Loading...
  OR  Zero-K Name:    Password:   

Post edit history

Help! I'm overrated!

To display differences between versions, select one or more edits in the list using checkboxes and click "diff selected"
Post edit history
Date Editor Before After
1/28/2018 3:45:01 PMDErankBrackman before revert after revert
Before After
1 @Godde your arguments about separating 1v1 and team ratings are valid. And there are also arguments against it. For example it can be good to separate ratings for players who play 10 games a day, but there are also players who play 2 games per month. If they are played on troll maps, it can be better to include them to everything. 1 @malric those are good questions. It seems casual rating is impacted by all games ( not sure about noelo and silly maps, though) . @Godde's change seems to be due to the hourly recomputation based on his 1v1 games. AFAIK team balance only uses rating average currently. See [url=http://zero-k. info/Forum/Thread/24625]this thread about team skill[/url].
2 \n
3 @Godde your arguments about separating 1v1 and team ratings are valid. And there are also arguments against it. For example, it can be good to separate ratings for players who play 10 games a day, but there are also players who play 2 games per month. If they are played on troll maps, it can be better to include them to everything.
2 \n 4 \n
3 How all those arguments weigh for prediction/balance quality on the whole can only be shown mathematically. I suppose this is what @DeinFreund did, so I trust what he says. 5 How all those arguments weigh for prediction/balance quality on the whole can only be shown mathematically. I suppose this is what @DeinFreund did, so I trust what he says.
4 \n 6 \n
5 What we can discuss about is the convenience of a system even if that worsens the prediction/balance quality or new systems that could improve quality but are not tested yet. 7 What we can discuss about is the convenience of a system even if that worsens the prediction/balance quality or new systems that could improve quality but are not tested yet.
8 \n
9 Furthermore, I think that subtracting one time stdev from mean to get shown rating is enough, rather than two times.