This page contains a short rundown of some of the advanced statistics for rugby league that used on the Eye Test. The first three have links to longer articles that explain them in greater detail. All statistics (unless otherwise noted) are taken from the Fox Sports statistics page
Tackle % – Also referred to ask Tackle Rate. An advanced statistic for rugby league that is used to quantify how often a player makes a tackle whilst on field, normalised by the estimated number of “plays” defended during their time on field. After all, you can’t complete a tackle if your team has the ball. The idea behind this metric (and the following two) is to provide a way of quantifying the effort of middle forwards who don’t play big minutes and are often overlooked for players who put up bigger raw numbers in larger minutes. The average Tackle % for a middle forward is in the 25-26% range, indicating they complete a tackle on one out of every four plays their team defends. The 2020 leader for Tackle % was Jai Whitbread of the Gold Coast Titans at 35.93%.
Run % – Also referred to as Run Rate. An advanced statistic for rugby league that is used to quantify how often a player completes a run whilst on field, normalised by the estimated number of “plays” the players team had whilst they were on field. Like Tackle %, the idea is to show which players are completing a high rate of runs for their time on field, as middle forwards usually play fewer minutes than the rest of a team. The average Run % for middle forwards is usually in the 10-12% range, meaning they complete a run in one out of every ten plays. The leader for 2020 was Cronulla’s Andrew Fifita at 17.34%.
Involvement Rate– Combines Tackle % and Run % to give a holistic metric for player involvement during a rugby league game. Like the other two advanced statistics, this is normalised by the number of plays whilst on field. The average Involvement Rate for middle forwards is between 17-19%, meaning they either complete a run or tackle on nearly two out of every five plays whilst they are on field. Generally Tackle %, Run % and therefore Involvement Rate generally decline as minutes increase. The leader for 2020 was Jaimin Jolliffe from the Gold Coast at 21.93%
Net Points Responsible For (NPRF) – A metric for a players overall contribution to a teams performance. Each score by a player is valued as it is on the scoreboard (try – 4, goal – 2, field goal – 1), plus 4 points for every try assist and try contribution. Four points are removed for every try cause a player concedes as a way of quantifying their defensive contribution. This total is then divided by the number of games played to get a plus or minus total points per game that a player is “responsible” for. The NPRF leader for 2020 was Penrith’s Nathan Cleary at +9.72 per game.
Error Rate – The number of possessions a player has divided by the number of errors they have made. The end result is X number of possessions by a player for every error generated. The worst performing players each season usually have an error rate of <10 possessions for every error committed. Nene McDonald from the Sharks was the 2020 leader, committing an error every 5.33 possessions in his two games.
One of the things I’ve noticed over the past few rounds is that the average time of ball in play has dropped slightly to the pre Rugby League 2.0 levels. This comes after a decent increase earlier in the season once the rules were changed. Focusing just on time in possession, the last NRL three rounds haven’t had more than 57 minutes of ball in play, the three lowest rounds this season and both before V’Landysball was introduced in Round 3.
This led me to investigating why, and I put together the below chart plotting time in possession (sourced from NRL.com) against points scored per game. The blue line represents average time in possession for the first 14 rounds of the 2019 and 2020 seasons, and the yellow bars represent the average points scored per game in each round (by both teams). There’s a reference line on these bar charts as well to show the average for 2019 and 2020. For points its about the same – 38.8 in 2019 and 39.9 in 2020.
Initially I thought that the amount of points scored was reducing the time in possession, with more tries and conversion increasing the amount of time the ball was doing nothing. But if you look at the above chart, it’s not really apparent – Rounds 8 and 11 had average game scores below 40 points, but time in possession above 62 minutes, significantly higher than other rounds this season.
I should note at this point I’ve filtered out any golden point games to normalise minutes per game. A great example of why is Round 3 2020, where the average goes from 58.81 to 61.46 if you include the Panthers v Knights drawn match which had a whopping 80 minutes of time in possession. Another note is that Round 12 2019 had only four games played due to State of Origin, which is why it looks like an outlier.
It’s not due to tries either, see below for the chart that shows why. Round 7 2020 had 8.3 tries and nearly 58 minutes of time in possession, while Round 9 this year had 6.7 tries but 62.7 minutes of possession. Again, this makes sense with the previous chart as points are a factor of tries scored.
My next thought was maybe there are fewer penalty goals? There are fewer penalties being called, so it makes sense that there were fewer penalty goal attempts this season. Whether or not that’s a good or bad thing is another discussion, especially in those instances where a team is down 2 inside the opponents 20m zone and gets a set restart. But that’s another matter for another time. Below is penalties awarded plotted against time in possession.
This led me to look at penalty goal attempts against time in possession. The data checks out – 1.6 attempted penalty goals last year against 1.1 in 2020. And they’re being taken at a lower rate too. In 2019 penalty goal attempts comprised nearly 20% of all shots at goal. In 2020, that number has dropped to just 13.6%. So that’s the likely reason for the increase in time in possession, right? Less time standing around waiting for a kick at goal.
Hang on, let’s look at something a bit closer on that chart. Round 1 and 2 had time in possession of 58 and a half minutes and an even 57 minutes, respectively. That’s more time in possession than the last three rounds under one referee and with set restarts. There’s actually been six rounds since Round with less time in possession than Round 1.
Yet Round 1 and 2 had over 2 penalty attempts per game, far higher than the rest of 2020 and more than most rounds last year to the same point. How did those two rounds still have high time of ball in play yet more penalty goal attempts?
Maybe the time elapsed during a penalty goals is counted as time in possession? If that were the case, that wouldn’t explain Round 12 having 56.6 minutes in play with 1.4 penalty goal attempts per game, while Round 8 had almost 63 minutes in play with just 0.6 penalty goal attempts per game.
Maybe the game is just faster? In this “faster pace” era, everything is up, and more stuff is being done. So far this season we’ve seen an increase in time that the ball is in play. There’s an increase in runs and play the balls as well. Although not an increase in play the ball speed.
But we do know from the first chart that the ball is in play more this season by about 8% compared to 2019 for the first 14 rounds. Runs are up nearly 10% compared to the same point in 2019. Passes are up 5%, line breaks are up 7% and tries are up 10% Everything is up! More stuff is good!
Kicks are also up 7.5% year on year, with long kicks up 17% and attacking kicks are up 7%. More stuff! But hang on – weighted kicks are down 20%. That’s strange. Forced dropouts are down 2.1%. Kicks dead are down 3%. Why would those kicks be down year on year if everything else, including other types of kicks, has increased?
The fact there’s not a corresponding increase in weighted kicks, kicks dead and dropouts, and a higher increase in attacking kicks than other statistics indicates something has changed. You might be slowly getting at where I’m leading with this and why its taken over 700 words.
To save you anymore of this shaggy dog story, here’s my crackpot theory – teams have gotten more efficient and accurate at aiming their attacking kicks just outside the goal area to avoid a seven tackle set. The rule change which came into effect in Round 1 that gives airborne attacking players the same level of protection as airborne defensive players is surely a driver for this, as Daniel Tupou was showing before succumbing to injury.
This explains the drop in weighted kicks but the large increase in attacking kicks. Fewer kicks reaching the in-goal area leads to fewer dropouts which can take up to 45 seconds each. By aiming them a bit shorter than the try line, at worst a team will give up possession less than 10 metres out or a scrum at the same point. This is a much better result than a seven-tackle set from the 20-metre line.
Why does this make such a difference in time in possession? A drop out usually takes 40-45 seconds off the clock, because the NRL has a rule saying you can take that long (another rule change with unintended consequences). In the first two rounds this season, there were 20 fewer forced dropouts than the first two rounds last season than in 2019. If you are generous and say each one takes 40 seconds, there’s 920 seconds saved across two games. Divide by 60 to get minutes and then divide again by the eight games per round and you get an extra 57 seconds saved on average per round purely from fewer dropouts.
This would account for some of the time in play change for Rounds 1 and 2 this year compared with last year. It also explains why Round 3 had only a slightly higher time in possession than Rounds 1 and 2 – the time savings from reduced penalties was cancelled out by having over five dropouts per game that round. The chart for average dropouts against time in possession is below.
These first two rounds this season serve as my exhibit A, albeit with a small sample size. There is similar average time in possession to post Round 3 (excluding the golden point draw), but there were still two referees and no set restarts. A comparable number of penalties were awarded as previous seasons yet more penalty goals attempted. The key is fewer dropouts in Rounds 1 and 2 compared to 2019, and below the average for 2020.
Need more proof that a reduction in forced dropouts might be part of the increase in the time of possession? Exhibit B – the last three rounds have had the three lowest time in possession averages this season, all under 57 minutes as noted in the first paragraph. In the last two of those rounds (13 and 14, factoring out Round 12 due to fewer games), dropouts are up 31% year on year and weighted kicks up 11%. As opposed to down 2% and 20% for the season so far. Goal attempts were down 3% over these rounds too, ruling that out as a cause as well. Why the change in kicking? Teams may be finding that their tactic of launching more bombs aimed outside the try line hasn’t been as successful and are adapting. Whatever the cause, there’s another link between time in possession and dropouts taken.
The increase in time on possession hasn’t isn’t just a result of rule changes, but a larger and more complicated combination of change in playing style to suit for these rule changes. The consistent attribution of faster “pace” and more “stuff” being done given solely to set restarts and one referee is proving to be a false equivalence, but one that will get a lot more airtime to boost agendas. If you really wanted to speed up the game, you’d halve the clock on dropouts.
The North Queensland Cowboys announced interim Warriors coach Todd Payten as their coach for 2021 on Friday, and using the Eye Testtm it’s easy to see why. The Warriors have improved on the field under his watch and are showing a lot more enthusiasm and commitment, even after a coach they supported was removed. Their fightback against the Eels showed a desire that Warriors sides haven’t shown late in a game for quite some time.
This led me to have a look at what changed under Payten and how the Warriors improved under his tenure by looking at the teams per game statistical averages under Kearney compared with Payten’s performances.
First up I’m going to qualify everything below with a small sample size disclaimer – we’ve got six games for Kearney and eleven for Payten to analyse. These aren’t representative, more they are indicative of their performances, but within those games there are hundreds if instances of runs and tackles which gives me some comfort. It’s not like I’m writing an article purely on the win/loss percentages of teams where one win or loss would throw out my premise completely. But I digress.
Below is a chart of the percentage change for a number of statistical categories for the Warriors in 2020, with te blue dots represent the percentage difference between Kearney and Payten, and the orange dots represent the difference between Payten and Kearney. The further the circle is to the right or left, the larger the difference. Whichever colour sits on the right-hand side shows which coach had an advantage in that area, and at first glance you can see by the number of organe points on the right hand side that Payten has outperformed Kearney in the majority of these statistics.
I wanted to note that I’ve chosen percentage change because on a per game basis it’s hard to get a scale that fits average total run metres in the thousands (1400m+) with average metres per run in single digits (8.5-9.0). Otherwise it would be impossible to see some of the changes. Another reason, which if you’re becoming familiar with my posts you’ll know, is that the exact number isn’t as important to me as the size or direction of the trend. I’m looking for how much things have changed under Payten.
WIth that out of the way, let’s delve in a bit deeper to the differences and start with the first line for points scored per game – the orange dot on the right shows Payten had an increase of nearly 50% in points scored per game, with the Warriors going from 12.2 under Kearney to 18.4 per game under Payten’s stewardship. Looking at from the other perspective, the Warriors scored 34% fewer points when coached by Kearney in 2020.
Tries and line breaks were also up significantly under Payten, whilst a possession statistics like play the balls and total sets were down between 1-4%. From this group of stats, you can see that not only were Payten’s Warriors scoring more, they were doing so with less possession. That is countered by the fact they had slightly better field position, as play the balls inside the opponents half and opponents 20 metre area were up 2% and 1% respectively.
The next set of stats I wanted to focus on were runs, run metres and passing. Both teams averaged the sane number of runs (168 per game), which makes a fantastic baseline to use.
There was no change there from a quantity point of view but it’s very clear they’ve changed how they were running the ball, and its effectiveness. The first is that dummy half runs were up 44%, from 7.5 to 10.8 per game. One pass runs, your standard hit up, were down 13% under Payten, whilst general play passes were up 17%. The increase in passing wasn’t just mindlessly throwing the ball around either, as offloads increased dramatically after the coaching change, with nearly 70% more total offloads (6.7 to 11.6) and a triple figure percentage increase in effective offloads.
This would suggest that he has given his dummy halves more freedom, allowing them to skip out of the ruck and engage the line before spreading the ball wide, and the increase in passing stats shows they were playing a more expansive game compared to the safe conservative style under Kearney. He’s also unlocked their ability to promote the ball with offloads as well. Given these changes it is not surprising that they may have scored the try of the year against the Eels on the weekend.
Although as you’d expect, moving the ball around more often did create more errors, which increased by 17%, similar to the increase in passing but that’s just the cost of doing business to improve a teams attack.
On the run metres front, it’s a blanket increase of 2-3% under Payten, and the increases to Post Contact Metres could indicate players increasing their effort as they hit the line and trying to push through initial contact.
Finally, I wanted to look at how their kicking changed, which has seen a drop under Payten in total kicks and kick metres. This lines up with the above changes, showing the team passing the ball more and in better field position, reducing the need for long kicks (or kicking the ball at all) to end a set. Total kicks are down nearly 6% And when they were kicking, they are doing so more accurately and effectively – fewer kicks dead (down 45%) and more forced dropouts (up 63%).
Things have also improved defensively for the Warriors under Payten as well in a few key areas. Total run metres conceded are down 6.7%, post contact metres by opponents are down 9% and offloads have dropped 8.4%. Clearly Payten’s Warriors are putting more effort in defense, reducing opponents gains after contact, and wrapping up the ball carrier more effectively.
And Payten has achieved all of this with a similar line up to Kearney. If anything, you could argue he was dealing with a weaker hand – Ken Maumalo, David Fusitu’a, Agnatius Paasi and King Vuniyayawa all returned home mid-season. They were replaced with loan players like Jack Hetherington, George Jennings, and Daniel Alvaro, who have been fantastic additions but came without knowing the structures and combinations the departing players already understood. Despite these issues, their discipline has improved with 15% fewer penalties conceded per game.
The fact that he has been able to squeeze this extra performance out of a squad with significant challenges living away from their families is incredible. Next season he’ll inherit a stronger Cowboys squad that desperately needs a new direction and an injection of belief and creativity. Payten has shown in just 11 games so far this season that he’s just the man to deliver all three, and it’s encouraging to see North Queensland give him a chance instead of giving a problematic and divisive former coach another run around.
In last weeks trends and notes post, I showed there was a negative correlation between set restarts and margin, which had been positive from Round 3 to Round 7. With another week of matches completed, I thought I’d dig a bit deeper into this to see if I could find out what was leading to this.
The reason I find this so interesting is that it doesn’t conform with traditional rugby league thinking. Possession is treasured, and statistics like run metres correlate with winning games (keep in mind correlation doesn’t equal causation). Yet by giving away more set restarts, you’re giving possession and therefore more metres to the opposition. Surely that would result in giving up more points?
Looking at net margin plotted against net set restarts after Round 9 shows a similar chart to last round. I’ve named the quadrants as well to make it easier to identify what the chart is showing and the bigger the dot the more set restarts conceded.
As with last week it appears that “winning” the set restart count is inconsequential, with only two teams having any significant net margin despite coming out ahead with set restarts. The top left quadrant – “Conceding and winning” – is the one I want to focus on though, given the makeup of teams within that area.
Last week there was only one top four side in that top left quadrant, which was the Panthers. This week they’re joined by the other top four sides – Melbourne, Parramatta, and the Roosters – indicating that giving away set restarts is a genuine part of their strategy. And that is only counting the set restarts given, not intentional slowing down of the ruck that isn’t called.
The Panthers are a curious case. Penrith not only have the largest difference between set restarts conceded and awarded at -23, they have also conceded the most in the NRL at 43 and been awarded the fewest at 20.
I noted last week that the only time they’ve come out ahead in set restart differential (Round 6 against the Eels), they lost by 6. This continued in Round Nine, with Penrith conceding three more set restarts than Cronulla, yet still beating them by 32 points.
In addition to the Panthers, among the other top four sides, the Roosters have the second fewest restarts awarded (23), the Storm are third fewest (24), while the Eels sit seventh (27).
With the limitations on publicly available data it’s hard to see exactly why they’re benefiting from these restarts, but we can use what is available to craft some ideas. One theory is that the early set restarts that are in vogue help the defensive team get set, limit chances for the team with the ball to gain momentum and exploit any breaks in their line.
A way of quantifying this is could be by examining the amount of runs under and over 8 metres conceded against the raw number of set restarts. As a team concedes more set restarts, there is a small positive correlation with runs shorter than 8 metres conceded (top of chart below), and a small negative correlation with runs longer than 8 metres (bottom of chart below).
Given this correlation, you could assume that the more set restarts you concede, the more likely you are to give up shorter, ineffective runs than longer more damaging runs.
This makes sense – giving up a set restart on the first or second tackle on your opponent’s 20m line and contain them within their own half is preferable to allowing them to string together a number of longer runs and push into your territory for an attacking kick or set piece. It also enables teams to maintain a defensive structure and limit any gains from broken play.
Examining when set restarts are given and the outcome of the consequent set compared to the average set would show if this is successful or not, but again we’re limited by publicly available data.
On the other end of the scale, the Bulldogs position on this chart is just another sad indictment on their run under former coach Dean Pay. They play a very conservative brand of football, limiting defensive mistakes and (attempting to) maintain possession and complete sets. As much as he has been able to get his players to show up every week under trying circumstances, this style of play hasn’t yielded any results and the constant switching of combinations appears to be actively hurting their performances. And let’s not even talk about the Queensland sides.
I’m not arguing that conceding another set of six to their opponent is the reason that the top four sides are sitting where they are. But there is something in the new ruleset for rugby league that is widening the gap between the haves and have nots.
At a surface level there’s a minor increase in win percentage if you lose a set restart count. Removing drawn games and even set restart counts, teams who had a negative set restart differential have won 56% of their games this season. In Round 9, six of the eight games were won by teams with a negative set restart differential. The only teams that won with a positive set restart differential on the weekend were Souths and St George Illawarra.
Conceding more set restarts to your opponent isn’t going to win or lose games for you, but strategically conceding them appears to be part of the game plan for the successful clubs in 2020.
This week should have been a fantastic matchup of halves, with Shaun Johnson of the Sharks facing up against the Panthers and Nathan Cleary. Cleary has been one of the standouts this season, whilst Johnson is having a great season, but you wouldn’t necessarily know it from the way he’s often covered by the mainstream media. I’d pointed it out previously with a radar chart comparison in early August:
Johnson ended up missing the game due to some minor injuries and the birth of his child (congratulations Shaun!) and Cleary had another strong showing. Despite this we can still have a look at their statistical output for the season and use it as a test case for counting stats and raw volume statistics not telling the full story.
If you’ve been following for me for any length of time you know that I’ve put together some advanced statistics for rugby league, as using raw numbers mean players who spend the whole 80 minutes on the field usually dominate. If you’ve not read them, I’d recommend checking out my articles on Run %, Tackle % and Involvement Rate on the website, as they’re all incredibly useful in identifying high
But back to the topic at hand. It’s lazy analysis to only use counting stats without context but is more palatable to the wider viewing audience so I’m not going to deride them for dishing up what the consumer wants and easier to digest in small doses.
Comparison of raw numbers – Cleary seems ahead. More passes, more runs, more kicks, more attacking kicks, more tries, more try contributions, more line breaks and more line engagements. Johnson is only ahead in try assists (20 to 14) and weighted kicks (17 to 10).
Per game stats will give a slightly better comparison, although it paints a better picture for Cleary who has only played 12 games compared to Johnson’s 14.
When players are compared on the usual pregame shows, it’s assumed that all players in a certain position play a similar game or similar role within a team, leading to scorching hot takes like this on social media from mid-June:
And if you look at the raw volume or counting statistics at that time without any context you’d probably agree – Johnson isn’t impacting the game as much as Cleary is.
But there’s one variable that’s not usually discussed (although the wonderful Jason Oliver pointed it out in his Round 15 preview on SportsTechDaily, another must read each week), is possessions. The amount of times a player gets his hands on the ball plays a massive part in his statistical output. The basketball adage of “you can’t rebound the ball out of the basket” can be applied here with a twist, you can’t do more in attack in rugby league without the ball in your hands.
For the season Cleary has 964 possessions, compared to just 733 for Johnson, a difference of 231 possessions. On a per game basis, that’s roughly 73 possessions for Cleary and 52 for Johnson. Cleary has his hands on the ball nearly 30% more than Johnson on a per game basis.
Knowing this, what if we looked at the same statistics again for Cleary and Johnson on a per possession basis. Would it show anything?
Not initially, as those numbers are essentially meaningless – 0.016 line break assists or 0.187 line breaks per possession isn’t really meaningful. You can’t create 20% of a line break.
Instead we’ll normalise it to take out any bias that having more possesions per game creates. I’m going to pick a set number of possessions in between both players to even things out. The number doesn’t matter so much as long as we use the same number for both, and for this exercise I’m going to use 60 per game, since it’s a nice round number that falls between both Cleary and Johnson’s season average.
Now we can see that they’re not performing that differently. Cleary has an edge with kicking, especially long kicks, whilst Johnson leads on weighted kicks and try assists. Yet for all the calls that Johnson needs to run the ball more, his runs, passes and line engagements are very similar to Cleary’s. Does he really need to “do more”?
That leads into the other part of player this analysis their positioning and their role. As mentioned above it’s assumed that all #6s and all #7s should play identically but this is rarely the case.
This was shown in a great article by Jack Snape from the ABC showing the locations NRL halves are receiving the ball. If you apply the Eye Testtm during Sharks games you’d know that Johnson sticks primarily to the right side of the field, while Cleary tends to operate on both sides. The above article shows this, with the locations of Cleary’s touches coming evenly across the field whilst Johnson’s are mostly on the right. You could then argue that if Johnson had a similar level of freedom as Cleary, he would probably increase his raw statistics.
The other part of the role is not just what side they’re playing on but how often they’re involved and relied upon for their team. Cleary takes about 16% of the Panthers total possessions, makes 36% of their passes, 75% of their kicks, 66% of attacking kicks and 44% of line engagements. Johnson on the other hand, takes 12.6% of Sharks possessions, 29% of their kicks, only 52% of their kicks and 45% of attacking kicks.
This again ties back to role. Similar to a sport like Formula One, the Panthers play with a clearly defined lead half in Cleary, with Jarome Luai supporting him. The Sharks more often than not play with their halves on a closer to equal footing, a 1a/1b type scenario, with each either sticking to their side of the field or sharing in the playmaking duties.
Asking Johnson to “do more” or run the ball more won’t necessarily help his game. More stuff isn’t always better. Johnson will turn 30 in a few weeks and it could be that he doesn’t have the same explosiveness and has developed more as a player is now picking his spots and interjecting himself at the right time. Just because he’s not shredding teams anymore with one of his explosive runs and making defenders look as if their feet are stuck in cement doesn’t mean he’s not impacting games.
Whatever the reason, the outcome of this specific comparison of Cleary and Johnson is that they’re both having amazing seasons for their team, and the difference in their statistical output is simply down to the role they play for their side, which can be accounted for by looking at a per possession basis.
Claiming one is better based on one number or a group of counting statistics won’t prove anything other than teams and players are different and may play different styles. Hopefully we’ll see some more analysis based on possessions than just counting stats moving forward.