There was some talk recently on NRL twitter about short turn around times after Ricky Stuart mentioned about having only five days between games. The question raised was if there was any data to back up claims that they have a negative effect on performance. Turns out there was quite a bit.
NRL Fanalytics has previously posted a fantastic chart that showed the win rate for teams with various days of rest from 1998-2019.
Their conclusion was “The longer the rest, the lesser the home ground advantage”. The chart also shows that there is practically no difference in win percentage on a five-day turnaround (50.5%) compared to seven days (50.0%).
Andrew Ferguson from the best rugby league site on the internet chimed in as well, posting a fantastic chart on twitter showing win percentage by turnaround time from 1998-2020.
It’s worth opening the image to see the win percentage by each team for each day turnaround. Yet another piece of data showing that there’s very little difference in 5-8-day turnarounds for clubs.
The evidence that five day turnarounds don’t impact winning, which corresponds with an article from the Sydney Morning Herald in 2016 which came to a similar conclusion.
NRL Physio has also posted some data on turnaround times which shows there is no substantial difference in injury rates when teams back up after 5 days.
The lack of a difference in in injury rate was also noted in an acadmeic paper by Nick Murray, Tim Gabbett and Karim Chamari in 2014, titled “Effect of Different Between-Match Recovery Times on the Activity Profiles and Injury Rates of National Rugby League Players”.
You’ll notice from the above NRL Physio Twitter conversation that I’ve also done some previous analysis for NRL Supercoach scores which also shows a very minor change in player scoring with extra rest.
It should be noted there is a danger in using fantasy points scoring as a proxy for performance because the points are awarded to make scoring within the game interesting and not necessarily as reflection of player worth. Good fantasy sport players often do not make good players in real life – there’s a reason the term “empty stats” gets thrown about regularly.
For example, one of the statistics that correlates the best with winning games, run metres, isn’t a scoring category in Supercoach (unlike NRL Fantasy). However, each runs over 8 metres are worth two points as opposed to runs under 7 metres being worth one point.
Still, it corroborates the other evidence that turn around time has little effect on performance from a players perspective. That gives us three data points (win percentage, injury rate and player performance) that is virtually unchanged with less rest.
This got me thinking, is there any other underlying evidence that short turnaround times affect performance? Maybe there’s a change in how games are played under these conditions compared to the standard rest time?
To look at this, I’ve put together the data from 2014-2020 (up to Round 11), removing Round 1, as well as the first game back from the season resumption in 2020 since they have no turnaround time. Games on 10-17 days rest have also been excluded due to a smaller sample size.
Here’s what the average NRL game looks like from a number of statistics based on turnaround times from 5 days up to 9. The sample size is what we in the industry would call “robust” if we were pitching some data to a sales and marketing team. The term “games” is probably a misnomer, as it is the instances of a team playing on each turnaround time, since the team playing on five days is usually facing a team that has had a longer recovery.
Looking at the averages, teams playing under a five-day turnaround average slightly less “stuff” than six and seven days but very similar to eight days rest, and more than nine days. For example, there are around 11 fewer runs per game from teams on a short turnaround, which results in three fewer completed sets, with six fewer passes, which leads to 1.5 fewer kicks per game. These are all intertwined so the very small drops across the board make sense.
Despite doing less “stuff”, teams on a short turnaround seem to play a slightly safer style of game. They have a slightly higher completion rate, fewer offloads, higher metres per run, miss fewer tackles (as a percentage of made tackles), and concede fewer penalties.
Now that we’ve looked at the raw numbers, lets now look at the same statistics with the percentage difference between each days rest. Each column below shows the percentage difference from 5 days to 6 days, then from 6 to 7, and so forth.
Again, we can see there’s minimal changes across the board. Most of the increases are a few percentages at most, with the highest being in the 6-8% range from five to six days, and then declining by a few percentage points with each extra day to recover.
Looking for the biggest changes, there’s only three statistics with a double figure percentage. For each of them, there’s an easy explanation.
Long kicks and attacking kicks increasing by 30+% makes sense when you pair it with the difference in runs. As mentioned above, there’s an 11 run per game difference when moving from 5 to 6 days, which is about 2% of total runs and about three sets of six if you take the 3.6 tackles per set.
If you’re adding three sets of six, those sets need to end somehow and the most likely ending is a kick. And with only four kicks per game, even adding one more would be a 25% increase.
Missed tackles increasing by 12.5% looks like a decent increase, although tackles increased by 8.5%. This indicates that teams with longer rest miss more tackles than those on five-day rest. We’re talking fractions of a percent here, from 92.3% to 92.0%, but it may again point to a safer game plan or more emphasis on defense with a short turnaround.
The takeaway from this, other than there being no substantial statistical difference in performance when looking at the break between games, is that those on a five-day turnaround may be implementing a slightly more conservative game plan. This makes sense as you have less time to game plan and focus more on recovery.
Whilst all available evidence points to short turnarounds having minimal effect on games, this is not to say that there is no impact. Smarter people than I with access to better information will know for sure, and it’s highly likely that some players aren’t hitting their maximum speed or tracking the same distances in these games. There is also the mental side of backing up, which could influence decision making which would not necessarily be show statistically.
So, to answer the original question, how do games change with shorter turnarounds? To put it bluntly, at appears they don’t. They don’t result in a lower win %, injury rate isn’t changed, player performance doesn’t vary significantly and teams laregly output the same amount of “stuff” as they would with any other amount of rest.
This is more for the broader rugby league community who tend to blindly repeat the short turnaround myth despite an overwhelming amount of evidence that fails to support it.
If anything, the bigger story is that teams with a nine-day turnaround have slightly lower statistical averages across the board than teams playing on 5-8 days rest. But no coach is going to complain about having too much time off, are they?
Penalty and set restart update
We’re now approaching a point where there’s little need for additional penalty and set restart analysis, things are quite stable week to week. This week we had a penalty or set restart every 18 tackles, up from 16.5 last week. Here’s the breakdown of penalties and set restarts by round and you can clearly see we’ve fallen into a stable pattern.
Looking at referee averages and it’s a similar story with little change. The only difference is Adam Gee dropping his spot as the top caller of set restarts to Chris Butler, although Butler has only officiated two games in Rounds 8 and 9.
Speaking of Gee, in some big news, he broke his streak of six straight games with at least seven set restarts called in the first half, calling just seven in the whole game during the Panthers/Titans matchup. Who says referees aren’t consistent?
Net Points Responsible For
A few weeks back I went through some data on Points Responsible For as a way of highlighting playmakers who don’t always get the credit for a try through the awarding of a try assist. Scott Drinkwater was the notable example there, who had set up a number of Cowboys tries but did not receive a try assist for them.
This week, I’m looking at Points Responsible For (PRF) again, but this time including the Try Cause statistic that Fox Sports keeps showing the value of playmakers on both sides of the field.
Net Points Responsible For (NPRF) follows the same formula as before – tries *4 + try assists *4 + try contributions *4 + field goals but adds in a negative four points for every try caused.
Here’s the top 20 after Round 11.
The difference can be seen with Drinkwater in this chart, who has 8 try causes for the season. Under the normal PRF calculation, he would be ranked sixth at 7.56 PRF per game. Once you add in the defensive side, he drops to 15th with an average of just 4.00 NPRF per game. Benji Marshall is another who suffers, moving from fourth in PRF at 8.14 per game to ninth at 4.71 per game.
By far the biggest loser though is Matt Dufty who drops from fifth under PRF to 35th on the back of having 12 try causes attributed to him. His average plummets from 8.00 to 2.67 per game.
The other advantage of using NPRF is that you can also look at players who are a net negative, giving away more points than they score or contribute to. Here’s the top 20 after Round 11.
Unsurprisingly there’s a number of Broncos and Cowboys topping the list, led by Jesse Arthurs who let in 8 tries in four games, with only a try assist on the positive side. Esan Marsters with 14 try causes is also having a difficult season, with a negative NPRF of 4.4 per game thanks to having 14 opponent tries attributed to his defense, or lack of it. .
Blake Ferguson showing up on this list is more an indication of how little he is seeing of the ball this season than any defensive deficiencies.