It’s amazing who you get sat next to at the footy when you go as a journalist. I tipped up to Shark Park a few weeks back to watch them face Manly and found myself next to Marcel Noronha, one of the analysts at the Cronulla Sharks. I couldn’t have been better placed.
The first half was somewhat distracted by Siosifa Talakai running rings around Morgan Harper, but after the break, I noticed what my neighbour was doing. He was taking a tally chart.
Now, being the sort of journalist who spends their days diving through spreadsheets of player stats, I had assumed that data gathering was a precise, techy kind of business, so I was somewhat taken aback to see it being done with nothing more complicated than a pen and paper.
Naturally, I had to lean over and ask Marcel what he was counting. “Push supports” was the reply. I had to admit that, even as a stats and tactics guy, I wasn’t entirely familiar with the concept.
Well, let’s rephrase that. I was aware of the concept, but not that it was a countable thing, or indeed, that anyone was counting it. After a period of feeling quite stupid, I pinged into Google ”push supports rugby league” and got…very little. Nice to know I’m not the only one.
What you get is a few comments on forums and, crucially, a few choice quotes given from Steve Price, then of Warrington and now an assistant at Cronulla, following a victory in April of last year.
That game was three weeks ago. I’ve watched, I think, about 20 rugby league matches since and push supports have been front of mind throughout. They might be the most important stat that nobody has heard of.
First, a definition: a push support is related to the decoy runner and the support runner, except they might get the ball and sometimes do. It’s someone running through the line, like a decoy does, but also the player outside giving an option and the player running the line alongside the ball carrier threatening the tip on.
It’s a good counterpart to line engagements, which measure how much a player takes the ball to the line and makes a defender commit, except with push supports, the thing being measured is players without the ball.
It’s players in motion, doing anything that distracts or engages a defender, whether moving their hips, making them make a decision or simply keeping them honest.
How often have you heard a coach say ‘we have to be accountable for X’ or ‘it’s those little effort areas’? Think of the converse in attack: how do you make a team have to be accountable for someone? How do you make them make those efforts? That’s what push supports do.
Take this Sharks try against Manly: there are five push supports in the play before Talakai even makes the break.
It’s a stat that clubs measure and that is provided by the NRL to them in the comprehensive platforms that the clubs pay for, and their data does often vary – as anyone who compares NRL.com’s stats and Fox Sports Lab will know, this is an occupational hazard across the whole business.
They also don’t count in real time, which is why Marcel is sat in the media box at Shark Park with a pen and paper, so that he can give that data to the head analyst at halftime, who in turn gives it to Craig Fitzgibbon.
The whole experience got me thinking. Cronulla are one of the highest-ranking teams in the comp in crucial attacking statistics: tries scored, of course, in which they are third, but they are first for tackle breaks and general play passes (GPP) and second for line breaks and line engagements.
But they’re also not particularly highly ranked in other things that you might think correlate with attacking success: tackles in opposing 20m (tenth), attacking kicks (tenth), forced drop outs (11th) and dead last for dummy half runs.
(Fun fact: the best attacking side, Melbourne, are top for scoots from acting half, while the second-best, Cronulla, are last. There’s many ways to skin a cat.)
Reading those stats and combining them with the eye-test you get from attending a lot of games in person, what I surmise is that Cronulla and Melbourne are really, really good at creating deception and thus score from further out.
For individual players, the results bear that out. In the top ten players in the comp for line breaks, there are two Sharks (Ronaldo Mulitalo, first, and Sione Katoa, third) and four Storm players (Cameron Munster, Ryan Papenhuyzen, Nick Meaney and Xavier Coates).
For line break assists, Will Kennedy is top and Nicho Hynes is second, with Munster in third and Harry Grant and Papenhuyzen not far behind.
One of my colleagues in the media box, Adam Pengilly of The Sydney Morning Herald, has written about this for the Sharks in terms of one specific move they like to unveil, known as ‘the slingshot’.
In that, the Sharks start narrow and end wide, with Kennedy coming into the line and the attacking gradually broadening to the point where Katoa or Mulitalo score at the corner post.
Undeniably, they do this, but it only tells half the story. Cronulla’s men in motion make this possible, and (as Adam acknowledged in the piece) lots of teams try to do this. The difference, as far as I can see it, is that when Cronulla do it, defences are kept honest and stopped from immediately dropping and sliding by the men in motion.
Nicho Hynes explained it as something that they focussed on specifically because they didn’t have a massive forward pack who could dominate the middle through size, and instead, sought to create individual battles.
“It’s a big style of our play,” said the halfback. “We have agile forwards in Cam McInnes, Toby Rudolf, Jack Williams and Dale Finucane to name a few.
“We’re not the biggest pack but we’re one of the most agile, so you push into each other and take flies off the defensive line and create one-on-ones to get fast play-the-balls, which our spine can play off.
“The Storm do it very well, and they’re one of the best in the competition at it. It’s just what we have been gifted with man mountain players, so we have to be good at it. We have to do it every week.”
It works superbly for the Sharks. Melbourne missed 25 per cent more tackles in their game against Cronulla than they do on average, while Manly, who miss on average 27.1 per game, missed 23 in a half. The Tigers average 27 too, but missed 41 against the Sharks.
The question that came to mind through this is not simply if it was the secret to the Sharks’ success, but how replicable it was across the league.
If, as Marcel said, it was something that everyone counted, was it something that other teams were not concentrating on to the same extent?
With that in mind, I performed a little experiment. I was going to sit there, with a pen and paper, just like Marcel, and count.
The next weekend, one of the games I attended was perfect: the Bulldogs, a notably terrible attacking team, against the Roosters, a non-terrible attacking team but one that had been misfiring.
Then, this weekend, another opportunity: South Sydney, a team that (using my eye-test and data) I think are better than they are performing in attack, versus Brisbane Broncos, a team that perhaps over-perform their underlying data in terms of try scoring.
Some ground rules: Marcel kindly shared with me the totals from both halves of the Sharks-Storm game in Round 6, so we have those numbers.
For the Bulldogs, Roosters, Souths and Broncos, I took just the first half – when, theoretically at least, they are better at implementing a game plan free of scoreboard pressure – and then equalised across all halves for play the balls to ensure similarity of sample size.
Then I counted them manually, as Marcel did, to get the number of push supports. From that, I created a metric that I’m going to call this the Deception Score (DS), which is simply the number of push supports per play-the-ball, so that we can accurately compare half to half while accounting for a wide range of total plays per half.
The Deception Score is simple: if there are 75 push supports and 75 play-the-balls in a half, the DS would be 100, but if we assume that two or three plays will be simple hit-ups early in a set, a score of 50 would be about par and anything above that quite good.
The Sharks, with 75 push supports off 77 play-the-balls in their first half against Melbourne, were a DS of 97, while in the second half, they were actually better despite the poor possession and got a score of 109 – indicating that they actually did more with the football that they got. This was reflected in (relatively to amount of possession) more line breaks and tries in the second half.
Let’s go through the results: Souths were the best, with 39 push supports from 66 play the balls for a DS of 60, followed by the Roosters on 56, Brisbane on 52 and the Bulldogs on 34. You can insert your own joke about inventing a new attacking metric for the Dogs to be bad at.
Watching all three halves back with a real focus on deception, some things became very, very clear.
Firstly, the Sharks are exceptional at creating attacking deception.
I watched their first half with Melbourne after generating my own numbers, in the assumption that I was counting incorrectly because of the discrepancy between my count of the other two games and Marcel’s. In fact, I think Marcel might have undercounted. That’s how good they were.
Even in simple hit-ups, the Sharks had several men running alongside the ball carrier, creating accountability. In turn, they had multiple occasions of passing from forward to forward, close to the line, because a support runner was there to be passed to.
Watching that first half with the Storm back, Greg Alexander even mentions it on commentary: “They squeeze plenty of footy into every set, there’s no dead plays, there’s players moving off the ball,” he says.
The average set metres back this up: Cronulla hit 45m – away at Melbourne, remember – compared to 40m from the Roosters and Broncos, 35m from the Bulldogs and just 33m from the Bunnies.
It shouldn’t really be surprising that Cronulla are really good at push supports: Craig Fitzgibbon comes from the Roosters system, who have historically been very high in the similar, but not identical ‘supports’ metric that the NRL produces, and Steve Price, their assistant coach, is the only person quoted about push supports anywhere in a real article on the internet.
South Sydney are an interesting case, too. Their attack has actually been excellent this year – top three for line breaks, tackles inside 20 and line engagements – but they just drop the ball too much. In the half I studied, they bombed three tries. The NRL supports metric actually has them top for 2022.
The Broncos are in the middle of the pack, though it should be mentioned that, early on against Souths, they adopted a policy of going wide early – with lots of men in motion – before winding back later in the half and playing more conservatively.
They also completed poorly, just 69 per cent, and had just eight inside 20m play-the-balls, limiting the amount of good ball, shape-driven play that they could make. It might be inferred that in other situations, they might have done more with the ball given the opportunity.
That brings us to the Bulldogs. It’s not a surprise that they’re bad at attacking at this stage, because it’s been done to death that they had the highest completion rate but scored the fewest tries in the NRL last year. Watching back, however, it’s clear that they really don’t bother that much with trying to make the opposition think.
They had five sets with zero push supports at all – no other side went more than one set without any – and even managed several with one or none when in good ball. In the half studied, they had 41 opposition half play-the-balls and 25 inside 20m play-the-balls, but failed to ask any tough questions of the Roosters’ defence.
As Cronulla have an agile pack, so the Dogs have a big, slow one. That’s not necessarily bad – Tevita Pangai Junior, Paul Vaughan and Luke Thompson are all good at what they do – but it does look quite exposed in this model.
If you were wondering why their halfbacks never get any good ball, the fact that the defence always knows exactly where the footy is going might go a long way to explaining it.
The study in itself was an interesting exercise in why rugby league needs an xG model – discussed at length here – because Souths scored one try and created four, whereas the Dogs scored two and created next to nothing. There was no question about who attacked better.
Naturally, in the second half of both games, the Bunnies went further behind while chasing the game and dropping the ball, while the Dogs completed even higher and defended for their lives to win the game.
The push supports metric isn’t a magic bullet stat – such things don’t exist – but if you’ve ever wondered why certain teams are really good at isolating defenders and creating play, it’s as good a measurement as any. It should probably be more widely known.
Certainly, in a world of the VB Hard Earned Index and the Telstra Tracker, it can go a lot further to explaining the whys of attacking. Cronulla are a good example of it: you just have to sit next to the right person to learn why.