Following the Six Nations 2017, I read this interesting webpage explaining how World Rugby's rating was calculated (also thanks a lot to L'Equipe for pointing out the webpage). I was then wondering how much the ranking was sensitive to the method of computation, and in particular, I was eager to test the ELO method applied to chess for instance. I am grateful to this Github repo which provides with the Web database for rugby (and more) matches: this avoided me to write text miners in order to extract the results from the World Rugby webpage.
Recomputing the ranking since 2003, and trying alternative methods of ranking:
Position | Official WR ranking | Alternative WR ranking | ELO ranking | Alternative ELO ranking |
---|---|---|---|---|
1 | NZL | NZL | ENG | NZL |
2 | ENG | ENG | NZL | ENG |
3 | AUS | AUS | AUS | AUS |
4 | IRE | IRE | IRE | IRE |
5 | WAL | WAL | WAL | FRA |
6 | RSA | SCO | FRA | SCO |
7 | FRA | RSA | SCO | ARG |
8 | SCO | FRA | RSA | WAL |
9 | ARG | ARG | FJI | RSA |
10 | FJI | FJI | GEO | GEO |
Ranking as on Feb, 19, 2017.