FIFA rankings - why they are important and how to beat the system

[Kent]

I've talked before about how the vast majority of international games are within your own confederation. And for any result (excluding knockout rounds) takes points from one nation and gives them to the other. Which means that each confederation kind of has a pool of points that it spends most of the time just shifting around from team to team. That pool of points can grow in knockout rounds of their competitions (Nations League, or the continental Cup), friendlies against teams from other confederations, and the World Cup.

I was very interested in seeing how CONCACAF did in regards to their pool of ranking points at the World Cup... but then Canada lost all 3 games and well, I didn't get around to doing the work until Canada got some wins and moved back up the rankings a bit. Below is the number of ranking points each confederation changed by during the World Cup window (including friendlies in that window).

So Africa were the big ranking points winners (Morocco accounted for 109 points) and CONCACAF were the big losers (Canada accounted for -43 points).

Since the new ranking system came out in August of 2018 until now, here is the change in ranking points for each confederation.

[eramosat]

Panama win should net Canada 11.49 points and a 2-place bump to #45.  Now ahead of Romania and Cote d'Ivoire.

[Kadenge]

1 minute ago, eramosat said:

Panama win should net Canada 11.49 points and a 2-place bump to #45.  Now ahead of Romania and Cote d'Ivoire.

Kone is a smart guy. Sali it's time to come home.

[Kent]

11 hours ago, eramosat said:

Panama win should net Canada 11.49 points and a 2-place bump to #45.  Now ahead of Romania and Cote d'Ivoire.

Thanks for this. Looks like my go to website doesn't have the CONCACAF Nations League games for some reason. https://football-ranking.com/fifa-rankings

I dug up my old spreadsheet for calculating rankings, I think a win on Sunday would get Canada another 17 points or so.

In looking at that old spreadsheet, it reminded me that Gold Cup games are worth more ranking points than Nations League. Even a Gold Cup group stage game has a higher importance factor than the Nations League playoffs.

[Redpunkfiddle]

5 hours ago, Kent said:

Thanks for this. Looks like my go to website doesn't have the CONCACAF Nations League games for some reason. https://football-ranking.com/fifa-rankings

I dug up my old spreadsheet for calculating rankings, I think a win on Sunday would get Canada another 17 points or so.

In looking at that old spreadsheet, it reminded me that Gold Cup games are worth more ranking points than Nations League. Even a Gold Cup group stage game has a higher importance factor than the Nations League playoffs.

 

I ran the numbers a bit back. Here is what I came up with, with some commentary on our total ranking from that time:
 

Here are our possible point gains for CNL and Gold Cup, give or take- with some variability based on ranking point changes as teams win or lose over the next few months.

CNL SF vs Panama: +11.47

CNL Final vs USA: +17.30 or Mexico: +16.84

Gold Cup group vs Guatemala: +8.96. vs Cuba: +5.18.

vs T&T +9.86 or vs Guyana +5.00 or vs Grenada +4.76 vs Guadeloupe +zilch

Gold Cup QF vs Jamaica: +15.46

Gold Cup SF vs Mexico: +26.95

Gold Cup Final vs USA +27.68

Current ranking and points: 1442.66 (42).  The current 30th ranked team Ukraine is on 1530.04

Total points from matches we are favoured in (Panama CNL, Gold Cup group and QF vs Jamaica)  is about 53. That gets us to about 35th or so.  CNL Final and Gold Cup SF/Final wins push us up further. Total point gains from winning out is about 120, which gets us to 22nd or so territory.

USA and Mexico are on 1653 (13) and 1631 (15) points. It will take a couple of cycles of CNL and Gold Cup success to achieve that level.

[The Beaver 2.0]

3 hours ago, Redpunkfiddle said:

 

 

I ran the numbers a bit back. Here is what I came up with, with some commentary on our total ranking from that time:
 

Here are our possible point gains for CNL and Gold Cup, give or take- with some variability based on ranking point changes as teams win or lose over the next few months.

CNL SF vs Panama: +11.47

CNL Final vs USA: +17.30 or Mexico: +16.84

Gold Cup group vs Guatemala: +8.96. vs Cuba: +5.18.

vs T&T +9.86 or vs Guyana +5.00 or vs Grenada +4.76 vs Guadeloupe +zilch

Gold Cup QF vs Jamaica: +15.46

Gold Cup SF vs Mexico: +26.95

Gold Cup Final vs USA +27.68

Current ranking and points: 1442.66 (42).  The current 30th ranked team Ukraine is on 1530.04

Total points from matches we are favoured in (Panama CNL, Gold Cup group and QF vs Jamaica)  is about 53. That gets us to about 35th or so.  CNL Final and Gold Cup SF/Final wins push us up further. Total point gains from winning out is about 120, which gets us to 22nd or so territory.

USA and Mexico are on 1653 (13) and 1631 (15) points. It will take a couple of cycles of CNL and Gold Cup success to achieve that level.

Top 35 seems reasonable for 2023. That's a solid goal, among other KPIs.  Top 30 by end of 2024, then top 25 as we head into WC2026

[youllneverwalkalone]

I feel like we haven't beat the system quite yet. 

[Kent]

2 hours ago, The Beaver 2.0 said:

Top 35 seems reasonable for 2023. That's a solid goal, among other KPIs.  Top 30 by end of 2024, then top 25 as we head into WC2026

Keep in mind, if we qualify for the Copa America, it will be hard to not lose points in that tournament in 2024. Progress is rarely a straight line.

[Jedi Ram]

http://www.football-rankings.info/2023/06/fifa-ranking-july-2023-probable-ranking.html

This one gives the projected points gained from 2 tournaments.

[maplebanana]

12 hours ago, Kent said:

Ah, that reminded me what my snarky comment in my notes meant. I was thinking it was about teams like Guadeloupe and French Guiana, that are CONCACAF nations but not FIFA nations. It was actually directed towards FIFA nations that aren't in CONCACAF. What the heck is New Zealand's CONCACAF ranking points? Do they just use FIFA points in the absence of CONCACAF points and presume they are weighted the same? That would make Argentina an underdog against USA and Mexico. So that can't be it. Or maybe they use FIFA points for both teams if there is a non-CONCACAF team involved, and then just apply those points to the CONCACAF set of points?

Bringing this discussion over from the gold cup thread, I've come up with a first pass at deciphering the confusing CONCACAF index. This may be interesting to only @Kent and me and is a lengthy post. Feel free to skip if you are not Kent.

The points gained as stated by CONCACAF is P = K * (R - E), where K is a weighting factor for match importance, R is the result, and E is the expected result.

K = 17.5 (friendly) or 35 (CNL, WCQ, GC)
R = -1 for loss, 0 for draw, +1 for win
E = Probability of win (Pwin) - probability of loss (Ploss).

Now here is where it gets difficult. I assumed that Pwin is calculated in an elo-style rating like the FIFA ranking, and that Ploss = 1 - Pwin. So -1 < E < +1 as is consistent with CONCACAF statement. Further I assumed that the formula takes on the same form as the FIFA ranking, only with home advantage modifying the input team rating. So if home advantage is C, and Canada is A pts then at home it's like a team that is A+C in rating at a neutral site. There is also a denominator (D) that scales the teams in the calculation, and for FIFA ranking that number is 600. So for a team of rating A at home vs a team of rating B, the formula then is:

P(win) = (10^((B-(A+C))/D)+1)^-1

So to test this out, I looked at the old versions of the CONCACAF page on wiki (thanks @h coach) to get the rankings on Feb 28 and compared them with the rankings on March 31. I looked at 4 teams: Canada, Honduras, Mexico, Jamaica. This had a mix of friendlies, CNL, home/away, neutral site games. And since they were all playing each other, it's a good chance to look at internal consistency. I only have whole numbers for the rating so I can't get super accurate, but using a home field adv of C = 175 and a denominator D = 750 gives me pretty good agreement for the March window for all 4 teams:

Canada = 1713 -> 1743 (actual); CUW away win (17.8) + HON home win (12.7) = 1743.5 (predicted)
Honduras = 1399 -> 1403 (actual); ELS neutral win (16.2) + CAN away loss (-12.7) = 1402.5 (predicted)
Mexico = 1960 -> 1939 (actual); SUR away win (7.3) + JAM home draw (-27.5) = 1939.8 (predicted)
Jamaica = 1486 -> 1479 (actual); T&T home loss (-27.7) + T&T home draw (-9.2) + MEX away draw (+27.5) = 1476.6 (predicted)

Going further back, if I look at your question about NZ ranking, it says they are 'assigned' a ranking. Back in 2018 we had a ranking of 1540 and to gain 14 pts from a neutral site win using my formula/weightings, that would have meant NZ was around 1400, which puts them in the El Salvador area, which sounds about right.

Now looking at CNL that just passed, we would have gained 32.4 pts for our PAN win and lost 25.7 for our USA loss (assuming both were called neutral site because that's so CONCACAF). So if I'm right, we're looking at a current CONCACAF index of 1750. We need to do notably better than CRC in the gold cup to leapfrog them (at 1796 before they lost to GUA and ECU this window) to get into 3rd place seeding for CNL '23-'24 (and Copa America qualifying). Wins against Guadeloupe (+5.9), Guatemala (+18.3), and Cuba (+10.4) will help.

If anyone is interested in the Excel file I put together to play around with the numbers, let me know.

[Kent]

That's awesome! On my first pass I seem to be messing up somewhere because I'm not getting the same results as you. I wouldn't mind seeing the spreadsheet or even just the formulas in excel format that I can just copy and paste to avoid screw ups. Otherwise I'll take another crack at it later today.

[Rivaldo]

On 4/11/2023 at 6:00 AM, Kent said:

I've talked before about how the vast majority of international games are within your own confederation. And for any result (excluding knockout rounds) takes points from one nation and gives them to the other. Which means that each confederation kind of has a pool of points that it spends most of the time just shifting around from team to team.

...

Since the new ranking system came out in August of 2018 until now, here is the change in ranking points for each confederation.

Ouch. That will hurt CONCACAF nations when the seeding is done at World Cups (fortunately we shouldn't have to worry about 2026) and possibly the 2024 Copa America.

Some of the ranking change came from the World Cup, but much more of it is coming from somewhere else. Is it mostly because other confederations are playing more knockout rounds where points are awarded, but none subtracted?

Is the ranking change just the difference in the sum of all nations in the confederation? If so it's a less significant change when spread among 54 nations in CAF than 10 in CONMEBOL. 

[Kent]

1 hour ago, Rivaldo said:

Ouch. That will hurt CONCACAF nations when the seeding is done at World Cups (fortunately we shouldn't have to worry about 2026) and possibly the 2024 Copa America.

Some of the ranking change came from the World Cup, but much more of it is coming from somewhere else. Is it mostly because other confederations are playing more knockout rounds where points are awarded, but none subtracted?

Is the ranking change just the difference in the sum of all nations in the confederation? If so it's a less significant change when spread among 54 nations in CAF than 10 in CONMEBOL. 

Yes, it is just the difference in the sum of all nations in the confederation. Good point about the average being spread against different numbers of nations.

The difference is due to a few factors. The smaller knockout round is a factor, kind of. It's cancelled out by the fact that we have 2 Gold Cups per 4 year cycle where UEFA, and AFC have just 1 (CONMEBOL seems to have no discernible schedule, they just do it whenever). CAF has the best of both worlds with a round of 16 and the Cup of Nations happening every 2 years.

Another factor is inter-confederation friendlies (CONCACAF lost just about each one this past window for example). Also, the more upsets there are in the aforementioned knockout rounds will result in a larger point gain for the confederation. For example, if Antigua beats USA in the quarter finals of the Gold Cup, they will gain more points (with USA losing none) than if USA had beat Antigua. Upsets I would presume are less likely in CONCACAF because there is such a quality gap between the top 2 and the rest of the field. Also, those top 2 teams are the host (USA) and de-facto host (Mexico) for every knockout round game, which helps make upsets less likely.

Then yes, World Cup results as well of course. Qatar coming to the Gold Cup could eat away at our continental point total (or if Qatar does poorly, CONCACAF could steal some points from AFC). Next Copa America will be a chance for CONMEBOL and CONCACAF to try to steal some points off of each other.

[Kent]

6 hours ago, Kent said:

That's awesome! On my first pass I seem to be messing up somewhere because I'm not getting the same results as you. I wouldn't mind seeing the spreadsheet or even just the formulas in excel format that I can just copy and paste to avoid screw ups. Otherwise I'll take another crack at it later today.

Nevermind @maplebanana. I was in less of a rush and I've got it now. I needed to write down on paper the Pwin formula to better visualize it, then I ended up doing the calculations by hand (with a calculator). I think I made some sort of mistake with the calculator. I've put it in my spreadsheet now and I'm getting the same results now!

Really outstanding work!

[maplebanana]

2 hours ago, Kent said:

Nevermind @maplebanana. I was in less of a rush and I've got it now. I needed to write down on paper the Pwin formula to better visualize it, then I ended up doing the calculations by hand (with a calculator). I think I made some sort of mistake with the calculator. I've put it in my spreadsheet now and I'm getting the same results now!

Really outstanding work!

Excellent! My wife caught me working on this last night and said 'don't post this, they'll think you're a nerd!' To which I replied 'I'm pretty sure that ship has sailed'.

Now the next question is whether or not we lose points for knockout round losses, which we would not for the FIFA ratings. Let's circle back after they publish the next batch of CONCACAF ratings.

 

[Kent]

1 hour ago, maplebanana said:

Excellent! My wife caught me working on this last night and said 'don't post this, they'll think you're a nerd!' To which I replied 'I'm pretty sure that ship has sailed'.

Now the next question is whether or not we lose points for knockout round losses, which we would not for the FIFA ratings. Let's circle back after they publish the next batch of CONCACAF ratings.

 

5 years ago when I first tried to crack the code, I enlisted a bit of help from my wife. Today I told her there was a breakthrough for soccer nerds, and she was happy when she found out what it was. She remembered the problem!

[Jedi Ram]

On 6/22/2023 at 3:19 AM, maplebanana said:

Bringing this discussion over from the gold cup thread, I've come up with a first pass at deciphering the confusing CONCACAF index. This may be interesting to only @Kent and me and is a lengthy post. Feel free to skip if you are not Kent.

The points gained as stated by CONCACAF is P = K * (R - E), where K is a weighting factor for match importance, R is the result, and E is the expected result.

K = 17.5 (friendly) or 35 (CNL, WCQ, GC)
R = -1 for loss, 0 for draw, +1 for win
E = Probability of win (Pwin) - probability of loss (Ploss).

Now here is where it gets difficult. I assumed that Pwin is calculated in an elo-style rating like the FIFA ranking, and that Ploss = 1 - Pwin. So -1 < E < +1 as is consistent with CONCACAF statement. Further I assumed that the formula takes on the same form as the FIFA ranking, only with home advantage modifying the input team rating. So if home advantage is C, and Canada is A pts then at home it's like a team that is A+C in rating at a neutral site. There is also a denominator (D) that scales the teams in the calculation, and for FIFA ranking that number is 600. So for a team of rating A at home vs a team of rating B, the formula then is:

P(win) = (10^((B-(A+C))/D)+1)^-1

So to test this out, I looked at the old versions of the CONCACAF page on wiki (thanks @h coach) to get the rankings on Feb 28 and compared them with the rankings on March 31. I looked at 4 teams: Canada, Honduras, Mexico, Jamaica. This had a mix of friendlies, CNL, home/away, neutral site games. And since they were all playing each other, it's a good chance to look at internal consistency. I only have whole numbers for the rating so I can't get super accurate, but using a home field adv of C = 175 and a denominator D = 750 gives me pretty good agreement for the March window for all 4 teams:

Canada = 1713 -> 1743 (actual); CUW away win (17.8) + HON home win (12.7) = 1743.5 (predicted)
Honduras = 1399 -> 1403 (actual); ELS neutral win (16.2) + CAN away loss (-12.7) = 1402.5 (predicted)
Mexico = 1960 -> 1939 (actual); SUR away win (7.3) + JAM home draw (-27.5) = 1939.8 (predicted)
Jamaica = 1486 -> 1479 (actual); T&T home loss (-27.7) + T&T home draw (-9.2) + MEX away draw (+27.5) = 1476.6 (predicted)

Going further back, if I look at your question about NZ ranking, it says they are 'assigned' a ranking. Back in 2018 we had a ranking of 1540 and to gain 14 pts from a neutral site win using my formula/weightings, that would have meant NZ was around 1400, which puts them in the El Salvador area, which sounds about right.

Now looking at CNL that just passed, we would have gained 32.4 pts for our PAN win and lost 25.7 for our USA loss (assuming both were called neutral site because that's so CONCACAF). So if I'm right, we're looking at a current CONCACAF index of 1750. We need to do notably better than CRC in the gold cup to leapfrog them (at 1796 before they lost to GUA and ECU this window) to get into 3rd place seeding for CNL '23-'24 (and Copa America qualifying). Wins against Guadeloupe (+5.9), Guatemala (+18.3), and Cuba (+10.4) will help.

If anyone is interested in the Excel file I put together to play around with the numbers, let me know.

This is very interesting... I have created FIFA rankings excel worksheets... But it's very time consuming... I would love to have a look at it from CONCACAF perspective. 

I would presume that they use the same methodology for club rankings. 

[InglewoodJack]

New FIFA rankings just dropped. We inched up to 45, just below Norway and above Slovakia. 
 

Costa Rica drops to 42, Mexico up to 14, US to 11.

 

https://www.fifa.com/fifa-world-ranking/men?dateId=id14058

[Kadenge]

and for those that say that we failed at the WC, because  we should have at least got a point in the 3 group games, take note:

Belgium is ranked #5, Croatia #6 & Morocco # 13

 

[InglewoodJack]

2 hours ago, Kadenge said:

and for those that say that we failed at the WC, because  we should have at least got a point in the 3 group games, take note:

Belgium is ranked #5, Croatia #6 & Morocco # 13

 

Not to mention, that’s with both Belgium and Morocco dropping. I think Belgium was 4th and Morocco 9th after the World Cup?

[Kent]

14 minutes ago, InglewoodJack said:

Not to mention, that’s with both Belgium and Morocco dropping. I think Belgium was 4th and Morocco 9th after the World Cup?

Belgium dropping doesn't strengthen the "our group was super strong" case. They are undergoing a correction at the moment. They were ranked number 2 in the world at the World Cup, but realistically they should be significantly lower than that, but it will take time for the rankings to reflect their true current level.

[InglewoodJack]

5 minutes ago, Kent said:

Belgium dropping doesn't strengthen the "our group was super strong" case. They are undergoing a correction at the moment. They were ranked number 2 in the world at the World Cup, but realistically they should be significantly lower than that, but it will take time for the rankings to reflect their true current level.

What I meant was more that we played in a group with 3 teams in the world top 15, which is crazy, but in reality, we played 3 teams in the world top 10 which is even crazier. I don’t buy that Belgium was ever the second best team in the world (or 5th for that matter), but the point is that we had to play in a group with a top 5 team and two top 10 teams, and yet we’re pissed because we should’ve won one of those games and had a chance to draw another. That’s crazy.

[Kent]

Honduras' late equalizer against Qatar was a 17 FIFA point swing from AFC to CONCACAF in my nerdy CONCACAF FIFA points pool monitoring. If Qatar had won they would have gained 14 points, but the draw means they instead lose 3 points. Glad to see Qatar struggling for a couple reasons, one of which is that CONCACAF is stealing points from AFC's collection.

[eramosat]

More importantly than how well our World Cup opponents did...and continue to do...and which completely explains how far Canada fell and so fast....

#3,4,5,6 - CONCACAF has tightened up majorly.  Canada remains 45th, as tie vs Guadeloupe had no impact.  But 3rd place CONCACAF Costa RIca has fallen all the way to 42, while both next teams rise... Panama at 57 and Jamaica at 62.

So.  It's much, much tighter than just a month ago. The best have fallen, the pretenders have advanced.  Does it matter beyond our entertainment?  \

Yes?..but also not sure what the next event is that will rely on FIFA rankings that Canada cares about...could be beyond 2026...

[Kent]

1 hour ago, eramosat said:

but also not sure what the next event is that will rely on FIFA rankings that Canada cares about...could be beyond 2026...

It could also be as early as next year. In 2019 the Copa America used FIFA rankings, but in 2021 it looks like they did a North/South split (might have been for COVID reasons). Most importantly, they used the FIFA rankings for the 16 team Centenario tournament in 2016. They did 4 pots of 4 teams.

If the top 6 CONCACAF countries by ranking qualify for the Copa America, and all team's rankings stay the same from now until the draw (and they use FIFA rankings as they did in 2016), it would look like this.

Pot 1
Argentina 1
Brazil 3
USA 11
Mexico 14

Pot 2
Uruguay 16
Colombia 17
Peru 21
Chile 32

Pot 3
Ecuador 40
Costa Rica 42
Canada 45
Paraguay 48

Pot 4
Venezuela 56
Panama 57
Jamaica 63
Bolivia 83

It should be noted that in 2016 they didn't strictly just use FIFA rankings. They also used some shady decision making that would make CONCACAF proud. USA and Mexico were put in the top pot, because USA were the hosts, and Mexico... I kid you not, "is the most decorated nation from their region." Brazil was also put in the top for the decorated nation reason, Argentina also put in the top pot because they were the highest ranked. They then proceeded to put the other 4 CONCACAF nations in their own pot. I could see them doing the CONCACAF pot thing, to minimize CONCACAF vs CONCACAF matches in the group stage, but hopefully they don't do the Pot 1 shenanigans again. But I guess we have gone full circle now. If they do the CONCACAF pot and just pluck out USA and Mexico for whatever reasons they decide to make up, then I guess the rankings won't matter to Canada for the Copa America anyways.