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

Post edit history

Why don't I get ranked direct?

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
8/5/2019 4:36:09 AMAUrankAdminGoogleFrog before revert after revert
Before After
1 There is no minimum number of games. Unfortunately the rating system is fairly opaque and in many ways frustrates human intuition. There is no straightforward interpretation of it that maps to many of the common concepts and features that we like to see in rating systems. 1 There is no minimum number of games. Unfortunately the rating system is fairly opaque and in many ways frustrates human intuition. There is no straightforward interpretation of it that maps to many of the common concepts and features that we like to see in rating systems.
2 \n 2 \n
3 The rating system works by fitting a function that represents you skill over time to the entire history of games played on the server. This function should [b]not[/b] be confused with the human concept of a chart shows you your change in rank over time. Instead the function is fit all at once, in the present, to your whole play history. Your rating in "the past" can change based on the future performance of you and everyone you played with. The function is optimised under some assumptions, one of them is (probably, although this may also be too-human of a concept) that players don't change skill particularly quickly. It also assigns a certainty (or variance, or some other fuzzy human word) to its estimate at each point in the past based on some sort of maths. 3 The rating system works by fitting a function that represents you skill over time to the entire history of games played on the server. This function should [b]not[/b] be confused with the human concept of a chart shows you your change in rank over time. Instead the function is fit all at once, in the present, to your whole play history. Your rating in "the past" can change based on the future performance of you and everyone you played with. The function is optimised under some assumptions, one of them is (probably, although this may also be too-human of a concept) that players don't change skill particularly quickly. It also assigns a certainty (or variance, or some other fuzzy human word) to its estimate at each point in the past based on some sort of maths.
4 \n 4 \n
5 There will not be a simple reason for you having too much uncertainty to be ranked. Playing more games would help. Playing smaller games would help. Playing games with people who have more certain rankings would help. 5 There will not be a simple reason for you having too much uncertainty to be ranked. Playing more games would help. Playing smaller games would help. Playing games with people who have more certain rankings would help. I have no idea why your "75% confidence" number dropped by 200 points. Remember that the chart shown on your ratings page is [b]not[/b] a history of your rating and the uncertainty. It is a function, all computed in the present, that is fit to your past. Notice how you appear to start out above 1500 ELO. This makes no sense under a standard ELO implementation.
6 \n 6 \n
7 I found it was too silly for you to have played 60 non-AI games and not be rated. I pushed one of the only levers available to me and bumped MinimumDynamicMaxLadderEloStdev from 220 to 300. 7 I found it was too silly for you to have played 60 non-AI games and not be rated. I pushed one of the only levers available to me and bumped MinimumDynamicMaxLadderEloStdev from 220 to 300.