First thing is put this report into a separate topic in the Feedback->Bug Report category of the forum.
Honestly, if there is a problem with just one game, I would just let it go…
I’d rather see it reported - it indicates there’s a bug, so there’s no knowing how many games are being affected that we don’t notice because we’re not really looking.
1 Like
If every correspondence player has a problem with “just one game”
…that’s a lot of games.
6 Likes
after playing this new system for a while, I TRULY HATE IT! it does NOT reward you for beating higher ranked opponents, and punishes you severely for losing against same or lower ranked opponents, if this system is supposed to be fair, i don’t see it, i’ll be hard pressed to continue here on ogs
What’s an example of a game where you were not rewarded for beating a higher ranked opponent?
3 Likes
Example of the Glicko-2 system (excerpt, page 2)
We now want to update the rating of a player with (Glicko-2) rating µ, rating deviation
φ, and volatility σ. He plays against m opponents with ratings µ1; : : : ; µm, rating
deviations φ1; : : : ; φm. Let s1; : : : ; sm be the scores against each opponent (0 for a loss,
0.5 for a draw, and 1 for a win). The opponents’ volatilities are not relevant in the
calculations.
So does this means that a draw would count as a half win for each player in a match that ends in a draw?
1 Like
No.
The relevant factor in the calculation is (outcome - winning probability).
Let’s assume you played a game against someone and you probability is 0.6 (60%). If you win that game you get (1 - 0.6)=0.4 times something. If you loose it’s (0- 0.6)=-0.6 times … points (you loose more points because it was more probable for you to win than to lose.
Now in case of a draw it’s (0.5-0.6)=-0.1. You loose only a little points because this is the expected result.
4 Likes
Is there a specific theoretical or otherwise aimed-for correspondence
between Glicko2 rating difference and win probability?
If yes, is it the same as for elo?
https://en.wikipedia.org/wiki/Elo_rating_system#Mathematical_details
(I am asking so I can run the numbers from
https://forums.online-go.com/t/analysis-of-a-9x9-handicap-tournament/45408
for myself.)
1 Like
Apparently:
Let P2 = Expected outcome for player 2. Then:
P2 = 1 / (1 + e-A)
with
A = g(sqrt(r12+r22)) * (s2-s1))
and
g(x) = 1/sqrt(1+3x2/pi2)
I believe each player has rating si ± ri in these equations.
2 Likes
You have more )s than (s in
g(sqrt(r12+r22)) * (s2-s1))
.
So do you 
glicko and elo values are scaled equally.