Impressive how you manage to simultaneously underrate non-FIFA-window friendlies and overrate regular friendlies. To be fair, I have no problem with the former.
Because FIFA window friendlies overwhelmingly feature the top players of a country while the non ones do not. It's really not a hard concept to understand. I mean why not count U20 results and U17 results in the FIFA rankings as well. That is just as illogical IMO.
...often half-assing it and getting subbed out at halftime. Again, my point is that it's surprising that you understand the drawbacks with taking irregular friendlies seriously, but not the issues with taking regular friendlies seriously. Weighing them as equal to confederation qualifiers is non-sensical.
Yes I agree that weighing friendlies and continental qualifiers the same isn't ideal, but there is no other way to gauge strength between confederations except the WC every 4 years. So IMO the benefit of assessing strength between confederations outweighs separating friendlies and qualifiers. Until we have a global nations league I think it is important for this to be done.
Three teams? Morocco, Cameroon, and who else? Other than those two, Comoros's most recent draws are against Seychelles, Kenya, Madagascar, Mauritius, Gabon, and Togo. They got a scoreless draw hosting Ghana in 2015. I don't think they're a Top 100 team. We'll have to agree to disagree.
Yes they are and that is by design. The way new FIFA ranking is built means it will need quite a lot of results for there to be any kind of big moves in position. And that comes from the complaints that was constantly made against the old formula.
Why would you need to do that? Unless you use relative strength as a factor in the ranking calculations then it is irrelevant to gauge it. Besides the old FIFA ranking did factor in relative strength, it was probably the one thing that got the most complaints (certainly from non-UEFA people).
The core concept of any Elo-based system (like FIFA's) is that the difference in rating points represents the likelihood that the higher rated team will beat the lower rated team. This model has proven pretty accurate in chess where it originated. See: https://en.chessbase.com/post/sonas-overall-review-of-the-fide-rating-system-220813 FIFA chose a variant of the model using 600 in the formula instead of 400. This implies that a team has a 40% chance of winning if it is 106 points lower, a 25% chance of winning if its rating is 286 points lower and a 10% chance of winning if it is 573 points lower. Spain is rated 1594. Does Ukraine (1490) have a 40% chance of beating Spain? Does Zambia (1308) have a 25% chance of beating Spain? Does Tahiti (1020) have a 10% chance of beating Spain? I think the answer in all three cases is no. If they had used 400 instead of 600, the chances would have been 36% (Ukraine), 16% (Zambia), and 3.5% (Tahiti). These seem better to me. Notice that there's a bigger change in probability at the extremes (Tahiti) than for more closely matched teams (Ukraine). FIFA's formula uses nine different values for I, ranging from 5 to 60. This represents the amount of points at stake in each game. It would be better if there were only two values, one for when you care about the result (all competitive games, say I=40) and one when you don't (all others, say I=10). Does anyone really think that Sweden tried less than half as hard in their playoff against Italy (I=25) as they did in their QF game against England (I=60)? The treatment of knockout round losers is also bizarre. Losses to Croatia and Belgium mean England were not as strong as they seemed before those games were played. This should be reflected in their rating. Considering only the Sweden and Colombia games makes no sense. If this rating system had been in place in 2014, Brazil's 1-7 loss would have had no effect on Brazil's rating. FIFA seems confused about whether its ranking system is meant to measure the comparative strength of each team or to reward the achievements like progressing at major tournaments. Either system can be good but FIFA has produced a mishmash. The correct way to judge FIFA's system is on its accuracy at predicting future results. Proposed changes to the system (vary the k-factor based on the competition or the score or home/away or...) should be evaluated according to whether they would improve that accuracy.
I agree with your point, but I have a separate comment. In all variations of ELO, there are two probabilities, In soccer, draws are possible. If the rankings are 60% for Spain and 40% for Ukraine, those aren't the probabilities unless you assume extra time and penalty kicks if necessary and declare a winner and a loser regardless of the length. Having a 40% chance at beating a team if a winner must be determined means you have less than a 40% chance at beating that team if draws are possible. It would be great to have a spreadsheet of results that could be sorted by rating difference before the game to break rating gaps into groups such as 50 to 100 points and look at the records of the better ranked team in those games.
And then there is of course the factors that cannot easily be quantified. Replace Spain and Ukraine with Argentina and Bolivia. A mathematical equation could suffice and tell you that maybe they would beat Bolivia 6 out of 10 times in theory. Maybe even 7 out of 10. But if I told you the game were to be played at the Altitude of La Paz... ....all bets are off. Argentina usually plays like crap at altitude and no ELO or weird statistics can tell you what will happen in that match.
Surprising that none of Pekerman's successors copied his successful gamble of calling up Argentineans playing in the Bolivian league for that one game and coming out with the win.
The FIFA Rankings used to be released every month except for excluding July during the World Cup. Last year they skipped another month when they changed the formula. I wanted to see if January rankings were out or would come out this week, and I see that there won't be any. Now the frequency is down to 8 times a year. The 2019 schedule is: February 7 April 4 June 14 July 18 September 19 October 24 November 28 December 19
Makes sense - with rankings only changing when actual games are played, no need for releases in the fallow months of the international calendar.