So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.

Based on Elo ratings, these are the toughest GoDs since 1970:
YEAR GROUP TEAMS AVG. ELO
1970 3 England, Brazil, Czechoslovakia, Romania 1959
2014 B Spain, Netherlands, Chile, Australia 1931
1978 3 Brazil, Spain, Austria, Sweden 1923
1974 2 Brazil, Yugoslavia, Scotland, Zaire 1915
1978 1 Italy, Argentina, Hungary, France 1910
2022 G Brazil, Switzerland, Serbia, Cameroon 1901
2014 G Germany, Portugal, United States, Ghana 1896
2002 F Argentina, England, Sweden, Nigeria 1895
1978 4 Netherlands, Scotland, Iran, Peru 1894
1994 E Italy, Ireland, Mexico, Norway 1889
2018 E Brazil, Switzerland, Serbia, Costa Rica 1888
1982 6 Brazil, Soviet Union, Scotland, New Zealand 1885
1986 E West Germany, Denmark, Scotland, Uruguay 1885
2022 E Spain, Germany, Japan, Costa Rica 1885
2018 B Spain, Portugal, Iran, Morocco 1884

Two from 2022 made the cut.

On 2022-11-13 17:21, Al Kamista wrote:
> So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
>
> Based on Elo ratings, these are the toughest GoDs since 1970:

Are you defining group of death as the ones with the highest average Elo
ratings ?
But what is a group of death, really? Should it not perhaps be the most
competitive group, perhaps represented as the one with the smallest
standard deviation of ratings ?

> YEAR GROUP TEAMS AVG. ELO
> 1970 3 England, Brazil, Czechoslovakia, Romania 1959
> 2014 B Spain, Netherlands, Chile, Australia 1931
> 1978 3 Brazil, Spain, Austria, Sweden 1923
> 1974 2 Brazil, Yugoslavia, Scotland, Zaire 1915
> 1978 1 Italy, Argentina, Hungary, France 1910
> 2022 G Brazil, Switzerland, Serbia, Cameroon 1901
> 2014 G Germany, Portugal, United States, Ghana 1896
> 2002 F Argentina, England, Sweden, Nigeria 1895
> 1978 4 Netherlands, Scotland, Iran, Peru 1894
> 1994 E Italy, Ireland, Mexico, Norway 1889
> 2018 E Brazil, Switzerland, Serbia, Costa Rica 1888
> 1982 6 Brazil, Soviet Union, Scotland, New Zealand 1885
> 1986 E West Germany, Denmark, Scotland, Uruguay 1885
> 2022 E Spain, Germany, Japan, Costa Rica 1885
> 2018 B Spain, Portugal, Iran, Morocco 1884
>
> Two from 2022 made the cut.

On Monday, November 14, 2022 at 12:21:12 AM UTC, alka...@hotmail.com wrote:
> So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
>
> Based on Elo ratings, these are the toughest GoDs since 1970:
> YEAR GROUP TEAMS AVG. ELO
> 1970 3 England, Brazil, Czechoslovakia, Romania 1959
> 2014 B Spain, Netherlands, Chile, Australia 1931
> 1978 3 Brazil, Spain, Austria, Sweden 1923
> 1974 2 Brazil, Yugoslavia, Scotland, Zaire 1915
> 1978 1 Italy, Argentina, Hungary, France 1910
> 2022 G Brazil, Switzerland, Serbia, Cameroon 1901
> 2014 G Germany, Portugal, United States, Ghana 1896
> 2002 F Argentina, England, Sweden, Nigeria 1895
> 1978 4 Netherlands, Scotland, Iran, Peru 1894
> 1994 E Italy, Ireland, Mexico, Norway 1889
> 2018 E Brazil, Switzerland, Serbia, Costa Rica 1888
> 1982 6 Brazil, Soviet Union, Scotland, New Zealand 1885
> 1986 E West Germany, Denmark, Scotland, Uruguay 1885
> 2022 E Spain, Germany, Japan, Costa Rica 1885
> 2018 B Spain, Portugal, Iran, Morocco 1884
>
> Two from 2022 made the cut.

What about 1982? Brazil,Argentina, Italy

On Sunday, November 13, 2022 at 10:00:11 PM UTC-5, MH wrote:
> On 2022-11-13 17:21, Al Kamista wrote:
> > So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
> >
> > Based on Elo ratings, these are the toughest GoDs since 1970:
> Are you defining group of death as the ones with the highest average Elo
> ratings ?

Wasn't me, it was 538. Forgot to post the link: https://fivethirtyeight.com/features/whats-the-2022-world-cups-group-of-death-its-tough-to-pick-just-one/

On Monday, November 14, 2022 at 6:04:39 AM UTC-5, Mark wrote:
> On Monday, November 14, 2022 at 12:21:12 AM UTC, alka...@hotmail.com wrote:
> > So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
> >
> > Based on Elo ratings, these are the toughest GoDs since 1970:
> > YEAR GROUP TEAMS AVG. ELO
> > 1970 3 England, Brazil, Czechoslovakia, Romania 1959
> > 2014 B Spain, Netherlands, Chile, Australia 1931
> > 1978 3 Brazil, Spain, Austria, Sweden 1923
> > 1974 2 Brazil, Yugoslavia, Scotland, Zaire 1915
> > 1978 1 Italy, Argentina, Hungary, France 1910
> > 2022 G Brazil, Switzerland, Serbia, Cameroon 1901
> > 2014 G Germany, Portugal, United States, Ghana 1896
> > 2002 F Argentina, England, Sweden, Nigeria 1895
> > 1978 4 Netherlands, Scotland, Iran, Peru 1894
> > 1994 E Italy, Ireland, Mexico, Norway 1889
> > 2018 E Brazil, Switzerland, Serbia, Costa Rica 1888
> > 1982 6 Brazil, Soviet Union, Scotland, New Zealand 1885
> > 1986 E West Germany, Denmark, Scotland, Uruguay 1885
> > 2022 E Spain, Germany, Japan, Costa Rica 1885
> > 2018 B Spain, Portugal, Iran, Morocco 1884
> >
> > Two from 2022 made the cut.
> What about 1982? Brazil,Argentina, Italy

First round groups only. By design a second group stage would be much more competitive, as many weaker teams would have been eliminated by then.

On Sunday, November 13, 2022 at 10:00:11 PM UTC-5, MH wrote:
> On 2022-11-13 17:21, Al Kamista wrote:
> > So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
> >
> > Based on Elo ratings, these are the toughest GoDs since 1970:
> Are you defining group of death as the ones with the highest average Elo
> ratings ?
> But what is a group of death, really? Should it not perhaps be the most
> competitive group, perhaps represented as the one with the smallest
> standard deviation of ratings ?

But it should be some combination of the mean and the standard deviation, no? A very balanced group with 4 middling teams would not be considered a Group of Death. Nor would a group with two super-heavyweights and two middling teams.

On 2022-11-14 07:22, Futbolmetrix wrote:
> On Sunday, November 13, 2022 at 10:00:11 PM UTC-5, MH wrote:
>> On 2022-11-13 17:21, Al Kamista wrote:
>>> So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
>>>
>>> Based on Elo ratings, these are the toughest GoDs since 1970:
>> Are you defining group of death as the ones with the highest average Elo
>> ratings ?
>> But what is a group of death, really? Should it not perhaps be the most
>> competitive group, perhaps represented as the one with the smallest
>> standard deviation of ratings ?
>
> But it should be some combination of the mean and the standard deviation, no? A very balanced group with 4 middling teams would not be considered a Group of Death. Nor would a group with two super-heavyweights and two middling teams.

Good question. I think in most people's minds a group of death has some
alleged heavyweights in it, often judged more by reputation than current
form. There is also the issue of format. In 1986-94 , with 3 teams
advancing 2/3 of the of time, a group with three very strong teams was
not necessarily death to any of them.

Would a better definition of a group of death be one in which at least
one team ranked in the top 50 % of teams remaining in the tournament
must necessarily be eliminated. Or maybe 40 % to make it more
stringent. Or they have to rank in the top half in both Elo and FIFA
rankings ?

How would you combine mean and SD ?

>

Al Kamista has brought this to us :
> Based on Elo ratings, these are the toughest GoDs since 1970:

did you use the current ELO also for past GoDs or the calculationa has
been made using past ELOs?

--
/-\ /\/\ /\/\ /-\ /\/\ /\/\ /-\ T /-\
-=- -=- -=- -=- -=- -=- -=- -=- - -=-
............ [ al lavoro ] ...........

On Monday, November 14, 2022 at 10:22:57 AM UTC-5, MH wrote:
> On 2022-11-14 07:22, Futbolmetrix wrote:
> > On Sunday, November 13, 2022 at 10:00:11 PM UTC-5, MH wrote:
> >> On 2022-11-13 17:21, Al Kamista wrote:
> >>> So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
> >>>
> >>> Based on Elo ratings, these are the toughest GoDs since 1970:
> >> Are you defining group of death as the ones with the highest average Elo
> >> ratings ?
> >> But what is a group of death, really? Should it not perhaps be the most
> >> competitive group, perhaps represented as the one with the smallest
> >> standard deviation of ratings ?
> >
> > But it should be some combination of the mean and the standard deviation, no? A very balanced group with 4 middling teams would not be considered a Group of Death. Nor would a group with two super-heavyweights and two middling teams.
> Good question. I think in most people's minds a group of death has some
> alleged heavyweights in it, often judged more by reputation than current
> form. There is also the issue of format. In 1986-94 , with 3 teams
> advancing 2/3 of the of time, a group with three very strong teams was
> not necessarily death to any of them.
>
> Would a better definition of a group of death be one in which at least
> one team ranked in the top 50 % of teams remaining in the tournament
> must necessarily be eliminated. Or maybe 40 % to make it more
> stringent. Or they have to rank in the top half in both Elo and FIFA
> rankings ?
>
> How would you combine mean and SD ?

Not sure, but here is a different (and somewhat arbitrary) criterion: a group of Death is a group in which one of the top 5 teams by Elo rating entering the tournament has a more than x% chance of being eliminated at the group stage (probabilities based on my adaptation of the Elo formula, same one that I use for the SophCon). I happen to have the data for the 1998-2022 period. And let's set x at 25%. Based on this criterion, I have only 5 Groups of Death:

1. Netherlands 2014 (number 5 by Elo rating entering the tournament, Elo rating 1986, group with ESP-CHI-AUS). Elimination probability 39%
2. France 2018 (4th, 1987, AUS-DEN-PER): 27%
3. Argentina 2002 (2nd, 2015, ENG-NIG-SWE): 26%
4. Netherlands 2006 (2nd, 1995, ARG-SRB-CIV): 25%
5. Czech Republic 2006 (5th, 1972, ITA-GHA-USA): 25%

None of the 2022 groups make it. Let's compare with Alka's original list, taking out the the pre-1998 tournaments:

2014 B Spain, Netherlands, Chile, Australia 1931 --> Yes, right up there
2022 G Brazil, Switzerland, Serbia, Cameroon 1901 --> Brazil is too much stronger (8% elimination probabiliy) than the rest for this to be a GoD
2014 G Germany, Portugal, United States, Ghana 1896 --> Germany was at 17%
2002 F Argentina, England, Sweden, Nigeria 1895 --> Yes
2018 E Brazil, Switzerland, Serbia, Costa Rica 1888 --> Brazil 8%
2022 E Spain, Germany, Japan, Costa Rica 1885 --> Spain at 19%, Germany (not in the top 5) at 32%
2018 B Spain, Portugal, Iran, Morocco 1884 --> Spain at about 20%

On 2022-11-14 15:09, Futbolmetrix wrote:
> On Monday, November 14, 2022 at 10:22:57 AM UTC-5, MH wrote:
>> On 2022-11-14 07:22, Futbolmetrix wrote:
>>> On Sunday, November 13, 2022 at 10:00:11 PM UTC-5, MH wrote:
>>>> On 2022-11-13 17:21, Al Kamista wrote:
>>>>> So after agonizing for months about how I will manage to actually boycott this Abomination World Cup, I feel my will power weakening by the day. The football DNA is woven too deep into my molecules.
>>>>>
>>>>> Based on Elo ratings, these are the toughest GoDs since 1970:
>>>> Are you defining group of death as the ones with the highest average Elo
>>>> ratings ?
>>>> But what is a group of death, really? Should it not perhaps be the most
>>>> competitive group, perhaps represented as the one with the smallest
>>>> standard deviation of ratings ?
>>>
>>> But it should be some combination of the mean and the standard deviation, no? A very balanced group with 4 middling teams would not be considered a Group of Death. Nor would a group with two super-heavyweights and two middling teams.
>> Good question. I think in most people's minds a group of death has some
>> alleged heavyweights in it, often judged more by reputation than current
>> form. There is also the issue of format. In 1986-94 , with 3 teams
>> advancing 2/3 of the of time, a group with three very strong teams was
>> not necessarily death to any of them.
>>
>> Would a better definition of a group of death be one in which at least
>> one team ranked in the top 50 % of teams remaining in the tournament
>> must necessarily be eliminated. Or maybe 40 % to make it more
>> stringent. Or they have to rank in the top half in both Elo and FIFA
>> rankings ?
>>
>> How would you combine mean and SD ?
>
> Not sure, but here is a different (and somewhat arbitrary) criterion: a group of Death is a group in which one of the top 5 teams by Elo rating entering the tournament has a more than x% chance of being eliminated at the group stage (probabilities based on my adaptation of the Elo formula, same one that I use for the SophCon). I happen to have the data for the 1998-2022 period. And let's set x at 25%. Based on this criterion, I have only 5 Groups of Death:
>
> 1. Netherlands 2014 (number 5 by Elo rating entering the tournament, Elo rating 1986, group with ESP-CHI-AUS). Elimination probability 39%
> 2. France 2018 (4th, 1987, AUS-DEN-PER): 27%
> 3. Argentina 2002 (2nd, 2015, ENG-NIG-SWE): 26%
> 4. Netherlands 2006 (2nd, 1995, ARG-SRB-CIV): 25%
> 5. Czech Republic 2006 (5th, 1972, ITA-GHA-USA): 25%
>
> None of the 2022 groups make it. Let's compare with Alka's original list, taking out the the pre-1998 tournaments:
>
> 2014 B Spain, Netherlands, Chile, Australia 1931 --> Yes, right up there

And Spain did go out.
> 2022 G Brazil, Switzerland, Serbia, Cameroon 1901 --> Brazil is too much stronger (8% elimination probabiliy) than the rest for this to be a GoD
> 2014 G Germany, Portugal, United States, Ghana 1896 --> Germany was at 17%
> 2002 F Argentina, England, Sweden, Nigeria 1895 --> Yes

And Argentina did go out.

> 2018 E Brazil, Switzerland, Serbia, Costa Rica 1888 --> Brazil 8%
> 2022 E Spain, Germany, Japan, Costa Rica 1885 --> Spain at 19%, Germany (not in the top 5) at 32%
> 2018 B Spain, Portugal, Iran, Morocco 1884 --> Spain at about 20%
>
>
Great stuff and very interesting. Under this definition I suspect
genuine groups of death will become rarer and rarer - more teams,
seeding based on rankings and UEFA rankings more convergent with Elo
than they used to be (or are they ?)

Going back in history, though, this definition might yield a lot of
groups of death because in a 16 team world cup at least one group will
have two top five ranked teams in it... often more because hosts were
seeded, along with holders. Add to that that teams from minor
confederations were (since the 24 team era, approximately) seeded by
Geography rather than by actual strength (there is a big difference
between getting Mexico or Jamaica, for instance)