Test to compare this python script with updates done by @Martin0:
Previous ratinglist
captaintugwash,1889
HarshSaberTwitch,1686
JustARandomPatzer,1634
Marks1420,1533
comradM,1502
musketeerchess2017,1481
evert823,1349
1e4c6_O-1,1201
Older ratings which I also used in this test
Rook_Handler,1264
DukeOfHelsinki,1473
haoming09,1443
JamesAgadir,1441
Ilampozhil25,1251
aronian22,1246
PandasAwesome,1203
Games used for this update cycle - see the referenced post
Script output
captaintugwash,1900
HarshSaberTwitch,1686
JustARandomPatzer,1634
comradM,1607
DukeOfHelsinki,1494
JamesAgadir,1492
musketeerchess2017,1481
VM_LeaderAbhyudaya,1430
Marks1420,1375
haoming09,1344
evert823,1326
Puzzles,1281
KapitalP_Pro,1265
marcellothearcane,1264
Rook_Handler,1231
BattleChessGN18,1231
Ilampozhil25,1226
aronian22,1225
PandasAwesome,1201
1e4c6_O-1,1201
Lovely_Kitty19,1200
goodatchessplus,1200
There are deviations of 1 point. Only for comradM there is a deviation of 6 points.
(Also, I noticed a UC/LC typo in Martins post just for comradM)
https://github.com/evert823/forumgameratings
With the script in this repo it should be doable to start maintaining ratings for competitions, for example chess variant games in forums.
I have tried to copy the approach from the Bulldog chess ratings.
In order to test this, I have used archives of Bulldog games, and compared the results that this script calculates, with Bulldog ratings that have been published.
The observation is that this script comes to the same result with minor deviations (very often there is one point difference).
The repo contains testfiles with fictional games and players.
In order to use this yourself, you would require sufficient skills to
- install python, pip, pandas
- download code from Github
- manipulate csv files so that it aligns with what the script expects
- start a terminal and run the Python mainscript
So a wide audience can potentially do this!