+- One Man Left Studios Community Forums (http://www.onemanleft.com/forums)
+-- Forum: General (/forumdisplay.php?fid=1)
+--- Forum: Outwitters (/forumdisplay.php?fid=11)
+--- Thread: A good exemple of ranking problem. (/showthread.php?tid=581)
Pages: 1 2
A good exemple of ranking problem. - Mag!cGuy - 09-09-2012 08:39 PM
Hello there!!
I just wanted to give an exemple of the biggest problem of Outwitters. I really love this game (This is the only game where I bought in-app stuff, I bought the uber pack and the 35 games feature!), congratz to the developers.
My main game center account, W1cked21, is in clever league, with 150 win. I play all my other games with this account, that's obvious.
But, some days ago, I had the idea to create a second one, only for outwitters, named Zenitheon. Because I bought the 35 games feature, you can see it's not to play twice more games, in fact it's to see in which league I'd be now I really improved.
Remember I'm in clever league with 150 wins, and I can't reach gifted league with W1cked21. With Zenitheon, I'm in master league, I lost one time and won 10, I won against two masters and all my current battles against top 20 masters are at my advantage. That means, the more time you spend in a league, the smaller your chances are to be promoted. And I daresay a lot of clever players are way better than gifted ones.
I do know that at my very first battles, I almost only lost. I did 1 wins for 2 losses. So the problem is that, 100 battles after you were learning the game, these losses are still here to prevent you from going to gifted. You can't improve without having in your hidden matchmaking data the stats of the ancient bad games you did. You have to change of account so that your progress is visible. I'm not smart enough to imagine a system which understand you really impvoed and "forget" your beginners losses. Or maybe I am totally wrong?
Today I decided to stop to play with my W1cked21 account, because the battles aren't interesting at all anymore against clever. Because I don't have the time to play with two accounts (I bought the 35 games feature to award the developers and in case I'm bored and want to spend a lot of time on this game) and especially because playing costs money to the developers (if I understood, servers aren't free ^^).
I hope you won't ban me because I used two accounts during some days, it was a test and you can see the difference is amazing. Be sure I won't play anymore with W1cked21, and I'll update this topic telling you how I'm doing against masters.
RE: A good exemple of ranking problem. - Kamikaze28 - 09-09-2012 11:59 PM
The effect you are describing is not surprising to me. Let me give a little context first.
The system is supposed to gauge your skill and the only source of information it gets are your wins and losses and the skill rating of your opponents at the time the respective game was finished.
Other than the system's representation of your skill (the hidden matchmaking rating), we can imagine you have your true skill. That is to say the sum of your cognitive faculties, experience and strategic thinking which, unfortunately, can hardly be represented as a value or set of values.
Nevertheless, the system tries its best to keep the discrepancy between your true skill and the matchmaking rating as small as possible. This is tricky because it is easy to over- or underestimate you based on a winning or losing streak. Some 'stickiness' helps to smoothen your matches. But too much 'stickiness' will result in frustration because no matter how much you improve and how many faces you melt in your league, the system will not be convinced that a promotion is in order.
What you did with the creation of a second account is to re-initialize the skill gauging process. And because your true skill has improved since you started your first account, the system will happily place you in a higher league, because it doesn't know that you once had a lower skill rating.
A little fine tuning may be required to find the right amount of 'stickiness'.
RE: A good exemple of ranking problem. - Joggies - 09-10-2012 01:19 AM
Clearly an ideal system would give more weight to recent games to account for the fact that people get better as they play more.
Such a system might discount the relevance of a any result by 1/x^2 where x is how many matches you've played since. (You could obviously modulate the strength of the formula for the desired 'stickiness' of course)
RE: A good exemple of ranking problem. - Mag!cGuy - 09-10-2012 01:30 AM
I agree with you both!
Kamikaze, that's especially what I wanted to explain but you said it better
Joggies, even if I'm not 100% sure I think you're right x)
Oh, and I've been really surprised another time.. After my placement battles.
During these 5 battles, I won against another player which was in placement, 2 not that well ranked fluffy, and two pretty well ranked clever. I got to gifted league. That's okay. Then, I win a battle again a top 15 gifted. And what happenned? I've been promoted to master! Simply because I beat a gifted?
The game has been right to put me in master league (all the gifted I met lost against me after), but it's strange it did it so fast.
RE: A good exemple of ranking problem. - CombatEX - 09-10-2012 04:18 AM
Wow. You're really lucky. My friend went 5-0 in placements matches and got into Masters but then got demoted after beating 6 or 7 more Master players (he did lose to a Super-Titan a few games prior to demotion though). =/ Too bad he already paid for the teams or I'd tell him just to make a new account.
RE: A good exemple of ranking problem. - Mag!cGuy - 09-10-2012 04:33 AM
Then I'm gonna learn him something great: the purchases are kept from one account to another, because it's on the same iDevice.
Happy?
Oh, yeah I think I'm lucky because now I lost against 4 masters, won against 3 and many gifted, and I'm still not demoted. ^^
RE: A good exemple of ranking problem. - Demon - 09-10-2012 12:20 PM
Indeed. I was originally placed in fluffy but I've been beating top 5 clevers pretty consistently yet am still in clever.
RE: A good exemple of ranking problem. - Kamikaze28 - 09-12-2012 08:06 PM
(09-10-2012 01:19 AM)Joggies Wrote: Clearly an ideal system would give more weight to recent games to account for the fact that people get better as they play more.
Such a system might discount the relevance of a any result by 1/x^2 where x is how many matches you've played since. (You could obviously modulate the strength of the formula for the desired 'stickiness' of course)
I think the 'ideal system' you are referring to is a theoretical construct which is capable of matching people as perfectly as possible with all the knowledge it gathered so far. This of course is nigh impossible to implement.
Nevertheless, I agree with you. And before we descend into a sheer endless rabbit hole of possible mathematical descriptions for various aspects of the matchmaking system, I would just like to stop this topic of discussion in its tracks by saying that there are lots of different ways to modify, tweak, tune, fine-tune and re-tune systems of this sort. We don't need to attempt to enumerate them. Alex has probably done his research before his implementation and sees the same issues we see and discuss here.
RE: A good exemple of ranking problem. - Promicide - 09-12-2012 09:07 PM
Currently ranked #1 in Clever, 131 league wins, and honestly, my record wasn't that bad. They won't promote me to the next tier, I have well over a thousand points and a 300 point gap from the guy behind me. Hahah thanks for the trip!
RE: A good exemple of ranking problem. - Mag!cGuy - 09-12-2012 09:43 PM
Promicide, just do like me, create another account and see how it happens. You should get to master league if you ain't too unlucky