World Ranking

[Charlie Darwin]

Wales were top seeds, Austria second.

[DeLorean]

So they were. Still though, all the one really. We knew they were four relatively even teams when the draw was made.

[KrisLetang]

There's nothing more meaningless than worrying about who we would play in a playoff BC we all know it's inevitable anyway. Why worry about what you can't avoid?

[DannyInvincible]

We're expected to fall from 29th to 33rd in the next edition of the FIFA rankings: http://www.punditarena.com/football/sokeefe/ireland-september-2017-fifa-rankings/

That certainly won't be of help to us if we do somehow manage to qualify for the play-offs as FIFA announced today that the draw would be seeded.

[geysir]

Fifa to review rankings system (http://www.telegraph.co.uk/football/2017/09/04/fifa-review-rankings-system-friendlies-harm-englands-world-cup/) after friendlies harm England's World Cup seeding

'The much-derided Fifa rankings system is to be reviewed after it emerged playing too many friendlies was set to cost England a seeding again at next summer’s World Cup.'

[DannyInvincible]

Fifa to review rankings system (http://www.telegraph.co.uk/football/2017/09/04/fifa-review-rankings-system-friendlies-harm-englands-world-cup/) after friendlies harm England's World Cup seeding

'The much-derided Fifa rankings system is to be reviewed after it emerged playing too many friendlies was set to cost England a seeding again at next summer’s World Cup.'

Ah, the good ol' Telegraph... I love how they portray it as if FIFA have been compelled to act - for all that is right and good in the world - because poor England have been in some way uniquely shafted by the governing body's incompetence/shadiness. :rolleyes:


England have refused to exploit the same loophole [that has allowed the likes of Wales, Switzerland and Poland to climb the table by playing fewer and fewer friendlies], with manager Gareth Southgate recently making it clear they should play the world’s best sides in order to better prepare for summer tournaments.

Well, isn't that very honourable of England?...


The Football Association is also keen on fixtures against leading nations in order to raise more money to invest back into the game.

Ah, so it's actually profits that might be a guiding motive behind playing glamour friendlies then rather than the purported safeguarding of competitive integrity. OK.

As is clear from the FIFA spokesperson quote tucked away towards the end of the article, FIFA would be conducting a general review of the system anyway, regardless of England's alleged misfortune:


Fifa is reviewing the Fifa Coca-Cola World Ranking system and will take a decision after the completion of qualification for the 2018 Fifa World Cup if any changes are to be made to improve the ranking.

There's no specific mention of England at all.

As an aside, it is a bit mad that Poland are ranked fifth. Where on earth did that come out of? :confused:

[KrisLetang]

The World Ranking system is sponsored by Coca-Cola? That's just silly.

[pineapple stu]

As an aside, it is a bit mad that Poland are ranked fifth. Where on earth did that come out of? :confused:
As the article says, Poland, like Switzerland, have stopped playing friendlies. Only 1 in the last 14 months. They've also only dropped two points in qualifying (the defeat against Denmark isn't recorded yet) and reached the last 8 of the Euros, where they lost on penalties. On that basis, it seems quite plausible.

Elo rankings are the way to go really.

[Gather round]

Elo rankings are the way to go really

Morning all. After getting completely muddled up-thread I've been increasingly persuaded by PA Stu and others that Elo's system would be fairer and easier to understand.

That said, NI have played 31 games since WC 2014. That should be enough (according to Elo's advocates) to measure our strength relative to teams who didn't qualify for 2016 and are doing less well than us this time. Yet Scotland, Bosnia, Denmark and Netherlands are all still ranked ahead of us.

If Elo was used to seed the current second place sides, Portugal, Italy, Sweden and Iceland would be ahead of Slovakia, Bosnia, NI and Montenegro.

Competitive Games 2014-17

SCOTLAND 18 8 5 5 29
R IRELAND 24 10 9 5 39
N IRELAND 22 13 4 5 43
WALES 24 13 8 3 47
ENGLAND 22 17 4 1 52



Poor old England. Basically none of their games matter. In qualifying, they swat aside Azermolduania and co in their sleep; the supposedly glamor friendlies are half paced and often lost anyway; then they arrive at the finals expecting to go out to the first decent oppo with soapyhands Joe throwing at least one in the net :D

[geysir]

There's a sculpture to the side of the main road south out of Reykjavik, it's from 1974 it's called the falling/failing currency and has since lived through 3 or 4 currency collapses.
I think it's black comedy genius.

I think it also fits to nicely describe Ireland's falling FIFA ranking.

http://i148.photobucket.com/albums/s31/geithals/fallid%20gengi/Fallid%20gengi.jpg (http://s148.photobucket.com/user/geithals/media/fallid%20gengi/Fallid%20gengi.jpg.html)


What has changed with our cycles though is our highs are getting lower and our lows are getting deeper.

http://i148.photobucket.com/albums/s31/geithals/fallid%20gengi/irelands%20cycle.png (http://s148.photobucket.com/user/geithals/media/fallid%20gengi/irelands%20cycle.png.html)

[KrisLetang]

Didn't Bjork go skateboarding on that once Geysir?

[pineapple stu]

Morning all. After getting completely muddled up-thread I've been increasingly persuaded by PA Stu and others that Elo's system would be fairer and easier to understand
Welcome on board! :)


That said, NI have played 31 games since WC 2014. That should be enough (according to Elo's advocates) to measure our strength relative to teams who didn't qualify for 2016 and are doing less well than us this time. Yet Scotland, Bosnia, Denmark and Netherlands are all still ranked ahead of us.
I haven't done any full analysis of what you've noted - but have you checked the standard of the teams those sides are playing?

You will never gain points for a win or lose points for a defeat under Elo - but it is possible for, say, a win and two losses to increase your rating while a draw and two wins could decrease it.

So the strength of qualifying groups will be a factor. Third behind Germany and Poland could even be better than first ahead of Romania and Hungary. :)

Elo works - so there has to be a reason there somewhere. Just a case of trying to find out what it is.

[Gather round]

I haven't done any full analysis of what you've noted - but have you checked the standard of the teams those sides are playing?

No- I've just taken as my starting point that "Ratings tend to converge on a team's true strength relative to its competitors after about 30 matches" as Elo's website suggests. I even allowed a 3% margin of error with that 31st game ;)


You will never gain points for a win or lose points for a defeat under Elo - but it is possible for, say, a win and two losses to increase your rating while a draw and two wins could decrease it

Indeed, I broadly understand the science and remember it vaguely from A Level Maths. But surely we aren't looking at short-term variations, but long-term trends?


So the strength of qualifying groups will be a factor. Third behind Germany and Poland could even be better than first ahead of Romania and Hungary. :)

Elo works - so there has to be a reason there somewhere. Just a case of trying to find out what it is

Touche. Strictly, it potentially works for football matches as the football authorities prefer to take Coca-Cola's money than use something more scientific.

As I may have mentioned previously, Romania and Hungary (as well as Greece and Finland) ALL carried a better recent competitive record into 2016 qualifying than you, Scotland, Poland and Georgia did...

[pineapple stu]

No- I've just taken as my starting point that "Ratings tend to converge on a team's true strength relative to its competitors after about 30 matches" as Elo's website suggests. I even allowed a 3% margin of error with that 31st game ;).
That's got nothing to do with the point I made though.

You can get a higher rating even by not qualifying if you have a tougher group.

You can lose half your games and have a higher rating than a country who win 80% of their games if they're against teams of sufficiently high rating.

I reckon this is what's happened the north

[Gather round]

That's got nothing to do with the point I made though.

You can get a higher rating even by not qualifying if you have a tougher group.

You can lose half your games and have a higher rating than a country who win 80% of their games if they're against teams of sufficiently high rating.

I reckon this is what's happened the north

I understand your point. Mine (I think supported by Elo's own statement) was that differing standards of groups/ individual opposition) converge over time. 30 games over 3 years is a fairly long time.

Separately- why does Elo's scheme exclude Israel from European tables but include Kibris and Vatican City? Isn't that just silly?

[pineapple stu]

You're right in that second in a group should be roughly speaking consistent - the groups are designed to be even.

Obviously points scored will impact on that - second with 24 points is better than second with 18 points.

But looking at the groups, ye did have Romania as top seeds in 2016 qualifying for example. UEFA seed them highly - and this thread is in part about how that may be wrong - but Elo won't rank them as highly at all. So it's possible that actually, ye've had an easier group or two, which is reflected in the ratings. If I get time, I might try do some sort of analysis of that.

One potential drawback with Elo ratings is a similar thing to that seen in chess, incidentally, and that's regional inflation. If you have a pool of players which is mostly self-contained (Irish chess players; South American teams), it's possible that a rating of 1700 isn't comparable to a rating of 1700 elsewhere (e.g. among Spanish chess players or European football teams). This is because there isn't enough intermingling to really set a level standard. So Venezuela in particular may be overrated, albeit that they're picking up a few points in a bloody tough qualifying group.

No idea on Israel, although Australia are still in Oceania as well, so maybe they've left nations in their original confederations? (Although Kazakhstan are in Europe, so that doesn't seem consistent)

[John83]

I think there's enough intercontinental games to avoid excessive regional differences. If it was a concern, FIFA could sponsor some intercontinental cups - essentially lower tiers to the confederation's cup - once in a while to rebalance things.

[Gather round]

But looking at the groups, ye did have Romania as top seeds in 2016 qualifying for example

Romania were a third seed for 2016 qualifying, behind Greece and Hungary.

[Closed Account]

This guy runs simulations based on ELO ratings and gives a percentage on the number of times a country qualifies.
From his last run we've gone from a
64.65% chance of qualifying to
6.51% chance of qualifying.

https://media.tenor.com/images/a7297988cd2c2afd75026454a9dad477/tenor.gif

[swinfordfc]

I know this is impossible now as there are more games tonight .... but where would we be kinda placed in this? ... Top 25?

[DeLorean]

I was hoping you might know, Swinford.

[swinfordfc]

I was hoping you might know, Swinford.

Think 24-26? .... Very hard to be exact due to games tonight!

[dantheman]

FYI, not sure if posted before. Useful site below, seems to calculate world rankings "live":

https://www.fctables.com/fifa-rankings/

[BonnieShels]

FYI, not sure if posted before. Useful site below, seems to calculate world rankings "live":

https://www.fctables.com/fifa-rankings/

Thanking you. Always something that we need around here.

[Fixer82]

How are Netherlands 20th when they've failed to qualify for the last two tournaments?

[DeLorean]

How are Netherlands 20th when they've failed to qualify for the last two tournaments?

Seems a bit mad alright. Would their second and third place finishes at the last two World Cups still stand to them?

[BonnieShels]

FIFA rankings are in an 8 year cycle. I'd say they'll fall off a cliff edge come next year.

[seanfhear]

Probably tumble over a dyke .

[DannyInvincible]

How are Netherlands 20th when they've failed to qualify for the last two tournaments?

They'll move up to 20th (from 29th) on Monday, aided by victories over Belarus and Sweden, but it does seem very high for a team who have been stuttering through qualification and who have failed to make even the play-offs two campaigns in a row. They finished third in their group this campaign and fourth in the last campaign, having picked up only 13 points and lost 5 games.

The Elo rating actually has them ranked 13th in the world, which is even more surprising to me.

[Fixer82]

Probably tumble over a dyke .

Happened me once. She knocked the boll!x out of me when she realised I knocked over her pint of Bud

[geysir]

FIFA rankings are in an 8 year cycle. I'd say they'll fall off a cliff edge come next year.

It's a 4 year cycle these days, nevertheless the Netherlands' current confounding high ranking is on feet of clay.

[Gather round]

https://i.imgur.com/pCEgejY.png

Key:

FUBAR (Floreat Ultonia's Boolean Algorithm)

Teams are ranked by points won only in WC 2018 qualifying

FIFA

All games since October 2013 count towards ranking, with greater weighting for more recent and lesser for friendlies

UEFA

All and only competitive games since September 2012 count, with lesser weighting for those before September 2014. The 'Nations League' draw is based on this ranking and is color-coded

Elo

Similar to system used to rank chess players globally

[paul_oshea]

it should be about where you finish in your last campaign(s). And anything that says Boolean in an algorithm for calculation is definitely on a hiding to nothing.

We should be 2nd seeds for next qualifying, but something tells me after the loss to Serbia we probably wont be, although a qualification play-off win might help achieve that.

[geysir]

it should be about where you finish in your last campaign(s). And anything that says Boolean in an algorithm for calculation is definitely on a hiding to nothing.

We should be 2nd seeds for next qualifying, but something tells me after the loss to Serbia we probably wont be, although a qualification play-off win might help achieve that.
The seedings for 2018 -20 uefa nations league are already fixed, Ireland are 2nd seeds in league B

[Gather round]

And anything that says Boolean in an algorithm for calculation is definitely on a hiding to nothing

Agreed. It's pretty much f*cked up beyond all recognition...

[tetsujin1979]

Confirmed, risen to 26th in the latest rankings: http://www.fifa.com/fifa-world-ranking/ranking-table/men/index.html

[dantheman]

From here:

http://www.fifa.com/fifa-world-ranking/ranking-tools/index.html

2 wins (!) could bring us to 1344, which would bring us up to 5th! :D:D

http://foot.ie/attachment.php?attachmentid=2606&stc=1

[pineapple stu]

The Elo rating actually has them ranked 13th in the world, which is even more surprising to me.
I did a bit of analysis on their Elo rating (seeing as I don't really understand the FIFA calculations) to see why they're still 13th. I also compared them to Ireland and Iceland, who both qualified for the Euros, and who've made the play-offs and the World Cup proper, yet who are both ranked below Holland. (The comparison could be done with any countries if anyone else wants to do some maths). I've taken the World Cup Final as my starting point - so that's 35 games for Holland, 38 for us and 43 for Iceland. The Elo guys reckon 30 games is enough for a rating to converge; I think it's more and that the 2014 World Cup still has some impact on the rankings. But it won't be much.

After the World Cup, the ratings were -

Holland 2137 (3 in the world)
Ireland 1667 (42 in the world)
Iceland 1515 (76 in the world)

By comparison, the current ratings are -

Holland 1854 (13 in the world)
Ireland 1750 (18 in the world)
Iceland 1822 (24 in the world)

So first observation - Holland have dropped almost 300 points, which is a hell of a lot. They've just done so from a very high starting point.

I've split each country's results into 6 categories - friendlies 2014/15, Euro qualifiers, pre-Euro friendlies, Euro 2016, World Cup qualifiers and 2016/17 friendlies - and looked at their rating performance in each category -


Fr Euro Q Euro16 Fr WC Q Fr
Holland 1753 1555 1861 1711 1878
Ireland 1898 1767 1766 1763 1739 1706
Iceland 1496 1813 1909 1403 1843 1746

Basically, if the only games Ireland played were the ones in Euro 2016, we would have a rating of 1766, or if the only games Iceland played were friendlies pre-Euro 2016, their rating would be 1403.

So Holland had an awful Euro 2016 campaign, and though they improved for the World Cup, they still performed worse than Iceland and, albeit marginally, Ireland. So why aren't they rated behind either of us?

I think the key is the friendlies - there's fewer friendlies than qualifiers (15 of Holland's 35 games in the sample are friendlies), and they impact the ratings less, but Holland's recent friendly form is way above either Iceland or Ireland. It includes wins over Spain, England, Wales, Poland and Austria, which are results they'd love to have in qualifying. Ireland, meanwhile, have lost to Belarus and Iceland, and have only really beaten the US and Uruguay of note. Iceland's friendly form is actually terrible - they lost 6 of 9 friendlies in the run-up to Euro 2016, for example, against Poland, Slovakia, the UAE, the US, Denmark and Norway. (One of their wins was against Liechtenstein) Similarly, post World Cup, they drew with Canada and Estonia.

It's obviously only fair that friendlies impact the ratings, and they do so at a much lower level than competitive games (for which there's twice as many points on offer), but still, their friendly form is very good.

I also think the 2014 World Cup has some residual impact; their recent form has been below their rating, which would indicate they will continue to slide. Ireland's is too, which is why we're down 10 points in the last year, while Iceland's form in friendlies post the Euros has improved, albeit only over 5 games.

[Gather round]

If the Elo system's 'selling point' is that it's devised by mathematicians you'd think that convergence figure would be a bit more precise. It clearly isn't as per PAS's worked example above. Here's another: NI have played 33 games since WC 2014 (24 competitive, 9 friendlies), yet we're ranked below Scotland despite a much better competitive record.

Why is it obviously fair that friendlies impact the rating? Does that mean that for that reason alone UEFA's alternative is unfair?

[pineapple stu]

It's obviously fair that friendlies impact a rating because they're valid senior internationals. It's also fair that they don't count for as much as qualifiers, which they don't.

The North have gone from 300 points behind Scotland in June 2014 to 50 points behind now. Scotland have also been on a good run of form lately and only scored one point less in 2018 qualifying - so hardly a "much better" record for the North. (2016 is different obviously)

But the North's friendly record is appalling; wins over New Zealand, Slovenia, Latvia and that's it. Scotland by contrast have beaten the Czech Republic, Denmark, the North and Qatar (who drew with the North)

[Charlie Darwin]

New Zealand are a good side. Not many teams could put five past Cabinteely reserves.

[Gather round]

It's obviously fair that friendlies impact a rating because they're valid senior internationals

No, it isn't obvious and is widely challenged (including by UEFA, which you ignored in my previous point).


The North have gone from 300 points behind Scotland in June 2014 to 50 points behind now. Scotland have also been on a good run of form lately and only scored one point less in 2018 qualifying - so hardly a "much better" record for the North. (2016 is different obviously)

Look, either we compare their record for the last one or two tournaments (in both cases, NI are obviously better in that we got more points and progressed twice). Or we credit the Scots for being merely mediocre rather than hapless in the previous two tournaments 4 or 6 years ago. Which again obviously (sorry) contradicts the basic claim of rankings converging after about 30 games.


But the North's friendly record is appalling; wins over New Zealand, Slovenia, Latvia and that's it. Scotland by contrast have beaten the Czech Republic, Denmark, the North and Qatar (who drew with the North)

NI friendly record since summer 2014: 9-4-3-2 (you forgot our thrashing of Belarus just before they went down to Cork to hand out a footballing lesson ;))

Scotland similarly: 8-4-1-3

[pineapple stu]

No, it isn't obvious and is widely challenged (including by UEFA, which you ignored in my previous point).

Not sure what the UEFA alternative you refer to is? I don't see why you think they should be ignored entirely - but in any event, we can agree to disagree on the matter - it doesn't change the substance of the Elo system. If you want to have an Elo ranking system ignoring friendlies, you can.


Look, either we compare their record for the last one or two tournaments (in both cases, NI are obviously better in that we got more points and progressed twice). Or we credit the Scots for being merely mediocre rather than hapless in the previous two tournaments 4 or 6 years ago. Which again obviously (sorry) contradicts the basic claim of rankings converging after about 30 games.
I'm not sure why you're so adamant about this precise 30 game convergence. You say earlier that you think it should be precise because it's invented by mathematicians, but maths doesn't work that way. Should pi be exactly 3 because it was introduced by mathematicians? I don't think someone who has openly admitted to talking crap about the Elo system is really best placed to make these kind of demands of it.

Anyways, the simple fact that the North got more points and qualified twice (once to the play-offs) doesn't tell the full story. Nor does qualification for a tournament mean anything of itself - there's no bonus points going for it. And why should there? The points gained by virtue of a good qualifying campaign is statistically enough.

You can do the same performance stats as before to compare the North and Scotland (and something I've noticed here is that I'm using the chess formula where a win is 1 point and a draw is 0.5, whereas in football there's three times as many points for a win. There's also bonus points for away wins, extra goals scored, etc - none of that is relevant in chess, and all will impact my calculations, but they should still be decent guidelines)


Team Fr Euro Q Fr Euro16 Fr WCQ
N Ireland 1508 1764 1833 1613 1725 1628
Scotland 1765 1661 1811 1516 1689
So a couple of things jump out -

> I was too hasty about the North's friendly record; it's actually quite comparable to Scotland's. I only really had a quick scan when making my previous post. Scotland have only played one friendly since the Euros, so their 1516 rating performance (a draw with Canada) isn't a particularly reliable figure as the data sample (one game) is too small. But pre-Euro friendly form is very similar, and quite good for both.
> The North didn't actually perform all that well at the Euros - which we know because they lost three of their four games
> Scotland actually outperformed NI ratings-wise in the last WC qualifiers - but San Marino being really crap effects on that. It drags NI's average opponent rating down a fair bit - had ye played (and twice beaten) Malta instead, as Scotland did, the rating performances would have been much the same. This again is down to the small sample size (5 nations) - it means the performance ratings are only guidelines, but as the overall ratings take in far more data, it isn't an issue at that level. So to an extent, the North were fortunate that 6½/10 (6 wins and a draw in chess speak - see my earlier point on the limitations of my calculations) was enough for a play-off spot, while the same score (5 wins and 3 draws) wasn't enough for Scotland.

But basically, there's little enough to choose between the two. In fact, there was just 5 points before the last two rounds of WC qualifying - but then NI lost their last two games, while Scotland won one and drew one. (This shows how reactive the ratings can be, which is actually a trait you earlier argued for, albeit to a higher and unnecessary extreme) Progressing past Switzerland in the play-offs could well move ye past Scotland.


NI friendly record since summer 2014: 9-4-3-2 (you forgot our thrashing of Belarus just before they went down to Cork to hand out a footballing lesson ;))

Scotland similarly: 8-4-1-3
You're right on Belarus; apologies.

But my previous point on who you played is important, and so again, you can't just compare raw records like that. Otherwise, we'd be better than both of ye because we keep beating Oman every couple of months.

[DeLorean]

Down six places to 32rd - https://www.rte.ie/sport/soccer/2017/1123/922263-ireland-drop-fifa-rankings-england/

http://www.fifa.com/fifa-world-ranking/ranking-table/men/index.html

[seanfhear]

Down six places to 32rd - https://www.rte.ie/sport/soccer/2017/1123/922263-ireland-drop-fifa-rankings-england/

http://www.fifa.com/fifa-world-ranking/ranking-table/men/index.htmlFootball must be in a bad place if we are the 32rd team in the world .

It’s time to adapt the rules to help / encourage attacking football . How can Anti-football be at 32rd in the world .

[DeLorean]

Up one place to 31st - http://www.the42.ie/world-rankings-released-today-3852881-Feb2018/

http://www.fifa.com/fifa-world-ranking/ranking-table/men/index.html

[geysir]

Hmm, I see that Heart, Passion, Skill and Team Spirit are up to a record 18th spot.

[tricky_colour]

Up 8 to 26 a decent ranking.

http://www.fifa.com/fifa-world-ranking/ranking-table/men/rank=279/index.html

One ahead of the USA who were ranked a fair bit above us IIRC, well 6 above us, they moved up one from 28.

We are 2nd highest climber in the top 30 which is promising. Netherlands highest up 9 to 20.

Scotland up 14 to 29.

NI down 3 to 23

Wales down 1 to 14

England up 3 to 12

Going in the right direction.

Bit annoying Russia ranked 65 are at the World Cup and we are not, even worse for 9th ranked Chile who ain't going either.

[tetsujin1979]

tricky, those are the rankings from last October. This is the correct link: http://www.fifa.com/fifa-world-ranking/ranking-table/men/
We're still at 31 (same as in May - http://www.fifa.com/fifa-world-ranking/ranking-table/men/rank=286/index.html ).
USA dropped one to 25

North have dropped two to 29
Scotland have dropped eight to 42
Wales up three to 18
England up one to 12

[DeLorean]

Was wondering... sorry tricky but I had to remove my 'thanks'. :)