Latest Posts

Topic: Widelands player rating: call for games

king_of_nowhere
Avatar
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-08-31, 20:54

WorldSavior wrote:

Maybe some more rules from your tournaments should be "copied" more or less, for example the draw rule for very long games without a winner.

that could apply for games that were started and never finished. although the tournament rules state that one can claim victory if he has advantage there, and the games must be reviewed by the referee. we can't do that if we expand the ranked system

why?

because it would require having an official referee. not a problem now, but would be if more people play


Top Quote
trimard
Avatar
Topic Opener
Joined: 2009-03-05, 22:40
Posts: 230
Ranking
Widelands-Forum-Junkie
Location: Paris
Posted at: 2019-09-01, 13:00

Now updated with glicko2 and using the tournaments games face-smile.png


Top Quote
king_of_nowhere
Avatar
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-09-02, 02:28

well, at least now it looks more sane, although 5 games are not enough to differentiate. we should probably adjust what in elo rating is the K factor, not sure how it's called in gliko. or maybe we could try giving anyone the starting score calculated after the tournament, and recalculate their scores as if they played with that. should increase the gap between players

also, we need more games without worldsavior, as him defeating anyone tend to flatten the ranking.

Edited: 2019-09-02, 02:29

Top Quote
trimard
Avatar
Topic Opener
Joined: 2009-03-05, 22:40
Posts: 230
Ranking
Widelands-Forum-Junkie
Location: Paris
Posted at: 2019-09-02, 12:03

or maybe we could try giving anyone the starting score calculated after the tournament, and recalculate their scores as if they played with that. should increase the gap between players

I'm very much against that. Why should we want to increase the gap between players? We don't have enough data to know how far of each other each players are. The variation is very high, and the current score represent that. Yes we could manipulate the data, but that would just give us meaningless data.

We need more real games face-smile.png

also, we need more games without worldsavior, as him defeating anyone tend to flatten the ranking.

We need more games without you, worldsavior and the-x, or to be more precise, we need more games where you guys don't play only against eachother. That's why the-x is so low in the ranking when I expected him to be much higher for example.

we should probably adjust what in elo rating is the K factor,

We have a few factors to test yes:

  • number of games to examine at the same time ==> I'm working on that yet

  • tau and starting volatility values

  • starting stardard deviation


Top Quote
einstein13
Avatar
Joined: 2013-07-29, 00:01
Posts: 1118
Ranking
One Elder of Players
Location: Poland
Posted at: 2019-09-02, 13:36

@trimard and king_of_nowhere

Both Elo and Glicko-2 are ranking systems that were designed to collect data from bunch of games at once, then calculating the final rating score.

For Elo most of use cases are when you recalculate points after each win or lose. Then the equation is pretty simple.

For Glicko it is the same. We can simplify the equations and apply the system for one-game only and the points would change only for participants.


einstein13
calculations & maps packages: http://wuatek.no-ip.org/~rak/widelands/
backup website files: http://kartezjusz.ddns.net/upload/widelands/

Top Quote
trimard
Avatar
Topic Opener
Joined: 2009-03-05, 22:40
Posts: 230
Ranking
Widelands-Forum-Junkie
Location: Paris
Posted at: 2019-09-08, 18:57

I updated the score! face-smile.png

Because I adapted the technique to add games, I made less mistakes (I hope).

And glicko seems to give much truer results


Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-09-09, 09:18

trimard wrote:

I updated the score! face-smile.png

Because I adapted the technique to add games, I made less mistakes (I hope).

Good, but this is how it has really been:

  • kaputtnik played 4 matches, won 2: loss against KoN, forfait round, win against LAZA, loss against the-x, win against GunChleoc
  • LAZA won only 1 match (against watchcat). He lost against you, kaputtnik, hasi and modellbahner.
  • I won 1 rated match against the-x on Ice Wars, 3 against him on Crossriver, 1 against KoN, 1 against you, 1 against Hasi and 4 tournament games, so 11 games and not 10.

And maybe I overlooked some further matches, I don't know.


Wanted to save the world, then I got widetracked

Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-09-09, 20:31

WorldSavior wrote: And maybe I overlooked some further matches, I don't know.

I checked it again and it looks that I overlooked only matches of KoN.

He played 5 tournament matches (4 wins) and then 4 rated matches ( loss against me, two losses against the-x, one win against the-x).

But at the other hand, is there a necessity to include his three matches against the-x already in the ranking? I mean, the replays are not uploaded yet, and now you can even upload them easily here under the posts. (You see, I really want to watch those matches face-wink.png )


Wanted to save the world, then I got widetracked

Top Quote
trimard
Avatar
Topic Opener
Joined: 2009-03-05, 22:40
Posts: 230
Ranking
Widelands-Forum-Junkie
Location: Paris
Posted at: 2019-09-09, 23:52

1 against KoN

I thought I shouldn't count that one?

But ok for the rest! We'll fix that when I have time

Huh, maybe for the replay, we say it's a requirement for the game to be validated, but can we act that post act of the new rule? I dunno, seems it might be a reocurring problem in the future. huh.


Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-09-10, 19:15

trimard wrote:

1 against KoN

I thought I shouldn't count that one?

Yes, you should face-wink.png Why not?

But ok for the rest! We'll fix that when I have time

Okay

Huh, maybe for the replay, we say it's a requirement for the game to be validated, but can we act that post act of the new rule? I dunno, seems it might be a reocurring problem in the future. huh.

"Post act"? I don't understand...

Edited: 2019-09-10, 19:16

Wanted to save the world, then I got widetracked

Top Quote