## Analysis – The more competitive the season, the more bums on seats

Most rugby league commentators wouldn’t know what a linear regression is or how do one. I’m no different but I do like to compare two variables and see if they’re correlated. A scatter plot with a linear trendline and an R-squared – remember R-squared goes from 0, no correlation, to 1, perfect correlation; I usually need at least 0.2 to raise an eyebrow – is all I need to keep me entertained for hours on end.

Last week, we looked the concept of competitiveness and how to measure it. This week, I want to see if (more or less) competitiveness impacts on other aspects of the game. Using my preferred ratings gap as a proxy for how competitive a season is, this post looks at a few variables to see if they’re correlated.

If you want a specific variable looked at, give me a yell.

Draws

Surprisingly, there’s no link between the number of draws and how competitive the season is. There’s basically a correlation of nothing with an R-squared of 0.03 . I think draws are more about the specific teams in question and I think golden point may play a role but the overall season competitiveness doesn’t matter.

## Analysis – Is the NRL getting more competitive?

The short answer is yes and no. Yes, the NRL is more competitive now than when it started but no, it doesn’t seem to be a thing that improves consistently year-on-year.

Let me explain. On this blog, we use Elo ratings to measure teams’ performances and assess each team’s probability of winning a game in advance. Surely we can use our ratings system to assess the competitiveness of each NRL season.

Philosophically, what is a high level of competitiveness? It has to be a situation where the teams are fairly close in performance resulting in a hard to predict outcome. Here’s two ways of measuring that closeness of performance with pros and cons.

1. You could look at the spread of teams ratings. Pro – makes an assessment based on all teams. Con – if all teams are pretty average but one team is excellent and another awful, then there isn’t a big spread of talent. This would imply a highly competitive season even though there is really only one potential premiership contender.
2. You could look at the difference between top and bottom ratings. Pro – simple. Con – if one team is truly terrible and three or four are pretty good, then the season is pretty competitive but the difference between top and bottom might be exaggerated due to the crapiness of the bottom team. This would imply a not particularly competitive season despite there being multiple potential champions.

Which is better, measuring the spread or measuring the difference from top to bottom? Neither way of doing this is immediately obvious as a better method. Let’s look in more detail.