|Rank & points mechanics, and a lot more
|Page 1 of 1|
|Author:||Skaal Khan [ Tue Aug 29, 2006 8:30 am ]|
|Post subject:||Rank & points mechanics, and a lot more|
I haven't read anything acceptable about points and rank. I mean: yes I read the messages about Glicko's rating previously used in MA(N), but it is definitely(?) not used in MAN2. Proof: you begin with 0 point. So... how does it work? I do not NEED an answer though, it's just... curiosity.
|Author:||Nick_WN [ Thu Sep 07, 2006 6:01 am ]|
We substract some score from medium value until you get 15 games finished. We hope to add some experience-based score as well, so it will become impossible to get on the top of ranking ladder with just few victories.
|Author:||Skaal Khan [ Fri Sep 08, 2006 4:55 am ]|
That's a beginning of an answer, thanks!
Although I don't mind for playing (and loosing ), I'm wondering many things about rating. Here's a list, feel free not to answer.
Is global rating the average of ratings by planets, or calculations don't care of which planet you battled on? Alternative: is there a constant-multiplier by planet for earned/lost points? (I mean: it would be fair to earn less points on a XS map than on a XL map)
Kate & Emily
Well, are they ranked? I think they should. If not: when challenging them, you challenge someone unranked with average points, which is unfair. I mean: I'm pretty sure that AI is not an average player which should have average rank (... and maybe that's why you see them as Lieutenants on battle reports on the website?). What about that test: be an unranked player, play 15 times against AI, win 15 times --> which rank will you have? more than the average rank (lieutenant or so I guess) which is quite unfair compared to people who played 15 times only versus more skilled humans. Solution: rank them?
Rating & rating deviation, period for calculation
Do you really use 1500 & 350 ? Do you display rating or something else (r-2*RD or so) Does really rating deviation increase with time (or do you cheat by fixing it to its max value)?
Edit: Rank calculations are made each time you win/loose a game. What about, as Glicko tells, having a period (month, week,...) for calculations? It would prevent (for people who play much) the rank to change a lot, it would smooth the rank change, it would avoid people to go down/up at each game, also avoiding the top ranked players to switch theirs ranks often/always.
What happens with the 15th game?!
I was sergeant with 718 points within 14 games.
I won the 15th game,... and I jumped to captain with 1884 points!?
This is quite surprising cos' I never had a game with someone whose rank was higher than lieutenant.
I understood what you explained about points before achieving 15 games, but do you think your calculated rank/points before this 15th game is reliable (I think: no; displayed victory bonus tells a lot more about true rank)?
Edit: A bit more: I am now major (!) with 18 games (W:14,L:4). Still surprising, cos' I played only against lieutenants or lower . Is this reliable? to be a major, without beating any captain or major?
Rank for trial players: unreliable, big malus onto their points
My guess: only a registered player gets its full points after 15th game, whereas trial players and/or players with less than 15 games do still have a big malus (1000 points or so) preventing them to reach a higher rank than lieutenant. Anyway, when playing against such players, rank calculations are made on the real points (not those with malus). Implies: there are some trial lieutenants that have a higher real rank than the one displayed and shoud/could be colonel or even more. (*)
Cancelling a game
As explained in another post (thanks Tiger!), I know you will (won't you?) add a new button to cancel a game: no points at all for anyone, as if the game has not existed. I agree, good idea! Purpose one: training a friend, telling him (by private chat) what to do or not; such a game should not be used for ranking people. Purpose two: game's parameters were done wrong, the two opponents agree to cancel the game.
Surrender in early turns
As explained in another ooooooooooold post (thanks Tiger!), I think that surrending in the very early turns should be used for ranking but with a lower percentage (20-50%, don't know). I have a few games only, but I know that you may have a very bad allies' placement; knowing that, in these very limited cases, a cheap surrending very early should be allowed for a more fair game.
(* I read what I wrote: and yes, it's clear!)
|Author:||Skaal Khan [ Mon Sep 11, 2006 3:11 am ]|
I change the topic, which now includes a few general thoughts and not only rank questions. I agree I'm not an experienced player, my suggestions may be not as good as I first think. Just react to them!
(to Storm440: married... and 3 children)
|Page 1 of 1||All times are UTC - 5 hours|
|Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group