Free Allegiance Forums

We have a decent ranking system, but it has problems.

JG no one likes AllegAge

It's optional and just one of many suggestions.

Using an age based rank doesn't lead to balanced games.

Our current problem is that balancing on plain AllegSkill rank doesn't work.

Currently Allegiance is only aware of the player AllegSkill rank, with ACSS it will be also aware of player mu and sigma. This allows real team balancing.

PS:
Imago already implemented such a balancing algorithm into Allegiance in a past release of R6, but it was removed because it used data from a none FAO source...

ITT: RoyBrown doesn't understand statistics nor their implementation (and lack thereof) into Allegiance and ASGS.

EDIT:
Roy, if ranks had a high uncertainty, then they wouldn't be accurate.

fuzzylunkin1 wrote:QUOTE (fuzzylunkin1 @ Feb 17 2012, 06:20 PM) Roy, if ranks had a high uncertainty, then they wouldn't be accurate.

Sigh.


I guess my years of doing statistical analysis of electoral districts voting patterns in order to determine targeted advertising and fundraising is not important.

There's a lot of room from the near zero uncertainty we have now, to whacked out no uncertainty that you imply.

pkk wrote:QUOTE (pkk @ Feb 17 2012, 06:16 PM) Using an age based rank doesn't lead to balanced games.

We don't have balanced games now. Anyway, on the age principle it's only a part of a plethora of suggestions we can use, and it's not the rank itself it's a governor for low ranks (for example, so newbies aren't out of protection before they know what a nan is), which is one of the many current problems we have now.

RoyBrown wrote:QUOTE (RoyBrown @ Feb 17 2012, 12:27 PM) We don't have balanced games now.

fuzzylunkin1 wrote:QUOTE (fuzzylunkin1 @ Feb 17 2012, 12:20 PM) doesn't understand ... implementation ... into Allegiance and ASGS.

We can't have balanced games now, unless you manually assign people to teams after doing all the calculations yourself.

Well.. more accurately, we can't use the Trueskill method of balancing right now but we can implement a semi-usable balance algorithm with conservative ranks.

I'm personally ok with waiting for ACSS, doesn't seem too far off now. The Allegskill algorithm already allows for recent games to be weighed more heavily - we need to try that first, and if it doesn't help then think about a sliding window.

Spunkmeyer wrote:QUOTE (Spunkmeyer @ Feb 17 2012, 06:34 PM) I'm personally ok with waiting for ACSS, doesn't seem too far off now. The Allegskill algorithm already allows for recent games to be weighed more heavily - we need to try that first, and if it doesn't help then think about a sliding window.

I agree, that would be the best thing to try first. If MrC gets his way, ACSS will be out in a few months . . . .

Spunkmeyer wrote:QUOTE (Spunkmeyer @ Feb 17 2012, 11:34 PM) Well.. more accurately, we can't use the Trueskill method of balancing right now but we can implement a semi-usable balance algorithm with conservative ranks.

I'm personally ok with waiting for ACSS, doesn't seem too far off now. The Allegskill algorithm already allows for recent games to be weighed more heavily - we need to try that first, and if it doesn't help then think about a sliding window.

I hear you, and that's a good start. But we've been waiting when people including many that helped with the ranking system, said that a sliding window and increasing uncertainty would be good. I've heard that for a couple odd years now.

So, let's go ahead and try tweaking it, rather then totally leaving it to the current situation.

I mean fuzzy isn't a (7), we're talking just silly with it the way things are.