This material was taken from this most epic thread:
http://www.wargear.net/forum/showthread/1854/Debate:_Board_Championship_and_Global_Ranking
But needs to be focused into it's parts.
What should our ranks be based on?
CP? - GR? - An either/or/both system? - Some other score?
What's been said:
Carry on.
The specialist and duellers are over represented by Global Rankings too.
my opinion, is that before we try to figure out some kind of aggregate score, we should think about improving the components of the aggregate. For example the CP discussion that is going. The ideas for making team play easier/more fun.
And...
Maybe this is not the right thread for this (oh god not another thread!), but it seemed to me like there was a (start of a) consensus to use something like a TrueSkill ranking to calculate GR.
Basically TrueSkill stores two #s for your skill. A perceived skill & a confidence in that skill. This would help stabilize GR and make it more accurate.
http://en.wikipedia.org/wiki/True_Skill
It seems like a relatively simple change. Well supported by theory.
In difficult discussions like these, I think the best thing to do is find the smallest changes that can be agreed upon and work from there, instead of coming up with 100 various grand plans that no one agrees on. Bottom up instead of top down.
I think the first thing to decide is if you want the rankings (general, private, etc.) to be based on one metric or allow multiple.
My vote is get the aggregate stat right, and base it on that. If that can't happen (due to disagreement), my vote is to keep it based on CP or the higher of your GR & CP. I'd be happy with any of those.
AttilaTheHun wrote: I think the first thing to decide is if you want the rankings (general, private, etc.) to be based on one metric or allow multiple.
This is a tad further out of the box - but that's what I do. Most of it ends up in the trash, but that doesn't stop me..
The highest mean score of any of the following metrics determines rank:
CP/GR, Team/GR, Tourney/GR, Team/CP, Tourney/CP, Tourney/Team
Basically pick two..
There are some people who play in tournaments to learn a board without affecting their GRs and CPs.. On the other hand, there are some who do the opposite - they learn a board in regular play before playing in a tournament (presumably to rise in the tournament rankings). Such a system would enable most if not all players to preserve their goals and systems of play and be able to attain rank.
The problem is scaling CPs to all the other metrics, and I have a proposal for that..
Scale CPs proportionally along with whatever metric has the highest score. Currently this is Falker1976 with a 3424 GR.
3424 GR = 504 CP 1000 GR = 0 CP With a range of 2424 each CP point = 4.809 GR
So to compute the Adjusted CP: 1000 + (CP * 4.809)
This would put my 48 CP equivalent to a 1230 GR and ATH's 209 CP = 2005 GR (Not too far from his actual GR). That seems fair. I pretty much only play my own boards, so my ~1700 GR should be the better score for me.
Here are some numbers using the above system..
berickf GR:Tourney 2644:2482 = 2563
smoke CP:GR 2445:2014 = 2220
ATH GR:CP 2147:2005 = 2076
M57 GR:Tourney 1707:1308 = 1508
A bunch of Tourney players would raise significantly in rank. For instance Gill only has 19 CPs, but..
Gill GR:Tourney 2138:1602 = 1870
I like the gist of M57's last post. It seems like the right way to approach "Rank". Taking the max of some categories will make it so that a strong player who chooses to do one type of thing well doesn't end up ranked as a "Private". I prefer this idea to an aggregate, at least for determining ranks up to a certain level.
An alteration to my above proposed system (for the CP weighted crowd) could be to require CPs to be one of the components..
..or CPs + pick any two..
Though my personal vote would be for my original proposal.
Well, I would think picking your 2 best scores (on an individual basis) across all the categories should be possible right?
Then you normalize it and compare players to each other.
ratsy wrote:Well, I would think picking your 2 best scores (on an individual basis) across all the categories should be possible right?
Then you normalize it and compare players to each other.
I wasn't thinking of normalizing all of them -- I was only normalizing CPs to GR, but I like the idea of normalizing everything to whatever is highest.
Normalizing only CPs to whatever has the greatest max will have the effect of always weighting CPs more heavily than some categories. Presumably Tourney and Team play scores will catch up or the gap will lessen - but we don't know that. Not normalizing all categories might better please the CP crowd - if there still is one.
Okay so If I had a great CP score and a great GR (hah!-my hypothetical best) and you had a great Team and Great Tourney score (your two hypothetical best) - could we compare them to determine rank?
ratsy wrote:Okay so If I had a great CP score and a great GR (hah!-my hypothetical best) and you had a great Team and Great Tourney score (your two hypothetical best) - could we compare them to determine rank?
Currently, if ALL are not normalized - the CP:GR mean would eclipse the Team:Tourney Mean. If you were the top player in your two categories and I was the top in my two, you would have 3424 and I would have
( 2809+2385)/2 = 2597.
If ALL are normalized then we would both have scores of 3424
The difference is whether you want CPs to have more weight than Tourney and Team. GR is the normalization control by default because it is has the highest max score of all metrics using that system.
Tournaments give trophies... that's already cool. GR and CR really seem to be the key measurements of skill. Team play means you had good team mates, or at least ones open to communication.
Tournaments and Team games are stat fodder or trophy case in my opinion. I'm more impressed by someone's ability to win a high percentage of games over time (GR/CR) than to be lucky enough to win 3 of those games in a row to get the tournament.
I see Team Play as stats fodder and Tournaments as worthy of recognition, but I don't see why you would blend those with GR & CR. Why do we want to stick them all in the same equation?
Edit : Oops! I didn't realize this was for profile rank (General etc), I thought this was a discussion at first to replace the top GR/CR on the right.
+1 to M57s idea. I like the idea of normalizing CP with GR to average. I'm not sure about Team & Tourney.
So wouldn't it be interesting if you saw statbar like this: Team,Tournament,CP,GR
TM TR CP GR Avg
% 40 27 73 68 52
Your % would be the percentile you stand in the site above or below the norm with a final average that could be used for ranking purposes.
I think perhaps I'm just stating M57's post in a different way, convert each category to a % and tadaa.
This seems like another proposal for an aggregate system, doesn't it? Otherwise couldn't you just award titles independently within each category (GR, CP, TM, etc.) without having to normalize each one to the rest? I.e. top Team player is 5-star general, top GR IS 5-star general, etc.
Andernut's is an Aggregate proposal, which I like because it's inclusive of team and tourney. It is normalized as well if that's what folks want. Mine is a Rank proposal, where the low scores are thrown out in favor of finding the areas where the player performs best and awarding rank based on their areas of expertise. Remember it's just a rank, and I think more people should be able to get in on the fun.
My calculation method can also be used to determine an aggregate as well.
TM TR CP GR Avg
berickf 2385 2504 1188 2644 2180
ATM 1319 1510 2005 2147 1745
I have a slight preference for mine, but would recommend that if we do end up both changing the Rank system and employing an Aggregate - we use the same system for calculations. I think the advantages of mine are that people are use to seeing these types of numbers and thus have an inherent understanding of their values, only one metric (CP) needs to be 'converted,' and it offers the option of each category being normalized or not.
Andernut wrote:Team play means you had good team mates, or at least ones open to communication.
It could also mean that you yourself are a good teammate? Hence, the overall result of the team rank (not any individual team game) is measuring your ability to win those games and NOT your teammates. If you are a crap ally and your good teammate(s) pull you kicking and screaming to victory, your overall team score will not really rise that high in the long run because winning will not be a consistent result for you if you are the anchor on the team. If you drag bad teammates to victory, however, now you're doing yourself some good. If you and your teammates are both good, then, because of the diminishing returns built into GR calculations, eventually you'll both hit a rank ceiling and need to look for teammates who, whether good or not, have a lower rank to try and extend your ceiling, or you'll have to find higher combined-ranked teams to play against so as to risk less and possibly gain more if winning the more difficult game, which is exactly the same way that standard GR works. If neither you nor your teammates are good, well then, you need to go back to the drawing board when it comes to your individual team skills. Point is, it's not like having consistently "good" teammates is going to drag you to the top of the ranking if you don't have individual team skills yourself, and are not a good teammate. At some point we need to stop making excuses about teammates and take some personal responsibility for one's own individual team skill ability!
Also, I made a suggestion in the original GR/CP aggregate thread regarding having more control over who one's teammates are so that we can put that excuse behind us once-and-for-all. If everyone could choose their teams by first player in on a team given invitation privileges, then there would be no risk of getting strapped with a boot case or an teammate who doesn't even know it's a team game and attacks their own allies. I think that such could make more people feel comfortable to join team games, up the competition level significantly for team games, and make it an even more competitive rank and increase the team-GR point ceiling. Win-win-win.
M57 wrote:Andernut's is an Aggregate proposal, which I like because it's inclusive of team and tourney. It is normalized as well if that's what folks want. Mine is a Rank proposal, where the low scores are thrown out in favor of finding the areas where the player performs best and awarding rank based on their areas of expertise. Remember it's just a rank, and I think more people should be able to get in on the fun.
My calculation method can also be used to determine an aggregate as well.
TM TR CP GR Avg
berickf 2385 2504 1188 2644 2180
ATM 1319 1510 2005 2147 1745
I have a slight preference for mine, but would recommend that if we do end up both changing the Rank system and employing an Aggregate - we use the same system for calculations. I think the advantages of mine are that people are use to seeing these types of numbers and thus have an inherent understanding of their values, only one metric (CP) needs to be 'converted,' and it offers the option of each category being normalized or not.
Another nice thing about your style M57 is that one can also look directly at the numbers and know exactly where they need to target their efforts to improve their aggregate rank as well, as obviously, the lower the number the easier to improve it. This is where I get encouraged to work on my CP more so then because of CP in-and-of-itself. I have mostly thrown CP to the back burner, but, if it were my glaring weakness in an aggregate, then, I'd have to tighten my belt and put more of an attempt in at it because the rest of my ranks are largely maxed out until the rest of the field below rises to increase their ceilings.
I like how it's simple and they all automatically weight based on the possible ranges that the scores fall in. If a team-tourney score were added then perhaps an added weight could be applied and it could be the average of GR, CP, Team * .8, Tourney * .8, team-tourney *.8, or something like that, to keep the CP/GR guys happy.