NRL 2021: faster and more unpredictable than ever

This season, the NRL marketing department has decided to run with the tagline “faster and more unpredictable than ever” in one of its ads. Putting aside the fact that this is the blandest possible declaration that the NRL stands for nothing (indeed, tacitly endorses the racist view that a black man with an Aboriginal flag is a “political” statement with no place in the sport if 2020’s back pedalling is anything to go by), I dispute their very premise.

Let’s consider the two clauses to this statement:

  • Is the NRL faster than ever?
  • Is the NRL more unpredictable than ever?

In the Boolean sense, both parts would have to be true for the statement as a whole to be logically true.

Clause 1

The Rugby League Eye Test has done all of the research regarding the sport’s perceived fastness. There’s no need for me to re-create his work here but the cliff notes are as follows. The game is more condensed. This means that the first round of rule changes during last year’s covid break increased the amount of stuff being done in the same amount of time. The second round of rule changes during the offseason has decreased the allotted time available with the same amount of stuff being done.

Players expend energy more quickly and are more tired as a result, which is a great outcome if that’s your entirely mystifying goal. Ostensibly, it was for Peter V’Landys and his team. The problem is that fatigued players make more errors, have worse tackle technique and run slower. Shockingly, this hasn’t generated more exciting or better football for mine. The contrary argument is surely increasingly difficult to mount and even for the boomer enthusiasts (read: Phil Gould) who might like to intimate that the errors have created more opportunities for natural footy players, the reality is that line breaks are higher but not setting any records, one out hit ups are up significantly and handling errors – easily the most frustrating part of rugby league – were up 12% in 2020 on 2019.

It’s almost like the players were optimised for the sporting environment in which they plyed their trade and to change that in significant ways sees many of the players – both forwards and backs – short on the necessary aerobic capacity to do their jobs. I suppose its the clubs’ faults for not seeing these changes coming a few weeks out from the season starting and recruiting accordingly several years earlier.

Without the GPS data, it’s a tough call to make whether the game is faster. By definition, if you increase the amount of stuff being done and/or decrease the time it takes, you are increasing the rate of it and in common terms, making it faster. That said, the players may not be physically moving faster and may well actually be slower.

Clause 2

People will largely be familiar with form Elo ratings by now but if not, here’s the place to start. Form Elo ratings are optimised to return the highest reasonable head-to-head tipping rate through the regular season. Given the system has been more or less fixed for the entire history of the NRL, the success of the form Elo tipping can serve as a proxy for predictability. In a literal sense, the more successful at tipping the Elo ratings were, the more predictable the season was. It’s interesting then that 2020 was by far the most predictable NRL season.

Even noting that the seasons 1998 to 2002 are slight cheats in terms of predictive power for reasons I won’t go into now, 2020’s 71.3% success rate is a full 3.5% clear of the next best 67.7% in 2011. We’re currently tracking at 75% in 2021 in what is usually the least predictable part of the season (the latter stages can be 10% higher again) and several of the upsets have swung only on an unusually high number of injuries.

Considering a long run average of 61.8%, an increase of this magnitude is significant.

Last year, when I wrote about this same topic, I believed that the theory that rule changes encouraged blowout wins was at the very least exaggerated, if not wrong. I have flipped on that position.

What’s tripped me up is that the scorelines initially seemed to be similar to previous years, perhaps a little on the low side with slightly higher margins but well within what we’re used to seeing. This has held true. From 2007 to 2020, the 2020 season had the second highest average winning score (28.6 points, compared to 28.9 in 2008 and 27.6 over the long run average from 2007 to 2019) and the fourth lowest average losing score (13.2 points, compared to 12.8 in 2013 and 13.6 over the long run average), contributing to the second highest average margin over that period (15.4 points, compared to 15.4 in 2013 and 13.9 over the long run). In 2020, some of that will be explained by the absence of home ground advantage, typically worth four points to the home team, some of it will be the usual ebbs and flows of noisy data but the majority can be explained by fewer penalties, resulting in a collapse in penalty goals.

What’s completely fucked is how much favourites are winning by. If we label the team that is judged by their form Elo rating to be more likely to win the game as the “favourite”, then the favourite’s average margin (assessing losses as negative margin and wins as positive margin) has increased noticeably since round 3 of 2020.

What this means is that the favourites are not only more likely to win than in previous years, they are more likely to do it with more points in hand. By any slicing of the data, this is more predictable, not less predictable, and worse, less competitive.

FACT CHECK: the NRL receives 4 Pinocchios out of 5 for their claim

One of rugby league’s great strengths is that any team can win on their day. We can all remember games our teams had no right to win but did anyway, through perserverance, skill, hard work or just plain dumb luck. The new rules have made it harder for upsets to occur and weakened the game and its capacity to entertain us in doing so.

An alternative explanation for this data is that I follow one of the three or four historically bad teams we’re currently seeing, so this is all sour grapes. Historically bad teams come along once every ten years or so (e.g. mid-10s Knights, mid-00s Souths, late 90s Magpies, etc) and make up 1% to 2% of all seasons played. Each is a story of their own multitude of catastrophic failures. On that basis, estimating the odds of three or four arriving simultaneously, completely indendepently from each other, and by pure chance, ranges from one in 125,000 to one in 20,000,000. Sure it’s possible but it ain’t likely.

If such teams were common enough that four could form simultaneously, then the label historic would be unnecessary. They would merely be bad in the ordinary sense. If the outcome of the rule changes is to create such a change in the league that teams that normally would only be seen once a decade are now an annual ocurrence, then it’s obvious to me that the rule changes are a net negative to the league. Who wants to relive the experience of the 2016 Knights? Who’s that for?

Just for fun, let’s have a look at the distribution of who’s benefitted most from the sudden step change in the environment.

It won’t be hard to work out which fanbases will defend the changes the hardest and which ones won’t. There’s going to be some disturbing parallels between this argument and that of climate change. There is a small but definite effect at work which has clear, negative consequences for almost everyone. The flipside is that a small group of people benefit disproportionately from this effect, so deny its existence and attribute their success to their own innate virtue. Thus, a debate is formed when the evidence is clear and there is no actual debate.

It should be fun resolving that one. I look forward to “there’s always been blowout wins” becoming the new “we’ve always had bushfires”. Small changes to finely tuned ecosystems can have outsized impacts.

The NRL is probably not faster and is definitely not more unpredictable than ever. It’s almost as if this tagline was developed on the idea that the rule changes have made the sport more watchable because the administration is in the pocket of the free-to-air broadcaster. The NRL’s slide in the ratings when compared to AFL, starting from a position of relative parity in 2019 and then falling noticeably behind in 2020, suggests that the entertainment value is not what the authorities expected.

The tide against Peter V’Landys is slowing turning. The same cheerleaders in the media who were referring to him as a “Greek god” as little as eleven months ago are now questioning the value for money on the new broadcast arrangements. It won’t take long for people to realise the on-field product is obviously worse for the changes (or maybe it will take a long time, given just how unpopular my views are in general) and could have been avoided if any of the experts had been given more than thirty seconds’ notice to think about it. The media will eventually latch on to the idea that this strongman and his lack of process has achieved very little other than to bluff rubes, including most of the NRL’s alleged journalists and the decision-makers at Super League who adopted a lot of the changes and will provide a second set of data by season’s end, and then we will enter the final stages of V’Landys’ chairmanship.

In the interim, it remains to be seen if the ARLC will come up with some more brain genious solutions to problems that didn’t exist before they created them or if we will simply have to get used to this new environment and wait for the gap to close up again. We wait with bated breath.