Discipline update by team and referees, plus some Origin selection thoughts

There’s no team specific deep dive this on the Eye Test this week. Firstly, because I can’t put out one of those posts every week of the season or I’ll lose my mind. Secondly, I like to look at some bigger picture trends every now and then to make sure I’m not missing something.

Thirdly, I’ve had countless people DM or email me asking about penalty counts and referees, especially regarding this years “crackdown” on hip drops. Hopefully this will get it out of your system. And lastly, all anyone wants to do is talk about Origin selections, so I’ll double down and give you some of that content later on as well.

I’ve got a nice bonus referee chart in the middle of this post for those of the real stat sickos who stay with me past 2,000 words every week. I think this one is worth it, and will either prove or disprove whatever you think it does.

Before we get into that, I’d like everyone to check out Liam’s 2024 NRL rivalry survey results at the Maroon Observer. It’s a good insight into which teams fans consider rivals of their own team, and not just through a manufactured media lens. Everyone should be subscribed to Liam’s newsletter and support quality independent rugby league content.

Back to this weeks analysis, for the 2024 season we’re seeing an average of 15.8 total penalties (set restarts + traditional penalties) called per match through 12 rounds, almost flat on last years 15.9 average. However, the composition of them has changed slightly. All numbers are from Fox Sports Stats.

Last season we had 11.4 traditional penalties per game, which was dropped to 10.4 this year. Of course, this means the slack is picked up by set restarts, which have increased by 1.0 to 5.4.

The increase in set restarts is intriguing, since they had leveled off in the 4.5-4.7 range for the prior two seasons after the peak insanity of 2021 where there were 7.5 per contest.

My gut feeling is that some of the slight increase is due to leg pulls whilst a tackled player is on the ground being targeted more by referees, but the data I have access to doesn’t give me that sort of granularity. My other thought is that teams are encroaching on 10 metres more often.

Thank you for reading The Rugby League Eye Test. Subscribe below to receive new posts in your inbox

I can see that there’s a split between inside 10 metres and ruck infringements of 32/68, which is higher than 2023 (26/74) but in line with 2022 (32/68) through 10 rounds, which would indicate there’s been some more restarts awarded for being offside. That’s not specifying whether the ruck infringements are for slow peels, hands on the ball or leg pulls though for those type of restarts. If either of the data providers want to slide into my DMs to help me out, feel free.

If we break these numbers down into halves, the split between periods is still very similar – about 60% for first half set restarts and 53% for first half penalties. This continues with the analysis I did a few seasons ago where set restarts were far less likely to be called later into games once referees had established their limits.

The average tackle number of a set restart is 2.37, no change from the 2.3 I uncovered previously. As I said then, it’s not six again, it’s 2.3 again and most clubs would back themselves to defend and extra 2 or 3 play the balls.

Another thing I also like to look at is track how many play the balls occur per penalty awarded, to get an idea of how the game flows between transgressions. During the 2021 peak of set restart hysteria, that number was down to as low as 13-14 play the balls per infringement some rounds. Thankfully we’ve moved on from that but are back to around 18 play the balls or three full sets of six per penalty, as seen below.

Interestingly, 2018 with the penalty crackdown that broke the crisis merchants was the lowest in recent years, with one penalty (which were trad only at that time) every 16.1 play the balls.

If we look specifically by round this season, event that average of 18.8 hasn’t been completely consistent.

Round 7 peaked at 24.1 play the balls per penalty, which also had one of the largest average margins of victory in a while. It dropped as low as 14.7 play the balls in Round 9, when we had several very close games.

Now we’ve established how many set restarts and traditional penalties are being called, are being called per game, who is awarding them? Buckle up friends. Below is the average number of set restarts and traditional penalties called per referee this season.

Of the 12 referees used this season up to Round 12, Liam Kennedy has been the shrewdest with the whistle. On average he blows about 13 infringement a game and has the lowest average of set restarts per match at just 4.0.

At the other end of the scale we have Todd Smith, who absolutely blows the pea out the whistle compared to Kennedy. Smith averages 7.8 restarts per game, nearly twice as many as Kennedy, and another 11.2 traditional penalties per game for a total of nineteen.

Only Peter Gough (11.3) and Adam Gee (11.8) award more traditional penalties per match. And only Kasey Badger hands out more set restarts, at 8.67 per game from her three appearances as lead official. For the least traditional penalties per game, rookie Wyatt Raymond takes first there at just 7.6 per match.

If you’re enjoying the posts on this site and want to support independent rugby league content, please consider donating to The Rugby League Eye Test with a value of your choosing via the link below.


As a long time set restart stat sicko, I’m somewhat saddened by the decline of King Gee. No one loved a six again call than Adam Gee from 2020 to 2022, especially in the first half where he would usually dole out three or four within the first ten minutes. Even in 2023 he was still calling them more than most. Now he’s awarding the sixth fewest set restarts in the competition.

Breaking these down by half doesn’t show much other than Todd Smith loving a first half whistle and Kasey Badger being the only referee with more second half calls than first half.

We’ve looked at penalties awarded, but which referees have the best and worst challenge percentages this season?

Ashley Klein has the lowest challenge success rate at nearly 27%, which will probably annoy many fans with terrible recency biases. Peter Gough and Wyatt Ramond, the worst, both nearing 60%. Some of these numbers are skewed by terrible choices of what to challenge, such as anything by a prop or any Parramatta Eel in 2023, as well as small sample sizes for the likes of Sharpe, Raymond and Badger.

Next, we have an update of a chart I’ve used a few times before this season, splitting total infringements into conceded and awarded by team for the season. The results are below.

Canberra receive the most penalties this season at 9.4 per contest, and only give up 7.5 for a net of +1.9, best in the NRL. The Dolphins are given the least favourable treatment with just 6.5 infringements going their way, and they concede 7.5 themselves for a net of -1.0.

If you look at this data by net total penalties, you’ll notice that most of the top eight are sitting in the negative. As much as we’ve weeded out some of the more cynical set restarts and penalties, they’re still a part of the game as the better defensive sides in the competition are happy push the boundaries and trust their defense to hold under repeat sets.

The most penalised team? Canterbury, with a net of -1.9, thanks to 9.2 infringements conceded and 7.3 awarded.

For set restarts only, it’s a slightly different chart but mostly in the middle.

Canberra are still first, receiving 3.7 set restarts per game and conceding 2.6. South Sydney and Penrith are the only other teams to be awarded more than three set restarts per game.

Again, Canterbury are at the bottom of this list, giving up 3.5 set restarts per game but only receiving 1.9 themselves, resulting in a net of -1.6. That’s a significant change last year where they were awarded 2.3 and conceded 2.2 for a net of +0.1. Given their defensive improvement, you’ll see why the good teams aren’t concerned by conceding a few extra sets.

Looking at traditional penalties, Manly takes first place here as they receive 6.2 penalties per game whilst yielding only 4.3 themselves for a net of +1.9. The Warriors are the only other team to receive six or more penalties per game but allow 4.9 per game for a net of +1.1.

The Roosters unsurprisingly have the worst discipline in the league for traditional penalties, being called for 6.8 per game compared with the 5.3 they’re given.

There might be something about penalties being called for Queensland teams, as the Broncos, Dolphins and Cowboys make up three of the bottom four sides here, and the Titans are also in the bottom seven.

Breaking total penalties received down by half shows that most teams follow the overall trend of fewer total penalties being called in the second half.

The exceptions are Canberra, South Sydney, the Warriors and Melbourne, who receive more penalties in the second half, and the Dolphins who receive the same amounts in both periods.

Swapping that over to penalties conceded per half, there’s only one team that averages more total penalties in the second half than the first. You can probably guess who that is.

Yes, the Roosters are the only team giving up more infringements in the second half than the first, with 5.2 total penalties in the second half and 4.5 in the first. The only other team close to them is Parramatta, who average a similar amount in both halves, but their rank is equal lowest or first halves and 6th for second halves. Another data point for their 49th minute collapses.

We can also plot net set restart and net penalties (awarded minus conceded) against total margin to get an idea of which teams are correctly gaming the system, something I first identified midway through the 2020 season. First up, net set restarts v net margin.

It’s hardly shocking to see the “conceding and winning” quadrant full of teams in the top eight. Even Brisbane and Penrith are doing well this season by not giving away more set restarts than they’re awarded. Everyone is trying to behaving and failing.

And I’m not sure what the Titans are really doing, but I’ll cut them some slack by assuming their position is more due to the first month of the season than their last month, which was far more successful.

Here’s the same chart again with net penalties instead of net set restarts.

It’s a similar chart, with the top teams in the competition in the “conceding & winning” quadrant, while most of the bottom eight is in the not conceding & losing quadrant. I did look previously at whether ladder position influenced penalty counts and found that it didn’t, this is more about style and pace of play than reputation.

Now as an added bonus for readers this week, I’m going to include some numbers on play the ball speed. Before I go ahead, I want to re-iterate that the metric of play the ball speed has zero correlation with winning games, margin of victory or ladder position. That isn’t to say that the act of a quick play the ball isn’t useful, it clearly is from watching teams play and build momentum within a set.

But the metric of play the ball speed doesn’t correlate in any way. I have some plans to delve more into the metric of play the ball speeds to show under what circumstances does that metric provide a valuable insight. The real metric that would be useful is the time from first contact to play the ball, not from when the referee calls held to the ball clearing the ruck.

With that in mind, here’s the play the ball speed by referee this season.

Similar to team play the ball speeds, which range from 3.3 to 3.5, play the ball speeds per referee are very similar as well, ranging from 3.39 for Ziggy Przeklasa-Adamski and Adam Gee, to 3.54 for Kasey Badger. It’s a very small difference and unnoticeable to the naked eye. Again though, the lower values for the metric of play the ball speed doesn’t correlate well with positive results, even if individual quick play the balls have a big impact.

Some Origin selection thoughts

Before finishing up this week I wanted to post a few thoughts regarding some of the more contentious Origin selections. Partly because it’s topical, but also because I built some fancy new jitter plots to replace my radar charts (which I’ll use for social media based on engagement for both types) which I wanted to put to some use.

Leaving out David Fifita given the way he is playing is almost unfathomable. Below is charts of per game averages and percentiles for Fifita and Jeremiah Nanai for 2024, for both advanced and regular statistics. I’m not saying that Nanai should be the one to make way for Fifita, just using him as a comparison point because he was always going to be in the side.

Fifita is almost peerless for most statistics with the ball, he’s one of the most damaging runners in the game and for someone who is ignorantly classed as “lazy” by the legacy media, he gets through a mountain of work as well.

Over the past month you’d be hard pressed to find a better form forward in the competition, let alone another Queensland forward. It’s a very NSW selection to leave out such an attacking weapon, and Billy Slater’s rationale is a bit flimsy.

“Dave created a really high standard for his footy last year, he just hasn’t quite got to that standard this year,” Slater said.

Of all forwards in the NRL based on per game averages, Fifita ranks in the 99th percentile for runs, offloads, tackle breaks and tries & line breaks (combined), 97th for run metres per game, and 96th for try assists and line break assists (combined).

Another point is that the list of forwards in the whole competition averaging more line break causes and try causes per game than Nanai is Beau Fermor. That’s the list. Fifita is still in the bottom 10% of forwards, but when you’re picking five players from the second worst defensive side in the league (only ahead of Souths), then you can’t pick and choose when to apply that reasoning.

For clarification, the further to the right on these charts the better, and the vertical axis only exists to index the players, it has no bearing on any values.

If I go to this sites advanced metrics, he’s 90th percentile for ETPCR (Eye Test Player Contribution Rating) and 89th for RMOE/run (Run Metres Over Expected per run).

Those are not the numbers of someone you leave out of origin because you think “hasn’t quite got to that standard this year”.

If you’re concerned about Fifita playing 80 minutes in Origin then I won’t disagree with you, but not being able to find him a spot on the bench is crazy. I don’t know how much more he could do. You’d have to think he’d be straight into the Queensland side if they lose the opener or struggle to post points.

The other major talking point was the decision between Robson and Koroisau looks incredibly close numbers wise. The biggest difference data wise is in missed tackles per game. Koroisau misses an incredible amount, beating out perennial leader for this category Reed Mahoney as seen below in the breakdown of standard per game statistics.

Somehow Robson, despite being on the least serious team in the competition that rarely looks like it is considering playing defense (let alone actually doing it), has avoided most of the negative statistics that would come with a team routinely allowing 30 points per game.

Most regular readers will know how I think missed tackles is another cooked stat anyway, as you have to be in a position to make one. Additionally misses by hookers aren’t as bad as they’re book ended my middle forwards, and the initial contact can slow down the ball runner. A missed tackle on an edge is usually more punishing with limited support outside.

Even by advanced metrics, which don’t favour hookers because other than tackling they don’t generate a lot of quantifiable production, it’s a bit more in favour of Robson. Some of that will come down to how much value you put into running metres from a dummy half, and also the Cowboys being far more potent with the ball, creating more scoring opportunities for Robson.

Based on those numbers it’s close, but watching games this season, Koroisau has felt far more impactful. That could be due to the Tigers playmakers being largely ineffectual, especially inside 20 and Koroisau needs to create more with the ball. Again, like Fifita, he’d have to be first up for a recall in Game 2 if NSW falters.

The last one I wanted to examine was the fullback spot, where Dylan Edwards has edged out James Tedesco. Based on the percentiles and per game averages of their regular stats, it’s close but Edwards had the lead.

By advanced stats, it’s a little closer but you would still lean towards given he is more active as a runner and has been safer with the ball in his hands.

Either way, this was one selection where New South Wales couldn’t really make a bad choice. It felt like the writing was on the wall when it leaked that Jake Trbojevic was named captain, replacing Tedesco before the teams were announced.

The choice for Queensland number six came down to Ezra Mam and Tom Dearden, and statistically it was another close race, with defensive metrics most likely outweighing Mam’s superior running game.

There was little between them in advanced metrics, with the biggest difference coming from Run Metres Over Expected per run, where Mam was better than 80% of all NRL backs.

Lastly, Spencer Leniu’s position on the bench is a bit of a stretch, he’s only played a handful of games this season and his numbers don’t scream “pick me”. I’d put up a chart but almost anyone you could name has had a better season statistically. But Origin is a different beast and I fully understand that you need a player like Leniu on the bench when you have a lot of safe, traditional middles in the side as well.