I’m sneaking in some actual facts on a Friday while Hoss is tied up in Yowie’s basement. We’re going to look at the current ranking system, at least briefly, then look at some other possibilities that might work better.
Why now? We’re approaching the point in the cycle where the rankings, and the points that determine them, actually matter. If we’re going to make things better in the future, now is the time to do it.
Starting Point
I’m not alone in believing that the current system for ranking teams is flawed. But before we throw it out, perhaps we should look, broadly, at how it works, and what it’s meant to achieve. Then we can look at some other ranking systems and see what might work for rugby.
The Current System
Without going into all the fine detail, when two teams play each other the winner usually gains some points from the loser, the loser gives some points to the winner. The amount of points exchanged is based on the difference between the ratings. There are four main tweaks to this:
- The home side gets +3 points to their rating for home field advantage.
- If the winning side is enough (12) points higher than the losing side, they don’t gain any points. If SA play USA, we all expect SA to win, they don’t gain points for that.
- If team A wins by 15+ and is going to gain points, they get double points for a big margin.
- If there’s a points exchange at a World Cup, it’s also doubled. If there’s a big margin between two similarly ranked teams, then that’s 4x points.
World Rugby ignores these ratings except for once every four years: the seeding for the draw for the pools at the RWC depends on your standing at the time of the draw, at least for the nations that automatically qualify. Normally that’s the teams that finish first, second and third in the pools at the previous tournament, but in 2031 the USA as hosts automatically qualify for the men’s tournament.
Everyone else plays qualifiers and those qualifiers put them into pre-determined slots. They haven’t announced these yet but, for example, the top non-qualified three teams in the 2025 Pacific Nations Cup (played between Canada, Fiji, Japan, Samoa, Tonga and USA – Fiji and Japan are already qualified, so essentially the bottom team misses out) will all qualify but won’t be in the draw, they’ll have assigned spots.
So, for the men, that’s essentially World Rugby cares about the Tier One nations; for the women it’s not quite as settled yet. They also tweak the scoring system so it’s almost impossible to win the RWC and not end up rated number one in the world, although South Africa tried their hardest in 2023. Tying the 2024 series against Ireland in South Africa, with the ratings shift from playing at home, was enough to put Ireland back on top.
Desirable Features of a New System
The idea of the seeding is to give us a draw that separates teams ranked 1-4, 5-8 and to a lesser extent 9-12 from each other, so the teams in those groups don’t play each other in the pool stages, and until the right parts of the knockout stages. We don’t want the top four sides playing each other in quarterfinals for example, we want them, results permitting, to meet in the semifinals. (If Ireland are a top four team, obviously someone lower ranked will get through.)
It probably doesn’t matter too much if 4 and 5 or 8 and 9 are swapped (it might matter a hell of a lot to you if you should be ranked 4th but you’re 5th, you play 1 in pools and 2 in the quarterfinals instead of the semifinals in some ways, but, ultimately, you avoid losing a SF and the 3/4 playoff so you lose one less match…).
The 2023 seeding was a disaster because of Covid, they took the ratings really early because of the way the 2020 schedules were affected and the rankings had changed a lot by the time the World Cup started. I think, unusually for WR, they made the right call, but events overtook them. They are changing that for 2027. The ratings at the end of 2025 will be what determines the pools – which gives everyone enough time to book hotels and the like in Australia, and while there might be odd shifts in the rankings, they’ll be close enough.
It would be nice if the rankings felt responsive to a team’s improvement, and perhaps to their dips in form too. For example, Italy have been getting better, it’s clear to see, going from easybeats in the 6N, to tough opponents. But, mostly, they lose. It would be nice if, over the last decade or so, they’d seen a reward for the change from whipping boys to competing as well as for their victories. I’m not saying a huge amount, but if we expected the points exchanged for a win to be about 20, I wouldn’t object to that being 10 on an exchange system and 10 on a points share system. More on that later.
There are, regularly, complaints that “my team” (most recently South Africa) has gone down in the standings but didn’t even play. Bizarrely, we never hear anyone complain their team has gone up despite not playing. Australia have benefited from that over the July tests. It would be nice if they felt responsive system was more transparent. We’re not going to fix teams playing at different times of year, but we could make it more obvious what’s going on.
Some Options
Some of these might seem radical at first but remember, the rankings really only have value at the point of the draw for the RWC, at least in WR’s eyes.
Chess
It may surprise you to see on the list but several decades ago chess wanted to improve their ranking system so it gave a better way to predict the outcome of a match. When you first enter a ranking tournament you are given a rating of 1500 (in rugby a new nation gets a starting score of 40), and if you win you take points off your opponent, if you lose you give points to them. The points are bigger, there’s no cap on the exchange and, unsurprisingly, no home field advantage, but actually the two systems are remarkably comparable.
Underpinning this is the concept that a player has a quasi-magical “strength” at chess that we can try to measure by their rating. If they play enough games against a mix of other players with known ratings they’ll find their level. There is some maths that I won’t get into that suggests this is a reasonable assumption, at least for chess.
But is it reasonable for a rugby team? It’s the assumption that the current system uses although they don’t spell it out. But does it make sense?
Consider, for a moment, the various incarnations of South Africa we’ve seen since the World Cup as Rassie squeezed in about 70 new caps. Look at just the two matches between SA and Argentina last year. Granted, different continents, different supporters and so on. But hugely different teams from SA and different results. What does the concept of the team having a “strength” that the ratings reflect/measure in some way if one team is, essentially, 23 different players from one week to the next?
Or take over the last few years, 2023 to now, both Wales and Australia. Two teams that have had rather different fortunes across those years. Wales had a decent 2023, slightly disappointing at the end, kept their coach into 2025, then basically lost all their senior players and, thanks to injury, all their somewhat experienced ones and went on a record-beating losing streak, dropped their coach and finally won a match! Australia have kept some players, rotated quite a few, and changed coaches. They’ve also changed from not having a discernible plan under Jones to having one that they are gradually executing more and more proficiently, culminating in a great November series in 2024 and then a win against the Lions. For both these teams, what does their performance in 2023 mean in terms of their “strength” in 2024 or 2025 when so much has changed?
Since it’s relatively current and in people’s mind, how about the French? Consistently they send a young squad away in their summer, it’s part of the agreement between the LNR and the FFR in fact, about player release for the national team. But the “strength” of their touring squad compared to their 6N and November squads is clearly much lower. But not according to the rankings…
Final Thoughts on Chess (and the current system)
The chess system has some mathematical rigour for chess. While I’m not entirely sure a player has a constant strength if the points adjust quickly enough as a player gains experience and so “strength” their score will rise. I don’t follow chess, but I don’t imagine there’s a real dip with age. If there is, that would follow too. But the score follows an individual, that makes sense.
In rugby, on both a grand scale, four-year cycles and so forth, but also week to week, teams can change so much that the sense of a team’s strength is pretty much meaningless. But that is what WR use, whether they understood the assumptions or not.
A Squash Ladder
A squash ladder is common in social sports clubs, not just squash clubs. You have a ladder or list. Any player can challenge someone higher up and if they win, they take their place, everyone else moves down a place. Over time you end up with a list of players ranked in order – which is what we want.
Cons and Cons
While we want the end result, the rugby calendar doesn’t really lend itself to this approach. The November Tests are not bad for it with the Southern Hemisphere teams jumping around the Northern Hemisphere sides, but even these matches are fixed well in advance. The Six Nations is the next closest thing, each team plays the others once, but the fixtures are set for the foreseeable future – the announcement of the schedule only confirms dates and times. The July Tests and TRC don’t really offer much chance to mix things up, although Argentina would, for a week, have been world number one!
Ironically I think the WXV structure actually would make this work better, and while the traditionalists would clutch their handbags in shock, reorganising the men’s game in the same way would make this more viable. But it’s never going to happen.
The other big issue with this, if SA send a development team to Portugal ahead of the next RWC and lose Portugal suddenly becomes world number one! Is that possible? It seems unlikely, but Rassie sends out development teams so it’s not completely crazy. Portugal would deserve a reward for their achievement but I’m not sure they deserve that much of a reward unless they beat the full-strength team.
Tennis
Perhaps another surprising inclusion but tennis has some potential for a ranking system in rugby. In case you don’t know, tennis tournaments are all knock out, loser goes home but, win or lose, you get some ranking points. Keep winning and you get more, until the tournament winner gets the maximum the tournament can offer, as well as the prize money, a cup and the glory. In tennis, under normal circumstances, your points go away after a year – strictly when that tournament ends again next year, so if you go out in the quarterfinals of the Australian Open one year and don’t play the next year, you lose all those points on the ladder.
What you probably know, even if you don’t follow tennis, is that they have four major tournaments (Australian, French, Wimbledon and US), which are worth 2,000 points to the winner. But they have various smaller tournaments worth 250, 500 and 1,000 points.
Now, except for the finals stage of the RWC, we don’t play knockout tournaments in test matches. Going into the last weekend of TRC in 2024 there were theoretical ways for all four teams to win the tournament, despite Australia and New Zealand losing their first matches.
But if we adapt that idea slightly, and say that the 6N and TRC are 1800 point tournaments, the June internationals a 900 point tournament and the November internationals a 1200 point tournament you can work out the points for a win in each series, and say 1/4 of that for a loss. The benefit of this is that you can, as tennis does, build in scaling. The PNC and Shadow 6N get 900 (or perhaps 1200) points as their starting point. Matches in the June and November windows without Tier One involvement are likewise scaled. There are various development, supported and rising tiers recognised by WR, as well as “the rest.” Scaling the maximum points available down, based on the highest tier team involved helps maintain the hierarchy, but will reward teams that get promoted.
I think, because we’re looking at a four year cycle, it makes sense to have the points drop off after four years but I wouldn’t mind something where they’re halved after a year and fall to zero after two so they’re a reasonable reflection of current form.
There are some odd things with this approach. If November is a 1200 point tournament, and we look at Wales and Australia, in 2024 Wales were playing three November tests, Australia four. So Wales would have won 400 points if they’d won each match, Australia only 300. But awarding points for the tournament means you don’t have to scale.
We would have to decide what participation and invitation looked like. Japan played two games, Georgia and Portugal only one. France and Italy, like Wales played three. I think Georgia and Portugal are definitely invited teams, Japan is probably too, and that prorates their points based on their participation rate? Wales, France and Italy played for 400 points per game, so Japan, Georgia and Portugal do too. Or, simpler, they compete for the same points as their hosts. Portugal played Scotland who played four games, so they played for 300 points. Georgia played Italy who played only three games, so they played for 400. It definitely would have to be agreed before the change, but it could be worked out.
A Four Year League
We already talk about four year cycles, we could use test matches from after a RWC through to the next RWC as a four-year league, and go from there. The 6N would get off to a bit of a flier over the SH teams, yes, but it’s clear to everyone except the biggest muppet (sadly rugby fans include some muppets) that South Africa have games in hand and a chance to catch up.
I would like to see up to 10 points for a victory, simply doubling the league points we see in SRP would do, and shared 10 points for scoring too.
This sounds complicated but in each match you add the points scored together, then share the 10 scoring points out as evenly as possible between the teams. I’ve just watched a 40-0 whitewash, then the team that scored 40 takes all 10 scoring points. In an 89-88 win, they each get 5 points. In a 41-10 victory that would add to 51, so each point would be worth 0.196 ranking points. 10 earns you 2 and 41 earns you 8 points.
Remember, the team that wins is guaranteed at least 8 points for the victory, so even if they share the scoring points, and come away with 13, probably 15, they’re doing better than the losers with 5-7. But, harking back to Italy going from easy beats to hard to beat, they go from taking nothing away to taking ~4 points away from a match, maybe more with losing bonus points, and not scraping by in the standings.
Pros and Cons
This, except for the scoring points, is familiar and simple. The scoring points are not hard to understand in principle, even if you need a calculator/computer to work them out in practice. People generally understand having games in hand, so the mismatched seasons are less of an issue.
There needs to be a weighting system to accommodate different numbers of games played. In 2024, I think New Zealand and Australia played more games than anyone else, Wales the fewest. While Wales would have a terrible score regardless, in a different year they’d have no chance of catching the All Blacks because of that. So, at the end of each year, and just before the draw and at the end of the RWC, prorate everyone’s score by the ratio of their games to the lowest played in their tier. So if Wales played 11 and that’s the lowest, anyone who played more multiplies their league points by 11/x where x is the number of tests they played (I think 14 for the ABs and Australia) to get their adjusted score. (I prefer this to inflating everyone else’s scores but that works too.) You could also produce a percentage score, so Wales could if there are 20 points available per game, theoretically score 220 points. How many did they score as a percentage of this? Do the same for everyone else, and you have a fair ranking of performance (Wales would have been close to 0%). You can keep updating the percentage of total available points week by week, just like the current rankings system.
Everyone starts at 0 again after each World Cup and WR might think this is a problem: this system rewards all teams equally. A team ranked 50 in the world that plays 10 matches against their peers and wins them all comfortably ends up with more points than your choice of the best team in the world this year – I’m taking SA. This isn’t hypothetical: if you score on win-loss ratio alone Zimbabwe were the best team last year, winning 100% of their games.
There are ways to handle this. Perhaps most simple, if you end the league just before the RWC starts, in the same way SRP switches from a league to a knockout structure for the finals, but the teams that go to the World Cup gain double points for the next league. South Africa would have a 116 point winning bonus, and about 90 points from scoring (despite all those close wins in the knockout rounds they had three pretty big wins and a close defeat in the pool stages). So they’d start on 206 points. Zimbabwe weren’t at the World Cup so they start on 0. Teams that implode after the World Cup, like Wales, start with a buffer of ~150 points from the World Cup which, loosely, carries them over for not winning anything in 2024, and getting thumped several times. But over a four year league, if they don’t start to improve, that’s less than you’d get from ten wins: it’s a bonus rather than a guarantee of inclusion.
There are other approaches. You could award, for example, the top 20 teams 8+2 and 10 points, the next 20 teams 6+2 and 8 and the rest 4+1 and 5 points. At the World Cup everyone gets 16+4 and 20 points. This might rewrite who is in which stream for the next four years. This is not as terrible as it seems, automatic qualification for the RWC doesn’t include all Tier One nations, and certainly doesn’t routinely include Tier Two nations. Having them handicapped and unlikely to be at the top of the table makes things easier for the seeding. It may make things easier for the normalisation to the number of games played too, if you do it by bands of 20 teams.
Polo
Another sport that might surprise you to see it here. I don’t really know much about polo, but one of the random nuggets I knew and that I backed up with a bit of research, is that in some games they use a handicap system. Each player has a score from -2 (novice) to +10 (superstar). Both teams add the points for their players together, you work out the difference and the lower team starts with that many points on the scoreboard.
Adapting This To Rugby Rankings
There are a few parts to this as a process. One is a bit of a black box but it’s not impossible. So let’s start there.
We need a system that generates player ratings. These exist – anyone who plays fantasy rugby has seen one – working out a system that measures all contributions to a match and assigning a score to them is a bit of statistical black magic. (For those who grok statistics, it’s some sophisticated multivariate modelling. I don’t envy whoever has to do it, but all the data is available, there’s just a lot of it.) I think, although I don’t know how the model would work out, there might be some interesting outcomes.
- How would the bomb squad, with fewer interactions per game, rank compared to the Irish front row, who tend to play long minutes with more chances for both good and bad outcomes?
- Is there a way to measure the impact of the coach? You can’t really say it’s zero, look at the way South Africa, New Zealand and Australia have changed tactics this year, compared to last year, and how it’s affected them at various times.
- Who are the most valuable players on the pitch? Is it the same position for every team?
Because this would rate players based on their contributions to the team, it’s possible that changes in coaches and tactics are reflected in those contributions. South Africa don’t box kick and chase anything like as much this year as last year, so the ways the whole team gets points will change.
Before you say “but this is the same ‘measurement of strength’ as the current system” stop a moment. The current system abstracts strength for a team. This looks at strength of individual players. As and when Malcolm Marx retires, he takes his points with him. There is a team strength, a team ranking, but neither Matfield’s impact for South Africa nor Gregan’s for Australia count any more.
Then comes the problem of how you convert these player ratings into a world ranking. I would suggest an extended squad with a structure. Whilst I’m not fixed on this, something like eight props, five hookers, seven locks, 11 back rowers, four nines, four tens, six midfielders and 11 back three players. That’s 60 in total. They don’t all have to be fit to play right now, so France could have kept Ntamack, Baille etc over November, and you just add up the rankings of the 60 players to get the team ranking. Coaches announce them before a tournament/tour and can change them before each one. This would let Schmidt take his huge squad on the November tour and keep some injured players in his scoring squad.
If the modelling for the player rankings is good, and publicly available alongside the team rankings, it should be possible to take a team sheet, assign a ranking to it, and then make some good predictions about the score, in terms of probable outcomes. If you go in with that as the designed outcome, you should be able to get a good model to do that.
Teams that are improving get rewarded as their players get better scores in close defeats, even if they’re not winning yet, and better scores again when they start winning. So that complaint disappears.
It seems that this nicely fits the desired outcomes and, while it’s a bit complicated, a computer can crunch the data and give you player scores and team scores by noon GMT on Monday, when they’re announced currently, so it’s not delaying the changes in the new rankings.
Closing Thoughts
Personally I like the polo solution. It gives us some hard data to bring to arguments about best player in the world. It keeps the idea of a numerical measure of the strength of a team, but it bases it, more credibly, on the sum of the current players and how they’re playing. It seems somewhat stable in response to teams not playing
After that, the four year league comes second for me. Although there are a few issues that need to be worked out with it, they’re not that hard. The simplest part, the direct points for winning, are familiar. Adding on the extra points for scoring is definitely novel but it’s not complicated. It adds a few points to reward teams that are getting close but not winning yet. But it also rewards teams that get a blowout victory. A league doesn’t really give you any predictive power but, although the Chess system uses that, WR doesn’t really care about that aspect. A league would do the job just fine. Addressing the different number of games played and perhaps handicapping – whether by tiers, performance in the last World Cup or something else – are the big two
As I suggested above, ending it just before the RWC and counting the games at the tournament as double value for the start of the next league pretty much guarantees the cup winners start ranked number one, and gives a bit of a boost to teams that were there, but not enough to help over four years and ~50-60 games if they collapse like Wales have.
I don’t hate the adapted tennis system either. It nicely has tiers built in. Although it’s not instantly comfortable to think of the June and November tests as a “tournament” it’s not a huge stretch to think about them that way and see the points being awarded for the games played. The different points for the teams in the same game are a bit odd, but we’d get used to it.
Over to you
Obviously WR is not going to change. Which of these do you prefer and why? Or do you have some other system you’d like to see?
