicked 778 goals (slightly different numbers to the above table, due to the fact that Bartlett played one or two games where goals and behinds were not accurately attributed). Although primarily a rover, Bartlett spent considerable time as a half-forward, particularly in the latter half of his career. According to his Wikipedia page, he actually developed a reputation for not passing when within range of goal, which may explain why despite taking over 1500 shots in his career (sixth most of all time) he converted at a very poor rate of 49.7%. Had he possessed the accuracy of Lockett or Dunstall he would've surpassed 1000 goals.

A final point to make, which I alluded to earlier, is that of course not all shots are equally easy and some players good (bad) accuracy on paper, may in fact indicate they typically take high (low) percentage shots, rather than that they are accurate kicks. Shots from 30 metres out directly in front are far easier than those 50 metres out, on the boundary. Similarly, kicking for goal in windy and / or wet conditions will be far harder than dry and players who player more often in those conditions will see their accuracy numbers suffer, compared to those blessed to play frequently indoors at Docklands for instance. Think Matthew Lloyd, Tory Dickson, etc. I plan on doing a second blog post examining modern goalkicking accuracy in more granular detail soon. In particular, I want to account for the degree of difficult of individual kicks and aggregate this up to see more definitively which players are accurate and which players are just getting easy shots. Unfortunately the data required for this analysis only exists from 2014 onwards (at least as far as I can tell), so the insights gained will be limited to very recent past.

---