Life In 19x19
http://www.lifein19x19.com/

ELO rating vs. Ladder pos. (repost from DGS forum w. images)
http://www.lifein19x19.com/viewtopic.php?f=49&t=17540
Page 1 of 1

Author:  alisbarbe [ Sat Jun 06, 2020 6:00 am ]
Post subject:  ELO rating vs. Ladder pos. (repost from DGS forum w. images)

Hi, I repost my post from DGS forums([url]https://www.dragongoserver.net/forum/read.php?forum=11&thread=52070
[/url]) here, so I can add the images that I had posted on Facebook (see Alistair Fronhoffs https://www.facebook.com/groups/go.igo.weiqi.baduk/?multi_permalinks=10158652261481514&notif_id=1591265703312110&notif_t=group_highlights)

And also, get back a little more actively here on L19 ;)

From FB:
- There's a quite clear lower rank limit, for example 16 kyu's don't get higher than about 280th and a 6 kyu can hope for top 50.
It seems that in the top 100-200 positions there's some kind of correlation between rank and position, better players are higher placed and all are packed within about 5 stones.
Below 200th, players of all strenghts are competing to climb the ladder.

- Theoretical position based on rank is an internal DGS algorithm, more or less ordering players by their ELO rating as Matthew Frye (fb) supposed. But I don't know the details of the ELO-ladder position conversion.
I calculated the elo rating from the ranks because I think the first graph is easier to understand. Then plotted the theoretical position vs. position to see how that looks.

- so it seems strong players can blunder to drop lower than they "should" be but weak players never accidentally play "hand of god" moves :)
- Weak players can play "hand of god" moves, but such a move depends very much on the correct follow-up. Otherwise nothing comes of it.

- I don't know if those higher ranked players fell down the ladder, I think it's more probable that most of them are busy climbing up as all players start at the bottom of the ladder and start competing, challenging higher positioned players.

From DGS forum:
- Isn't the first graph what you would expect:
Roughly a straight line
with a few people slightly below it (who have done better than their rank suggests)
and a lot of people above it who have either just joined and are working their way up
or people who have stopped playing/winning and are drifting slowly down the ladder

You might expect the second graph to be the same (but inverted)
But I assume JUG just put everyone in order to get the theoretical position
So theoretical position vs rank is not a straight line, it's wiggly where
there is a higher (or lower) density of players near a particular level

- Indeed, it seems logical, but interesting to see where in the ladder ranks start to be more consistent with position ;)

Attachments:
theor_pos_vs_ladder_pos.pdf [69.3 KiB]
Downloaded 318 times
elo_vs_theor_pos.pdf [28.23 KiB]
Downloaded 320 times
elo_vs_ladder_pos.pdf [71.54 KiB]
Downloaded 323 times

Author:  BigElefant [ Sat Jun 06, 2020 6:51 am ]
Post subject:  Re: ELO rating vs. Ladder pos. (repost from DGS forum w. ima

alisbarbe wrote:
(...)
- so it seems strong players can blunder to drop lower than they "should" be but weak players never accidentally play "hand of god" moves :)
(...)


I'm not sure if strong players "blundering" their way down the ladder is what you really mean ;-). Looking at the ladder table, I think it's in many cases related to players being inactive and just dropping down because they're overtaken by active players.

Author:  alisbarbe [ Sun Jun 07, 2020 2:48 am ]
Post subject:  Re: ELO rating vs. Ladder pos. (repost from DGS forum w. ima

No, indeed it seems unlikely that they blunder down the ladder, it was one of the replies on FB :)

Page 1 of 1 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/