Forum for KGS Rating Math

Path: <= Rank =>

more updates for kgs 3 constants [#998]

Back to forum     Back to page

New reply

yoyoma: more updates for kgs 3 constants (2007-05-29 08:24) [#3408]

I finally got around to updating the "How many wins in a row to promote?" table with KGS 3 constants. The old table assumed k = 0.8 and half-life = 45 days.

I also noticed while redoing that I didn't state that when you go on your winning streak, it assumes you play one stone up. That is if you were 2.5d before, you start playing even games with 3.5d players and win them all. This probably isn't the normal case, and I think I'll probably change the table to do the normal case instead. It takes longer that way obviously (56 games for 2d+ instead of 20, and 12 games for 15k- instead of 6).

Calvin: Re: more updates for kgs 3 constants (2007-05-30 05:06) [#3412]

How about if you play even games? That may be more common.

yoyoma: ((no subject)) (2007-05-30 01:45) [#3411]

Actually on reviewing the math I did for the old results, it looks like it was done using 2.5d opponents both before and after. I think there is a mistake in either my old results or the new ones.

yoyoma: ((no subject)) (2007-05-30 08:01) [#3413]

I did some more testing on my new code, and found the problem. Actually it's more like I ditched the example code I was looking at because it simply doesn't appear to be correct...

The old results were done using 2.5d opponents before and after, and now my new results using the old k=0.8 halflife=45 days constants match.

Sorry for the confusion, hopefully this should be correct now.

Back to forum     Back to page

New reply

Path: <= Rank =>
Forum for KGS Rating Math
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library