Well ‘elo there (humour me). Glad you stopped by. If you’ve been reading this blog, you’ll remember that I introduced the concept of how important point differential is to determining a rugby team’s place in the standings. With enough information we can even build models that tell us how many Expected Points they might accumulate in the season.
But does that tell us the whole story?
Should we automatically assume that a team with a high point differential is really good? Maybe we need more information. Did they put up most of those points against bad teams? Which team is a bad team? Does a low point differential (or negative) mean that they are really bad? What if they only lost one game by a bazillion points, but otherwise have been better?
This is where using an Elo rating system can help. At its very basic level, Elo tells us the relative strength of a team. Teams that consistently beat other teams have high ratings, teams that consistently lose have low ratings. Winning games against tough opponents are worth more to a team’s Elo score than beating weak opponents (and vice versa).
The Elo rating system I use for rugby is based on point shares in a game (because as I’ve already stressed scoring more points than your opponent is fundamental to success). Just as I said that a game won against a tough opponent is worth more, so too is the share and total of points that you win by. Beating a great team 10-0 is better than beating a bad team 10-0. Likewise beating a great team 86-0 is better than beating a great team 10-0.
Below, I chart the 2013-14 Top 14 season week-by-week to show you how Elo works. Every team started with an Elo rating of 1000. Teams that hovered near 1000 for the whole season are average. The degree to which a team is above or below 1000 is an indication of their observed quality relative to average. Poor Biarritz hit 900 in the last week of the season. In contrast, Toulon, the league winner reached its peak of 1047 in the Playoff Final. One way to interpret those ratings is to think that Biarritz is TWO TIMES WORSE compared to average teams than Toulon is better than average.
What is interesting to me, is that as I mentioned when discussing Expected Points, Elo shows how competitive the Top 14 league is. Toulon isn’t clearly the overwhelming favourite (despite narratives), and aside from Poor Biarritz (TM), the other below average teams are actually quite close. Some (me) might argue that when looking at both Expected Points and Elo, Perpignan might have been unlucky to be relegated.
With this additional piece of information, making power rankings (a constant source of amusement for fans) becomes a bit more objective than just scanning the table or naming star players in each squad. Here is my stab at the end of the 2013-14 season:
Simple rules for end-of-season power rankings:
- Teams that win titles are ranked #1 (Trophies don’t lie!)
- Teams that are relegated are at the bottom (Even if both Elo and Expected Points like Perpignan more than reality)
Other than that, it’s fairly subjective (but informed).