We get this question a lot, you will never find a ranking solution that will satisfy all parties and we have had some rather interesting conversations over the last season. We felt it prudent to explain how our system works and let us not understate the fact there is NO perfect system that exists.
While we tried a grading system based on points and allocating teams categories and while accurate we felt that there were issues in scaling up and including as many schools as possible. We therefore put our money where our mouths are and explored which rating system would work best for us and purchased software in order to hasten the process thereby allowing us to include far more teams.
The rating system we felt best would reflect the rankings for school rugby was the ELO Rating System.
The ELO rating was invented by Arpad Elo and it’s use has mainly been in chess. However the system has been adapted for use in American Football, Rugby League and since the end of the 2018 World Cup FIFA have used ELO as their ranking system.
ELO is used to predict the outcome of a game essentially, when two teams/players of an equal score face off against one another it is expected that should they play on multiple occasions they would win an equal number of times.
When it comes to different ratings, the odds widen and therefore the winner stands to gain fewer points for a win a loss could punish them badly. Odds in terms of rating differences indicated below:
10 points: 51%
50 points: 57%
100 points: 64%
250 points: 80%
500 points: 96%
We do a pre-season ranking based on what we think the end of seasons rankings could be, these are done in increments of 10 points. The top ranked team starts at 1,490 with teams ranked below as mentioned in increments of 10, so a top 5 would have a pre-season rating as follows:
This essentially means that we believe that there is a 55% chance that the number 1 ranked pre-season team would beat the number 5 pre-season ranked team.
Teams are allocated pre-season points until we reach number 50 on the rankings chart who are assigned 1,000 points meaning we believe our pre-season #1 team would have a 95% chance of beating the #50 ranked pre-season team.
Teams ranked 51 – 100 are assigned 950 points with any team outside of the Top 100 starting the season off with a 800 point rating.
ELO IN PRACTICE
If we take a look at when we first started the ELO system with the South African school season in 2022 we can show examples of various fixtures and how it effected each teams Elo score.
TOP TIER: GREY COLLEGE VS PAARL GIM
At the time of this fixture both schools had sublime seasons and were among the top 2 in the nation, Grey at the time were ranked 1st with an ELO rating of 1,517 with Paarl Gim 1,509. The odds were 51% in Grey’s favour for this fixture but it was Paarl Gim who ended up victorious 22-17.
After this result Paarl Gim gained 15 points and Grey College lost 15 points with the post ELO rankings moving Paarl Boys into first position with Grey moving to second. A 51% chance and a score margin of 5 shows you how effective this rating system can be.
THE UPSET: RONDEBOSCH VS PAARL BOYS
The game started off with Rondebosch having a rating of 1,364 and Paarl Boys 1,508. The odds dramatically changed because of the difference in rating with Paarl Boys having a 69% chance of a win.
We know the result, Rondebosch defied the odds and beat Paarl Boys securing 21 points with Paarl Boys losing 21 points and Rondebosch marching into the Top 10 while Paarl Boys moved down from 3rd to 5th.
ELO SHOWCASE: REWARDING TOUGH FIXTURES
Stellenberg are an excellent rugby program, they have been unlucky with rankings in recent years primarily because they have not played in the Western Province “A” league and thus have a weaker fixture list although with the talent they have they could easily best some of the top teams in the nation.
We gave Stellenberg a pre-season ELO rating of 1,380 meaning a ranking of 12th, again this was mainly due to their weaker fixture list as the team should have easily been in the Top 10 this year with wins over the likes of Boland Landbou, Outeniqua, Menlopark, EG Jansen and Drostdy as well as pushing both Grey College and Paarl Gim.
They finished the season with an ELO rating of 1,411 which meant an 11th place on our national rankings, their fixture list if what cost them points and this will of course change in 2023 with entrance into the A League but the table below tells the full story of what a weaker fixture list and not winning some key fixtures can do to a team in terms of our system:
|Date||Opponent||Opp Rating||Outcome||Player’s Rating||Change|
|Aug 20, 2022||Durbanville||1,189||won||1,411||9|
|Aug 13, 2022||Strand||895||won||1,402||3|
|Jul 30, 2022||Boland Landbou||1,260||won||1,399||12|
|Jul 23, 2022||Outeniqua||1,443||won||1,387||24|
|May 28, 2022||Brackenfell||945||won||1,363||4|
|May 21, 2022||Tygerberg||988||won||1,359||4|
|May 14, 2022||Oakdale||1,398||lost||1,355||-18|
|May 7, 2022||Rondebosch||1,398||lost||1,373||-19|
|May 2, 2022||Menlopark||1,028||won||1,392||5|
|Apr 30, 2022||Noord Kaap||969||won||1,387||4|
|Apr 23, 2022||Bellville||975||won||1,383||4|
|Apr 9, 2022||Parel Vallei||861||won||1,379||2|
|Mar 29, 2022||EG Jansen||1,236||won||1,377||13|
|Mar 25, 2022||Grey College||1,484||lost||1,364||-14|
|Mar 18, 2022||Paarl Gim||1,490||lost||1,378||-14|
|Mar 12, 2022||Drostdy||1,240||won||1,392||12|