In round five at the SCG, in front of a television audience of more than a million people, two extraordinary passages of football unfolded in quick succession.
After intercepting a wayward kick from Adelaide defender Rory Laird near the middle of the ground, Sydney’s Lance Franklin weighed up his options. Instead of opting for a teammate, he unloaded at goal from 70 metres out.
He kicked truly.
Just 30 seconds later, his side in need of a boost, Crows captain Taylor Walker took a mark in the centre square. Franklin walked towards Walker and tried to engage in conversation. Walker responded not by words, but by drilling the goal from beyond 60 metres, restoring his side’s advantage.
The two goals not only had a major impact on this tight early season game, but also provided a window onto the sport’s future.
Pace and Space in another place
In few other sports has the analytics revolution been as impactful as in American basketball’s NBA. No sport is stagnant, but change is particularly visible in one of the world’s richest and highest-pressure leagues.
The Boston Celtics’ legendary coach, general manager and president, Red Auerbach, summarised the battle:
“It is like a war. You devise offensive weapons and then the defensive players devise ways to stop it. Sometimes they never do stop it. But sometimes they do: A guy has a two-handed shot, then they find they can block it, so he goes to a one-handed shot. The offense is the inventor and the defense catches up.”
When the NBA introduced the three-point line in 1979 it was largely seen as a risky novelty. For the first decade, players rarely shot the ball from deep; and the reward for doing so was unclear. Just 3.2% of shots leaguewide were taken from beyond the arc, with the Atlanta Hawks attempting less than one per game.
By 1998, the Houston Rockets led the league with 25.6% of their shots being three-point attempts. Ten years ago, the inside-out play of Dwight Howard’s Orlando Magic reached 32.2% of their total attempts being treys, on the way to an NBA Finals appearance.
The NBA’s attritional pressure has pushed teams inexorably into using statistics to identify and exploit the highest value shots on the court: the three pointer, and the dunk/layup. Analytics tells us these two shot types return the highest score per shot. Across the league this season, three-point attempts produced 1.086 points per shot. Shots within a metre yielded 1.316 points per shot. In contrast, mid-range attempts returned just 0.8 points per shot. It took a while for this knowledge to take hold.
Of the four remaining sides in the 2018 NBA Playoffs, all are top six for average shot length. Three-point attempts are the lifeblood which took the Houston Rockets to the best record in the regular season. They became the first team to shoot more treys (50.2% of shots) than conventional two-point field goal attempts. A further 28.8% of the Rockets’ attempts from the field came from within one metre of the hoop. Most of them were layups or dunks.
The Rockets’ strategy has been dubbed MoreyBall after the franchise’s general manager Darryl Morey. It’s a philosophy that favours high value shots at the expense of the midrange dead zone. On the sidelines, the Rockets are coached by Mike D’Antoni, who shepherded the precursor of the Rockets’ MoreyBall with the Phoenix Suns in the mid-2000s. The Steve Nash-led team used a style that became known as “Seven Seconds or Less”.
Neither strategy has quite delivered a title to their respective clubs yet, but the broader movement instigated by D’Antoni, the “Pace and Space” era of the NBA, has been responsible for the rise of sides like the Warriors, Cavaliers and Spurs in recent seasons. Rather than the isolation-heavy basketball favoured by Kobe Bryant, Michael Jordan and other past heroes, the ball darts between superstars and fringe role-players alike, to find the most open shot in the best-value position. But it took about a decade to go from the stylistic flirtations of the SSOL Suns to the Splash Brothers-led Warriors and the trends currently overtaking the league.
Some critics label it “basketball by math”, but it can also be beautiful to watch. The ball pings across the court at top speed to find the best, most open shot possible.
Which brings us back to Franklin duel with Walker, and the future of footy scoring.
Franklin and The Launchpad
Franklin’s importance to the Swans has increased this year, partly because of how he enables his team to score elsewhere. Since his arrival in Sydney, Franklin has significantly increased the number of inside 50s he contributes, something which also happens to be directly correlated to the Swans’ success up forward:
Season | Sydney Points Per Inside 50 | Franklin Total Inside 50s | Franklin Average Inside 50s |
2014 | 1.64 | 85 | 3.86 |
2015 | 1.59 | 64 | 3.76 |
2016 | 1.7 | 116 | 4.46 |
2017 | 1.78 | 124 | 5.17 |
Franklin’s shift up the ground has been somewhat deliberate from the Swans, with Swans forward coach Steve Johnson seeing the zone just beyond the arc as a “launch pad” for i50 entries. “I think what’s important is that you get marks in sort of a launch pad zone – between 50 and 80 metres,” he said. “The advantage of having some players who can kick long bombs is if they mark in that sort of area and there’s nothing clear on, they might take the shot at goal.”
Franklin regularly occupies this launch pad zone beyond 50, whereas most of his fellow key forwards usually patrol the area between 30 and 50 metres from goal:
Being outside the arc so often, Franklin in 2017 was the first key forward to be in the league top 20 for inside 50s since Adam Goodes and Travis Cloke in 2011. This year, he leads the entire AFL in average inside 50s per game. Many are passes, but a fair proportion of those are shots on goal.
Despite this, Franklin’s instructions aren’t driven by a conscious philosophy.“It’s not overly stats driven,” said Johnson. “Some clubs spend more time sifting through the data analytics. But to a certain degree we certainly have an understanding of where our goals are coming from and where you can use the percentages to your advantage and sort of picking up any trends, but there’s nothing that I’m getting fed throughout a game that says we need to start targeting a certain area to take down to the forward group.”
“It’s certainly not a directive from the coaching staff to say let’s take shots from outside 50 because the percentages are higher of us kicking a goal rather than us looking for an option inside 50 and then risking there’s a chance you might not hit a target and you turn it over – that’s certainly not the case.”
Little public data exists about the value of the nature of these long scoring shots, although one source has been Port Adelaide’s new Performance Data Scientist Rob Younger (disclaimer – Younger is a fellow co-author of Footballistics, which is out June 18 and can now be pre-ordered via the ABC Shop).
In a piece for his website FiguringFooty written in April last year, Younger revealed the two sides scoring the most goals outside 50 were Sydney and Adelaide. The dominant forwards contributing the bulk of these scores were Franklin and Walker. Those two sides have also been two of the more efficient up forward this season, despite an overall drop in scoring rate across the league. The long-term average for points per inside 50 since 1999 has been about 1.76 points per entry inside 50, but fell to 1.69 points between 2013 and 2017. This season it’s currently at 1.58 points.
Despite that league-wide decline, the Swans have converted at 1.69 points per inside 50, ranking them 5th so far in 2018. Still higher is Adelaide, which currently converts 1.78 points per inside 50, good for 2nd in the league.
But there’s no nine-point goal in the AFL
The most obvious objection to applying the logic behind the NBA’s three-point revolution to the AFL is that there’s no nine-point goal in footy. While true, “points per shot” can translate another way, using points per inside 50 as our measure.
Unlike in basketball, not all AFL possessions give an opportunity to score, and many possession chains end in turnovers. All scores, however, must be an inside 50 first, which gives a site at which to measure attacks. Inside 50s can occur in different ways, and each type of inside 50 has a score probability.
Direct shots at goal from outside the arc, like Walker’s and Franklin’s, make up about 12% of inside 50 entries.
According to Younger’s 2017 piece, shots outside 50 are converted at just below 50% (we’ve therefore estimated around 45%), yielding approximately 3.25 points per entry.
Marks inside 50 are high value. In 2017, about 23% of inside 50 entries resulted in a mark inside 50. Close range set shots are a near certainty, goals coming about 95% of the time within 20 metres, according to shot value charts published by Younger. Not all marks inside 50 are this close however, and many come in lower value areas, such as in the pockets. Generally, the higher value areas of the forward 50 are defended more fiercely, so a fair discounted accuracy of these shots would be 65%, which means a marked inside 50 entry averages about 4.5 points. (More information at bottom).
General play inside 50s are the leftovers, constituting all entries not marked or reaching the goals. They’re the most common outcome, about 75% of all inside 50s. But what happens to these general entries?
About 68% of all inside 50s became rebound 50s in 2017, repelled either via turnovers, opposition clearance, or a behind kick-in. The difference between inside 50s and rebound 50s can be found by subtracting goals from the inside 50 count, and accounting for quarters that end with an inside 50 without a score or rebound (about 0.7 per game).
Using the data above for marks and long shots, and a season average of 12 goals and 11 behinds per team per game, we can estimate that about 7.7 goals and 9.2 behinds are kicked per 100 general play inside 50s. This means an inside 50 which isn’t marked, and isn’t a shot at goal, delivers around 0.55 points per entry on average.
These low-quality inside 50s are the mid-range jump shots of the AFL.
All in all, direct shots outside 50 average about 3.25 points per inside 50 based on the 2017 data. All other shots average about 1.54 points per inside 50. That 1.54 points per inside 50 would be good for the second worst mark in the league in 2017, fractionally ahead of Fremantle, and 0.32 points per inside 50 behind league leaders Adelaide.
“(Long range shooting) does stop you from having to find another mark or another connection inside your forward 50, which when you’re doing that increases the risk of turning it over again,” said the Crows’ forward-line coach Josh Francou. “If there’s a free player, if he can find a connection inside 35 that is more likely to kick the goal than he is, then you’d want him to take the shorter one. But if you have a Paul Seedsman and he’s running from outside 50 and he can back himself to kick the goal, if there are no glaring options to kick to inside 50, instead of increasing your chances of turning it over and trying to get another connection, then have a go yourself.”
Before joining the Crows this past off-season, Francou spent the past four seasons as an assistant to Swans coach John Longmire. He says although he’s given his forwards a green light to shoot from beyond 50, they need to be judicious. “I guess it depends on the individual bit of play,” he said. “I mean if there’s a free player inside your forward line, which probably doesn’t happen a lot in the AFL, and he can increase the chance of hitting the scoreboard with a goal then you’d want him to take the shorter one.”
The Side Effects from The Other Tex
Walker’s shift up the ground has provided an incidental benefit for the Crows, with other goals created by their power forward. Last year was the first time since the statistic was first tracked that a key forward led the league in goal assists, and like Franklin he has rocketed up the inside 50 charts as well.
Which opens another weapon – the use of a key forward to create space for the other forwards. We wrote earlier this year on the Kangaroos’ ploy of playing three forwards to create space in the forward line to operate, but having a genuine goal threat at long range also recreates defensive structures as well.
“So, if you’ve got a Taylor Walker leading up to the kicker, I guess invariably he’s going to be further away from goal when he receives it,” said Francou. “So those times when he is 50 to 55 out, having the ability to be able to kick it from there consistently and with reasonable accuracy is a good weapon to have, absolutely.”
Most sides deploy their best defender in a one-on-one sense to the best opposition forward. Usually that defender is asked to help in other contests, to turn a mark inside 50 into a general play inside 50. But if the dominant key forward doesn’t operate inside 50, or near the other forwards, the defence loses this weapon. As long as the threat exists and remains credible, the attack disables a key defensive weapon, to borrow Auerbach’s phrasing.
This expansion of the area that must be defended should increase the points per inside 50 return of non-shot entries, which may create a domino effect. The offence almost always moves first when it comes to innovation.
To most neutral observers, the Crows play the most aesthetically-pleasing version of modern footy. They are doing it partly by becoming a leading side at playing the percentages game. Francou’s words comparing the probability of long shots versus shorter connection attempts suggest a style which is heavily influenced by analytical thought.
It also drives at the heart of poorly thought out punditry about the malign influence of statistics on sport. Auerbach’s theory of offense-driven development seems to hold true here. Nobody is going to put work into stopping a long bomb-focused game style until one is fully developed. If data and analytical thinking are now driving forward coaching, then they are pushing first and foremost to make football better to watch.
So far, we probably haven’t seen players have anywhere near full license to test this theory, because traditionally such play is stigmatised as selfish and lacking vision. Taking a shot that you’re less than 50% likely to make seems intuitively risky, but when weighed against the scoring probability of the alternative option inside 50, it should often be seen as the better option. There’s certainly more players in the league who can kick a 55 or 60 metre goal than just Franklin and Walker. Over time, we expect to see more sides observe the Crows and Swans use of Franklin, Walker, Seedsman and others, and shift their decision-making accordingly.
But there is always a counter
In the NBA there are already example of teams moving away from the strict three and dunk methodology developed in the “Pace and Space” era. The side the has arguably benefited the most from the analytical push towards bombing from deep or scoring under the hoop, the Warriors, are shooting more mid and long-range twos than any other time in their four year run at the top. It hasn’t shifted back to the levels of a decade ago, but they are exploiting a new area of attack.
The Warriors are also the most effective in the league at shots from more than 10 feet, but inside the three-point line. This is almost certainly linked to Golden State also being the most effective side from beyond the three-point line as well. It is a sign how the defence can overcompensate in its attempt to shut down the new weapon, and inadvertently open a new path of attack. The “MoreyBall “Rockets are the second-most effective team from both these areas.
If the long bomb trend takes off in the AFL, expect a similar bounce back to mid-range goals after the establishment of the trend. Around the arc and beyond, if clubs do kick from deep regularly, defenders are going to need to cover long kicks more closely. They’ll also need to be wary of players blocking for long-kicking teammates. Focus will have to shift to lateral kicks and recycling backwards to shift defenders, and the Franklin-style handball receives. All these adjustments should pull defenders away from other, formerly favoured areas and open these scoring options back up.
This is just guesswork and analogy at this stage, because the counters haven’t really developed yet. They haven’t needed to, beyond an awareness of Franklin’s left boot. However, if the trend we’ve observed at Adelaide and Sydney for increasing forward efficiency with long-range goal kicking takes hold and expands, we can expect to see new footballing trends, developed from both local and stolen ideas.
We’re a long way from the full “Pace and Space” era of the AFL, and we’re even further from its counters. However, the potential is there for AFL teams to exploit the surprisingly high scoring probability of long bomb shots relative to basic inside 50 kicks, much more than they currently do.
The interviews for this piece were conducted by James Coventry, author of Time and Space and the upcoming book Footballistics. James’ take on this trend will be coming out on Monday on the ABC website, keep watch for more info.
Many thanks to Rob Younger for introducing this concept to us via his work on FiguringFooty.
Note: Not all marks inside 50 come from outside 50 entries, and there are occasionally two marks inside 50 per score, however this is offset by the absence of free kicks, 50m penalties and other rare set shots inside 50 in this estimate. While we can’t show our full working, we are comfortable that the conversion rate is at least roughly accurate.