[email protected] wrote: > Hi, > > first FIDE-Rating begins 1971: > > 1971-1989 one rating per year > 1990-2000 two ratings per year > 2001-2008 four ratings > 2009 five ratings > 2010-? six ratings > > > > Korchnoi, Viktor #gm URS/XXX/SUI [2695] 1931 > %Bio BornCity: Leningrad, URS > %Bio FIDEID 1300016 > %Bio VIAF 42629859 > %Bio PND 118565605 > = Korchnoi, Viktor Lvovich > = Korchnoi, Victor L > = Korchnoi, VL > = Kortchnoi, Viktor L > = Kortchnoi, Victor L > = Kortchnoi, VL > = Kortschnoi, Viktor L > = Kortschnoi, Victor L > = Kortschnoi, VL > = Kortschnoj, Viktor L > = Kortschnoj, VL > = Kortschnoy, Viktor L > = Kortschnoy, VL > = Korchnoi, Viktor L > %Elo 1971:2670 1972:2640 1973:2650 1974:2670 1975:2665 > %Elo 1976:2670 1977:2645 1978:2665 1979:2695 1980:2695 > %Elo 1981:2650 1982:2645 1983:2600 1984:2635 1985:2630 > %Elo 1986:2635 1987:2625 1988:2640 1989:2610 1990:2625,2630 > %Elo 1991:2615,2610 1992:2585,2575 1993:2605,2625 1994:2615,2615 > %Elo 1995:2635,2635 1996:2645,2635 1997:2635,2610 1998:2625,2625 > %Elo 1999:2673,2676 2000:2659,2620 2001:2639,2643,2624,2639 > %Elo 2002:2633,2635,2626,2634 2003:2642,2632,2628,2580 > %Elo 2004:2569,2579,2568,2601 2005:2617,2619,2615,2620 > %Elo 2006:2608,2607,2600,2607 2007:2629,2623,2610,2611 > %Elo 2008:2605,2598,2602,2584 2009:2552,2552,2561,2566,2567 > %Elo 2010:2566 >
Hi Franz, This part I also understood from both the ssp file and the code. Thanks anyway, but I was just wondering _why_ it deviates from the official elo time scheme. Can you explain that? It is not that important, but if I change the code, I might as well make it better... FIDE did produce a 3rd ratinglist for 2000, in October. It was in fact the very first quarterly list, which you can compare to the very fist bi-monthly list of November 2009. Why include the one and forget about the other? Korchnoi was 2632 on that list. You can check on ratings.fide.com Ratings before 2000 have not been published by FIDE, as you are well aware. www.olimpbase.org has a nice overview of ratinglists before that date. I was wrong in my first observation: Half-year lists have only been published since July 1981. I cannot really imagine that you are not aware of this source, but if so, have a look (http://www.olimpbase.org/Elo/summary.html). Korchnoi's rating in July 1981 was (back to) 2695. I will adapt the scid code such, that the additional data *can* be added, but not *must* be. E.g, if the 3rd rating of 2632 were to be added to the 2000 list of Korchnoi in the ssp file, his games played in Oct-Dec of 2000 would be re-rated to 2632. Hence, still no need to 'escape' ("?") missing (or, rather, anticipated) ratings at the tail of any year list, just like it holds for the year 2010 in the current ssp file. This may sound complex, but I think it actually will make the code simpler. More generic and a lot faster. One related question however: In general, how to re-rate the game for a player who has no rating identified for (exactly) that period? In reality this is not very likely to happen, but it is a matter of robustness and it touches upon the above. I see two options: 1) Take the most recent rating available at that time. This is probably too harsh, and incorrect. 2) Leave it unknown (blank) and let Scid do the interpolation, using the - back to Korchnoi's October 2000 rating - "(2620)" indication in brackets. I feel this works well and is a more reasonable solution. Regards, Joost. > Am Dienstag 12 Januar 2010 00:37:12 schrieben Sie: > >> Joost 't Hart wrote: >> >>> Joost 't Hart wrote: >>> >>>> Michal Rudolf wrote: >>>> >>>>> FIDE changed the number of rating list per year. Alas, Scid seems to >>>>> use 4- lists per year layout even for new games. For example, adding >>>>> Elo for December's Ivanchuk's games used 2756: >>>>> >>>>> Ivanchuk, Vassily >>>>> 2009:2779,2746,2703,2756,2739 >>>>> >>>>> For 2009, there were 5 lists, for next years there will be six. >>>>> >>>> Ok, I was contacted by Alexander and will take this on. >>>> >>>> For it to work properly, the code needs a bit of redesign; its >>>> exception-based behaviour will become too complex (it seems to be >>>> incorrect anyway at this moment), so I will change it into a more lookup >>>> based scheme. >>>> >>>> Let's agree on some sort of reality: >>>> >>>> Source: www.olimpbase.org >>>> >>>> Initial lists: >>>> 1971/7, 1972/7, 1973/7, 1974/5 >>>> >>>> Yearly lists (Jan): >>>> 1975 - 1977 >>>> >>>> Half-year lists (Jan + July): >>>> 1978 - 2000 >>>> >> Haha! See remark below... >> >> ssp file has only one rating for the years 1978 - 1989. >> Again, was this deliberately or...? >> >> I propose to do the corrections compatible with both my original >> proposal and the current ssp files. This means that historical ratings >> can still be added later on. >> >> >>>> Quarterly lists (Jan + Apr + Jul + Oct): >>>> 2000/10 - 2009/7 >>>> >>> Hm, >>> >>> Franz's ssp file seems to ignore the October 2000 list. >>> Correct? >>> >>> Deliberately, or is this the consequence of the Scid implementation? >>> >>> >>>> Bi-monthly lists (Jan + Mar + May + Jul + Sep + Nov): >>>> 2009/9 - ???? >>>> >>>> Does this sound familiar? Corrections welcome. If not, I'll give it a go >>>> shortly. >>>> >>>> Cheers, >>>> Joost. >>>> >>> ------------------------------------------------------------------------- >>> ----- This SF.Net email is sponsored by the Verizon Developer Community >>> Take advantage of Verizon's best-in-class app development support A >>> streamlined, 14 day to market process makes app distribution fast and >>> easy Join now and get one step closer to millions of Verizon customers >>> http://p.sf.net/sfu/verizon-dev2dev >>> _______________________________________________ >>> Scid-users mailing list >>> [email protected] >>> https://lists.sourceforge.net/lists/listinfo/scid-users >>> > > ------------------------------------------------------------------------------ This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev _______________________________________________ Scid-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scid-users
