Interesting points. Your point about Germany starting having problems a year ago is food for thought. Something I've been looking at for college Division I women's soccer is how to spot when a long-term upward or downward slide starts for a team. I'm seeing that it often starts well before it becomes highly apparent. A good rating system can show when it started, from hindsight, but there's enough variability from year to year that trying to use short-term rating changes isn't a good a basis for early spotting of a trend. Of course, for college soccer, there's almost 100% player turnover every four years, which accounts for fairly high variability from year to year, so it takes a long time to decide whether what's happening is only normal variability or is a long-term trend. For national teams, the turnover period is much longer, so one would expect the year-to-year variability to be less. Indeed, Elo-based systems assume some level of continuity for teams, with the extent of assumed continuity determined by the K-factor -- one way to look at higher K-factors is that they assume a high level of variability so that recent games should be highly weighted. In terms of your reference to "the inertia of the ratings," that sounds like a suggestion that recent games aren't weighted highly enough, i.e., the K-factor assigned to the games isn't high enough. I've done studies for DI soccer, to determine what level of K-factor produces ratings that correlate best with game results. The K-factor my studies came up with is 70. From a FIFA rating system perspective (you used a K-factor of 30, which I assume is what FIFA uses for the SBC games), the 70 is extremely high. I assume that at some point FIFA has done its own studies that justify the K-factors it uses for games between national teams. With that in mind, I wouldn't use the term "the inertia of the ratings." Rather, I'd refer to "the inertia of the teams," meaning that at the international level, teams' levels of performance ordinarily change very slowly. In that respect, the changes in the four teams' ratings based on SBC, which are pretty minimal, reflect the teams' inertia. Or, to quote you, "the teams' strengths are actually fairly consistent and are simply better quantified now than they were before." Recognizing the normal inertia of teams provides a good way for fans to "check" their emotional responses to games to see whether their emotional responses have validity as measures of teams' true strength. As is apparent from posts here on BS, fans have a tendency towards "the sky is falling" when their team has a poor result or some poor results or even looks bad while getting a good result. According to the ratings, it's pretty highly likely that's an overreaction. Except in extreme situations, the sky simply doesn't fall. There's too much team inertia.
Alright, let's crunch the numbers for Cyprus now... FIN (1708) 0-1 (1635) WAL, K=15 :: exp. 0.604/0.396 :: act. 0.150/0.850 :: pts -7 (1701) / +7 (1642) ITA (1865) 3-0 (1860) SUI, K=15 :: exp 0.507/0.493 :: act. 0.960/0.040 :: pts +7 (1872) / -7 (1853) WAL (1642) 0-3 (1872) ITA, K=15 :: exp 0.210/0.790 :: act. 0.040/0.960 :: pts -3 (1639) / +3 (1875) SUI (1853) 4-0 (1701) FIN, K=15 :: exp. 0.706/0.293 :: act. 0.970/0.030 :: pts +4 (1857) / -4 (1697) FIN (1697) 2-2 (1875) ITA, K=15 :: exp 0.264/0.736 :: act. 0.510/0.510 :: pts +4 (1701) / -3 (1872) SUI (1857) 0-0 (1639) WAL, K=15 :: exp. 0.778/0.222 :: act. 0.470/0.470 :: pts -5 (1852) / +4 (1643) BEL (1788) 1-2 (1642) CZH, K=15 :: exp. 0.699/0.301 :: act. 0.160/0.840 :: pts -8 (1780) / +8 (1650) AUT (1810) 0-2 (1876) ESP, K=15 :: exp. 0.406/0.594 :: act. 0.080/0.920 :: pts -5 (1805) / +5 (1881) CZH (1650) 0-2 (1805) AUT, K=15 :: exp. 0.279/0.721 :: act. 0.080/0.920 :: pts -3 (1647) / +3 (1808) ESP (1881) 0-0 (1780) BEL, K=15 :: exp. 0.641/0.359 :: act. 0.470/0.470 :: pts -3 (1878) / +2 (1782) BEL (1782) 2-0 (1808) AUT, K=15 :: exp. 0.448/0.552 :: act. 0.920/0.080 :: pts +7 (1789) / -7 (1801) ESP (1878) 2-0 (1647) CZH, K=15 :: exp. 0.791/0.209 :: act. 0.920/0.080 :: pts +2 (1880) / -2 (1645) SVK (1502) 0-0 (1435) RSA, K=15 :: exp. 0.595/0.405 :: act. 0.470/0.470 :: pts -2 (1500) / +1 (1436) HUN (1538) 0-2 (1957) DPK, K=15 :: exp. 0.082/0.918 :: act 0.080/0.920 :: pts -0 (1538) / +0 (1957) DPK (1957) 1-0 (1500) SVK, K=15 :: exp. 0.933/0.067 :: act. 0.850/0.150 :: pts -1 (1956) / +1 (1501) RSA (1436) 1-0 (1538) HUN, K=15 :: exp. 0.357/0.643 :: act 0.850/0.150 :: pts +7 (1443) / -7 (1531) RSA (1443) 0-0 (1956) DPK, K=15 :: exp. 0.050/0.950 :: act. 0.470/0.470 :: pts +6 (1449) / -7 (1949) HUN (1531) 1-1 (1501) SVK, K=15 :: exp. 0.543/0.457 :: act. 0.500/0.500 :: pts -1 (1530) / +1 (1502) HUN (1530) 0-2 (1701) FIN, K=15 :: exp. 0.272/0.728 :: act. 0.080/0.920 :: pts -3 (1527) / +3 (1704) SVK (1502) 2-5 (1645) CZH, K=15 :: exp. 0.305/0.695 :: act. 0.056/0.944 :: pts -4 (1498) / +4 (1649) WAL (1643) 1-1 (1801) AUT, K=15 :: exp. 0.275/0.725 :: act. 0.500/0.500 :: pts +3 (1646) / -3 (1798) RSA (1449) 1-2 (1789) BEL, K=15 :: exp. 0.124/0.876 :: act. 0.160/0.840 :: pts +1 (1450) / -1 (1788) DPK (1949) 2-1 (1852) SUI, K=15 :: exp. 0.636/0.364 :: act. 0.840/0.160 :: pts +3 (1952) / -3 (1849) ITA (1872) 0-2 (1880) ESP, K=15 :: exp. 0.488/0.512 :: act. 0.080/0.920 :: pts -6 (1866) / +6 (1886) Overall: DPK -5pts, 1952 ESP +10pts, 1886 ITA +1pt, 1866 SUI -11pts, 1849 AUT -12pts, 1798 BEL -0pts, 1788 FIN -4pts, 1704 CZH -1pt, 1649 WAL +11pts, 1646 HUN -11pts, 1527 SVK -4pts, 1498 RSA +15pts, 1450 ....I messed something up. Thanks to the spate of 0-0 draws, we were supposed to lose 3pts overall, but those numbers above total losing 11pts overall. Shit. Well, there's definitely gonna be some small errors spattered about when I put all 28 countries into the new rankings after I calc' the Algarve matches....
Siberian, I remembered something related inertia and Elo-based ratings, from my work on DI women's soccer. Even using a very high K-factor (and thus a high weighting of recent games), at the end of the season I found outliers. Some of the outliers were teams that had made major improvements in performance, which were significantly underrated from a statistical perspective, and others were teams with major declines, which were significantly overrated. In DI, the average team plays 19 to 20 games in the season. That works OK for a team whose performance is within a normal range of improvement or decline, but its not enough games for a team with a major improvement or decline. This is related to the rule of thumb for Elo-based ratings, which is that it takes roughly 30 games for a new participant to get properly rated. In the DI women's soccer situation, the team with a major improvement or decline is somewhat similar to a new participant, and at 19 to 20 games it simply hasn't played enough games to move to its "proper" rating position. For national teams, there also is the possibility of a rather sudden major improvement or decline. This is not as likely as for DI women's soccer, but it can happen. Perhaps Australia is a current case (notwithstanding its Algarve performance)? If they have had a major improvement, it will take quite a few games before they will reach their proper rating position. Thus you're right, there can be inertia in the ratings. And, the lower the K factor, the greater the inertia, but it also can be there with a very high K factor. With FIFA using relatively low K factors, I could see it taking 20 to 30 games for a team's rating to become "correct" after it has gone through a major improvement or decline. For national competition, that's a pretty long time, especially for some teams.
Yeah, that's why I brought up Germany's decline in my earlier post. There was writing on the wall as early as the 2015 WWC for them, but it's wasn't until after the Olympics that a lot of pieces shifted and they suddenly - and consistently - started underperforming versus what they would historically expect, having trouble in the EURO, in WWCQ, and at SBC. However, despite the definite step downward back between the Olympics and the EURO, it was only after SBC that they've officially fallen out of the #2 rank. That was why I definitely feel there's inertia in the ratings themselves - and, as you point out, it's dependent on the K factor. I think the thing with the rankings is you have to balance the ranking's inertia so it matches the teams' inertia. NTs certainly have less turnover and more consistency than college teams, so a lower K value would probably make sense, but at the same time you'd want a K value big enough so that teams that rise or fall smoothly (NED and maybe AUS growing, and JPN declining) have their rating rise/fall mostly in lock-step with their actual performance. Every Elo system will have some inertia to it, unless you tune it to the point that it's very volatile, so sudden increases or drops (e.g. Germany) will almost never be fully represented. But since NTs are usually consistent, or at least have moderate rates of growth and decline, you want the inertia so any momentary bumps or dips in form don't push the ratings too out-of-line.
I am surprised Italy was punished so much by their loss in the final to Spain. At the beginning of the tournament there was a 16 points gap between them (1865 to 1881); by the time of the final it had been reduced to just 8 points (1872 to 1880), i.e. half of what it previously was in just a few days and three matches! A disclaimer: unlike many posters on this thread, I am completely incompetent as regards the system of calculation of FIFA ranking (but, unlike other posters, I quietly admit that ), so I am curious: why Italy is punished so harshly for a 0-2 loss to a team that was anyway ahead of them (if by a few points), to the point that, in front of a good performance at Cyprus Cup, they only advance by 1 point overall in FIFA rankings? I don't think the last match of Cyprus Cup was considered a "final" (and anyway the whole tournament, if I get it right, was just a set of friendlies for what concerns FIFA rankings), so I don't think there were particular multipliers involved. Is it because, being the teams so close in points (8 points is basically nothing) the system was strongly expecting a draw between them? Did the actual score count? a 0-1 loss would have been more forgiving? Also, let's imagine the same scores (ignoring any consideration about the formula of the tournament, it's just a mental experiment), but Italy losing to Spain at the beginning of the tournament, when the gap between them was 16 points: would it have changed anything? Thanks if anyone wants to take the time to explain that for dummies.
SiberianTT is the expert on this. On checking, however, goal differential and goals scored do play a part in the rating process. Also, 8 points is not a big change.
You've got most of the reasoning down already. Yes, the two teams were very closely rated at the beginning of the match, so a draw was expected. You're also right that the 0-2 decision hurt them more than an 0-1 decision would have. However, an 0-1 decision is still worth 85% of a match, so there still would have been a noticeable points shift - still 4 or maybe even 5, depending on rounding. (I don't have my calculator right now.) Considering that the traditional, non-soccer Elo ratings are a straight 0% for a loss, 50% for a draw, and 100% for a win, the FIFA Womon's Rankings are designed to have a big jump between a draw (~50% depending on the score) and a win (at least 80% for a 1-goal win, and at least 92% for a 2-goal win). TBQH, I think that even if the same score happened at the beginning of the tournament, you still would have had a 4- or 5-point swing, simply because of that big gap in the "actual result" between a draw and a win. Might have been only 3pts since the "expected result" equation is very non-linear near the middle, but I'd have to run the numbers exactly to make sure. Think of it this way: the ratings like having more matches. If Italy and Spain played, say, 5 or 10 matches, you'd have a mix of results that probably combine to nearly an even split of points and GD, with a slight edge to Spain. However, since they only played one match and Spain won it, Spain reaps benefits that might have been scaled back over a longer series of matches.
Alright, time for me to finally do the Algarve, right? The actual rankings will come Firday, so I gotta finish my predictions quickly! As I mentioned in an earlier post, I'm going to assume the "Top Ten" designation (as it applies to K=15 vs K=30) is something that stays with a team from the most recent release, even if their transient rating drops them below the #10 spot for any given time before the next release. This means that the Day 1 SWE-CAN game will have K=30 when I do the calculations below. If this assumption is false and DPK earned the "Top Ten" designation immediately after the EAFF tournament, then any calculation involving SWE, CAN, or any teams they later play will be incorrect - but still within a few points of what would be correct. POR (1598**1698) 2-1 (1871) CHN, K=15 :: exp. 0.270/0.730 :: act. 0.840/0.160 :: pts +9 (1607) / -9 (1862) AUS (2030) 4-3 (1877) NOR, K=15 :: exp. 0.707/0.293 :: act. 0.820/0.180 :: pts +2 (2032) / -2 (1875) POR (1607**1707) 0-0 (2032) AUS, K=15 :: exp. 0.133/0.867 :: act. 0.470/0.470 :: pts +5 (1612) / -6 (2026) CHN (1862) 0-2 (1875) NOR, K=15 :: exp 0.481/0.519 :: act. 0.080/0.920 :: pts -6 (1856) / +6 (1881) POR (1612**1712) 2-0 (1881) NOR, K=15 :: exp. 0.274/0.726 :: act. 0.920/0.080 :: pts +10 (1622) / -10 (1871) AUS (2026) 2-0 (1856) CHN, K=15 :: exp. 0.728/0.272 :: act 0.920/0.080 :: pts +3 (2029) / -3 (1853) KOR (1862) 3-1 (1720) RUS, K=15 :: exp. 0.694/0.306 :: act 0.911/0.089 :: pts +3 (1865) / -3 (1717) CAN (2023) 1-3 (1955) SWE, K=30 :: exp. 0.597/0.403 :: act. 0.089/0.911 :: pts -15 (2008) / +15 (1970) RUS (1717) 0-1 (2008) CAN, K=15 :: exp. 0.158/0.842 :: act. 0.150/0.850 :: pts -0 (1717) / +0 (2008) SWE (1970) 1-1 (1865) KOR, K=15 :: exp. 0.647/0.353 :: act. 0.500/0.500 :: pts -2 (1968) / +2 (1867) SWE (1968) 3-0 (1717) RUS, K=15 :: exp. 0.809/0.191 :: act. 0.960/0.040 :: pts +2 (1970) / -2 (1715) KOR (1867) 0-3 (2008) CAN, K=15 :: exp. 0.308/0.692 :: act. 0.040/0.960 :: pts -4 (1863) / +4 (2012) JPN (1958) 2-6 (1965) NED, K=30 :: exp. 0.490/0.510 :: act. 0.044/0.956 :: pts -13 (1945) / +13 (1978) DEN (1897) 0-0 (1814) ISL, K=15 :: exp. 0.617/0.383 :: act. 0.470/0.470 :: pts -2 (1895) / +1 (1815) JPN (1945) 2-1 (1815) ISL, K=15 :: exp. 0.679/0.321 :: act. 0.840/0.160 :: pts +2 (1947) / -2 (1813) DEN (1895) 2-3 (1978) NED, K=15 :: exp. 0.383/0.617 :: act. 0.170/0.830 :: pts -3 (1892) / +3 (1981) JPN (1947) 2-0 (1892) DEN, K=15 :: exp. 0.578/0.422 :: act. 0.920/0.080 :: pts +5 (1952) / -5 (1887) ISL (1813) 0-0 (1981) NED, K=15 :: exp. 0.275/0.725 :: act 0.470/0.470 :: pts +3 (1816) / -4 (1977) CHN (1853) 2-1 (1715) RUS, K=15 :: exp. 0.689/0.311 :: act. 0.840/0.160 :: pts +2 (1855) / -2 (1713) ISL (1816) 1-1 (1887) DEN, K=15 :: exp. 0.399/0.601 :: act. 0.500/0.500 :: pts +2 (1818) / -2 (1885) CAN (2012) 2-0 (1952) JPN, K=30 :: exp. 0.585/0.415 :: act. 0.920/0.080 :: pts +10 (2022) / -10 (1942) AUS (2029) 1-2 (1622**1722) POR, K=15 :: exp. 0.854/0.146 :: act. 0.160/0.840 :: pts -10 (2019) / +10 (1632) Overall: CAN -1, 2022 AUS -11, 2019 NED +12, 1977 SWE +15, 1970 JPN -16, 1942 DEN -12, 1885 NOR -6, 1871 KOR +1, 1863 CHN -16, 1855 ISL +4, 1818 RUS -7, 1713 POR +34, 1632 Ha! The net exchange of points was -3, which makes sense with exactly three 0-0 draws and no multi-goal draws to balance those out! (Of course, there's the SWE "Top Ten" issue that might still be throwing things off...) Gonna compile all three lists in the next post to see what movement we might expect in the release this Friday...
STT's predicted rankings + ratings for the top 25 (insert caveats about messed-up Cyprus #s and the SWE Top Ten issue): 1. == USA : 2116 (2114 +2) 2. << ENG : 2042 (2033 +9) 3. >> GER : 2033 (2052 -19) 4. << FRA : 2024 (2014 +10) 5. == CAN : 2022 (2023 -1) 6. >> AUS : 2019 (2030 -11) 7. == NED : 1977 (1972 +5) 8. << SWE : 1970 (1955 +15) 9. >> BRA : 1968 10. << DPK : 1952 (1948 +4) 11. >> JPN : 1942 (1967 -25) 12. << ESP : 1886 (1869 +17) 13. >> DEN : 1885 (1899 -14) 14. == NOR : 1871 (1868 +3) 15. << ITA : 1866 (1860 +6) 16. >> KOR : 1863 (1868 -5) 17. >> CHN : 1855 (1861 -6) 18. >> SUI : 1849 (1860 -11) 19. << ISL : 1818 (1818 =0) 20. >> NZL : 1816 (1830 -14) 21. << SCO : 1799 (1785 +14) 22. >> AUT : 1798 (1810 -12) 23. >> BEL : 1788 (1788 =0) 24. == COL : 1756 (1756 =0) 25. << MEX : 1724 26. >> RUS : 1713 (1730 -17) March Summary: Almost no change for USA, CAN, NOR, ISL, or BEL. Big winners: FRA, SWE, ESP, WAL, POR, and RSA. Big losers: GER, AUS, JPN, DEN, SUI, RUS, and HUN. ENG/GER switch spots. FRA/AUS switch spots around CAN. #6 is still within 100pts of #1! DPK/SWE switch spots before SWE storms back in to the Top Ten to fill the void left by a crashing JPN (and BRA has no clue what's going on). ESP/DEN switch spots over what had been a quite sizeable gap! Big gaps still exist between #6/#7 and #11/#12. The mid-1800's are a cluster**** and ITA takes full advantage of it. SCO also makes a nice jump up at the expense of NZL.
http://www.fifa.com/fifa-world-rank...king.html?_branch_match_id=381383981078809024 Leader USA (unchanged) Moves into top 10 Korea DPR (10th, up 1) Moves out of top 10 Japan (11th, down 2) Matches played in total 108 Most matches played China PR (9 matches) Biggest move by points Portugal (up 32 points) Biggest move by ranks Israel (up 3 ranks) Biggest drop by points Germany (down 19 points) Biggest drop by ranks Montenegro (down 8 ranks) Newly ranked teams Paraguay, Côte d’Ivoire, Uruguay, Morocco, Senegal, and Burkina Faso Teams no longer ranked due to prolonged inactivity El Salvador, Ethiopia, Nicaragua, FYR Macedonia, Tanzania, Rwanda, and Uganda
on behalf of Lechus, blissett, Gilmoy & all other Nadeshiko fans(which means probably whole BS); this is an outrage, Japan, should never be drop out of the top ten!) but on behalf of me; HIP HIP HOORAY!
Really? Well, I am happy you credit so many friends to us. Great! Anyway, it's true that reading that this is the first time Japan is out of top 10 since 2007 stung alot.
Alright, let's see... Clearly my assumption about how FIFA treats its Top Ten ranking is incorrect, since I apparently doubled the points shift between CAN and SWE. Not sure why NED and JPN are so far off, though - JPN shouldn't have been out of the Top 10 by their match, even though that would explain the difference... Although I can partially explain JPN since they would have been out of the Top 10 when they faced *CAN*, so that saves them points - but still doesn't explain NED, and would over-explain CAN. Not sure why I missed SCO so badly, either, especially considering I got NZL within rounding. 1. USA : 2115 v 2116 (r) 2. ENG : 2042 3. GER : 2033 5. FRA : 2024 v 2025 (r) 4. CAN : 2029 v 2022 (+7 v me) 6. AUS : 2018 v 2019 (r) 7. NED : 1972 v 1977 (-5 v me) 9. SWE : 1963 v 1970 (-7 v me) 8. BRA : 1968 10. DPK : 1955 v 1952 (+3 v me) 11. JPN : 1950 v 1942 (+6 v me) 12. ESP : 1886 13. DEN : 1884 v 1885 (r) 14. NOR : 1871 15. ITA : 1864 v 1866 (-2 v me) 16. KOR : 1860 v 1863 (-3 v me) 17. CHN : 1855 18. SUI : 1849 19. ISL : 1819 v 1818 (r) 20. NZL : 1815 v 1816 (r) 22. SCO : 1793 v 1799 (-6 v me) 21. AUT : 1800 v 1798 (+2 v me) 23. BEL : 1788 24. COL : 1754 v 1756 (-2 v me) 25. MEX : 1724 26. RUS : 1713 A bit embarrassing to have so much, including things not shown here (e.g. POR's +32 pts), so close to correct, and then miss not one but two rank orders just within the top ten (FRA/CAN and SWE/BRA), even if both errors are easily explained by the wrong assumption about the Top 10 designation in calculations... =-/ I'm guessing that whatever error I had in the Cyprus Cup calculations that lost me 8pts from nowhere is why DPK, ITA, and AUT are minorly off - and it was a FIN match somewhere, since poor FIN is 1691 instead of the 1704 I predicted. My KOR prediction was minorly off as a result of the CAN/SWE mistake.
meh..It was bound to happen for Japan considering their results of late so outrage?...really? Heh, Japan was 11th in 2007 when Sasaki took over after WWC in China (though having core team already selected by Ohashi for WWC 2007). In 4 years time he raised them to top 4. Takakura has just finished her selection for generation change. Now we will see how far Japan will go under her in next 2 years (Tokio OG 2020)
naw, my dad would of been serving in the Dutch Royal Air Force at about that time getting kicked much like U Frenchies did, lol played right after the last FIFA rankings. this competitive game(only 3 subs allowed) played on Japan's home ground should point out why NKR deserves the higher ranking. Then just look at their last 5 game record(NKR unbeaten since whereas JA had 3 out of 5 losses, all of them by 2 goals or more) https://us.women.soccerway.com/matches/2017/12/15/asia/eaff-east-asian-cup/japan/korea-dpr/2589181/ the J fans going to tell ya their were missing this or that player, but it's not like NKR have any hardships either, lol this vids way too long, but just looking at the first few seconds should tell you where woso is heading(the CA steal), Personally I don't think Canada deserves FIFA's #4 place on quality; starting fwds Beckie & Prince scored 5 goals between them last season for the poor performing NWSL/Dash, but it goes to show more how winning by physical strength is the IN thing these days'
Once again, you're assuming one match proves a ranking. Also, no-one was arguing that DPK didn't deserve the higher ranking. But it's due to many matches together, not that one match. I had done all the calcs at the end of the EAFF tournament, and right after the last rankings came out. Go back to the previous page and you'll see two huge posts from me all about it. And then I explicitly did the calcs again last week. Yes, DPK would gain ground and JPN would lose ground with that results - and, of note, PDK passed SWE with that result to return to the Top Ten. HOWEVER, thanks to the gap that had already been there between JPN and DPK, the results shouldn't have swung enough points to put JPN below both DPK and SWE. Even if I rounded incorrectly and it was a +10/-10 swing instead of +9/-9, JPN would have still been a handful of points above SWE. The only way for JPN to have fallen out of the Top Ten by their match against NED would be if SWE had passed them earlier in the day thanks to the SWE-CAN match - except that the JPN-NED match was played more than three hours earlier than the SWE-CAN match. In other words, I honestly think FIFA just tried to calculate all the matches on the same day in one bunch, and their algorithm ran with some order that wasn't time-of-day.
Siberian, there's something I don't get about what you're saying. The way an Elo-based system works, all that matters for post-game ratings, when two teams play each other, is what the two teams' ratings were going into the game. All the other teams' ratings don't adjust based on what those two teams' ratings were. So if there are two games on the same day, it doesn't matter what the order of the games are. The post-game ratings based on the two games are independent of each other. Is there something I'm missing here?
For most Elo system, yes. However, the FIFA women's rankings have the "Top Ten" multiplier, so for a small number of "friendly" games, it's not only the rating that matters going in to a game, but also the ranking. This is the reason why, in all of the calculation posts I've posted before, most games have K=15 but some (including all of the SBC games) have K=30. Friendlies between two teams within the top 10 are treated as twice as important as friendlies where one or both teams aren't in the top 10. As such, it in theory matters for a game between two teams in or near the top ten, if movement elsewhere in the rankings causes one of those two teams to cross the 10/11 barrier simply by another team rising or falling. It's the one very very non-Elo component of the FIFA women's rankings. It also rarely matters since there usually aren't days of the year where nearly every team in the top 30 is playing... But the proliferation of the spring tournaments has apparently made it an issue.
Thanks for that piece of info. I wonder if the FIFA system computes the ratings daily, which in turn also would determine the rankings daily. That's how I would do it, even if I were publishing ratings/rankings only monthly. If I recall correctly, that's a question you've wondered about on line. I really appreciate your willingness to give good explanations.
After Japan won AFC Asian Cup today in a final vs FIFA-6th-place-ranked Australia, can I safely assume that they're probably going to get back in top 10 at the next rankings' update? Continental competitions have a multiplier factor similar to WWC and Olympics? Also: let's assume Australia lost some boatloads of points from their draw vs Thailand in the semi-final: will this influence the entity of Japan's gain for their win vs Aussies? In other words, a team's point/position get updated in-between tournaments? Or just at the end of it? Thanks for any gracious answer to my doubts by usual statistical gurus.
Japan will have lost some points for having drawn KOR, but earned some points for drawing and then beating AUS, and for beating CHN. The amount of points JPN gets for winning the final will be less than maybe expected, since AUS' draw with THA might have bumped AUS to already be below JPN - so yes, ratings are updated after every individual game - but I don't think it'll be way less than expected, since we're not talking 60+ points of difference from before and after the THA draw. In contrast, this tournament (and all WWCQ tournaments) are worth 3x what a friendly is worth, so you'll see a lot of points overall transferred. Only the WWC itself and the Olympics are worth more (4x). But to answer your original question: yes, I fully expect JPN back in the Top Ten after this tournament.
I genuinely appreciated each and every little detail of your response, but this was the most important part to me.