Galcon 2: 41 players online!
:: Community ::
ForumsMessagesModsChat
 

Forums :: [OLD] iPhone :: Strategy :: Official Stat Warriors Leader Board

You must sign in to post.

Page 1 / 22 1   2   3   4   5   6   7   8   9   10   11   12   13   14   15   16   17   18   19   20   21   22 

Official Stat Warriors Leader Board :: Mar 4, 2011 @ 3:55pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

The Official Stat Warriors Leader Board



http://toomanywheres.com/galcon/LeaderBoard.aspx 



Notes:

1) It is NOT real-time though it is really close to beign real-time.  The process of importing games begins at the start of each hour, and usually finish in around 30-45 minutes.

2) A discussion of the formulas used to calculate ranks can be found here: http://www.galcon.com/forums/14/16/4268/ 

3) The Leader Board Ranks will always contain all flag types.  Each flag contains a proportion of the total players.  The higher up in rank a player goes, the smaller that proportion of players is allowed to be.  

Broken down as such:
GA - 0.25% of total players
3 Stripe - 1.5% of total players
2 Stripe - 3% of total players
1 Stripe - 5% of total players
Admiral - 10% of total players
Captain - 15% of total players
Commander - 25% of total players
Lieutenant - 50% of total players
Ensign - 75% of total players
Cabin Boy - 100% of total players


4) The rank you have on the Leader Board may or may not have any relevance to your in-game rank.  This is normal... do not panic.  The goal of this is to determine the best players over time, ignoring the current 10 day rule for in-game ranks.

5) Your in-game rank, and the in-game rank of your opponents, is the only thing that this Leader Board calculates against.  It does not (yet) calculate against its own rank indicators.

6) If you play against a certain flag, you will be scored at that flag level in this Leader Board.  Regardless of who actually won the game. It could be the pink flag that won the game - it doesn't matter for this ranking.  The goal of this chart is to determine how well you play against EACH flag type.  So suiciding the higher rank to make the lower one win, is not as effective anymore - it still counts as a loss to that higher flag.


HAVE FUN!
post updated on Mar 18, 2011 @ 9:50am
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 4:08pm

Cabin Boy quantum

Clan: Singularity
Joined: May 19, 2010
Posts: 454
Location: Every(No)where @ once.

This is great!! Btw, does this not have 2v2 games included?
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 4:24pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

Still working on the 2v2 side of things...  I'll add that in when its ready.  :)
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 4:56pm

Cabin Boy swollenpig

Clan: Zenith
Joined: Jun 12, 2010
Posts: 954
Location: Colorado

Rofl, all hail standoff!
Nice work, looks great.
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 6:01pm

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

Mastery = Number of Opponents / Games Played 
Field = Opponents Played Against / Total Opponents Possible 
Skill = Field * Mastery * Win Percent 
Score = Win Percent * Win Percent * Rank * Rank * Skill 
Note: must have 10 games or more against a flag to be counted.


So Mastery % goes down as you play more games against fewer oppnenents.

Skill takes into account mastery % to calculate skill level.  

Score takes into account skill to give you overall rank.

So basically, being respectful, and playing more games against your opponents screws your overall rank.
See my point?

Nice try, but people like me and esparano who play few people lots of games in 1v1 will never be ranked well in this system lol. 

Oh well, time to galcon it up!
post updated on Mar 4, 2011 @ 6:02pm
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 6:08pm

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

Oh maybe I'm not done.  Just wanted to show another thing.

My 1v1 3 stripe stuff:

7 three stripes played
190 games total
112 games won
58% win 
Skill: .24

pwn.myself (sorry just picked you randomly)

5 three stripes played
27 games total
18 games won
66% win
Skill: 1.19


So I've played way more games, against more opponents, have a very decent win %, yet pwn.myself is 5 times as good as me based on your skill score. 

I'm just ramblin cause I'm bored.  Not angry, just listing facts.  

Ok, nooow it's time to galcon it up!  Just wanted to give you something to think about.
post updated on Mar 4, 2011 @ 6:10pm
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 11:08pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

IonizedFire, great feedback!  Thanks!

Couple of thoughts:

1) being dinged for fighting friends more often is something I hadn't thought about.  Will give that some thought....

2) Your Mastery % is the only thing holding you back.  You have a MUCH better Field rating.  Which counts tons more towards the overall Skill number.  I didn't give the full formula there on the page...  ;) it is:

(Field as a whole number) * (Mastery / 100) * (Win Percent / 100)

Basically, its a whole number times two decimal numbers.  So really, a higher Field rating is MUCH better to have.

The Field number is ALWAYS going to be a low number due to the amount of players involved.  Only when you get into the higher ranks do the numbers begin the get anywhere above 10.  

So that 3% Mastery number of yours is really hurting your skill number.
Re: Official Stat Warriors Leader Board :: Mar 4, 2011 @ 11:43pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

The conceptual problem I've been wrestling with here is how exactly do you measure 1 persons skill as compared to anothers?  Phils current system is designed around setting arbitrary benchmarks you have to hit before leveling up.  This gives ZERO indication of ones skill compared to anothers... which is proven time and again here by people claiming the current GA's are not as good as the previous GA's.

The problem here is that there is really no good way to gauge the overall skill of a player WITHOUT doing some measure of comparison against a wider range of players.  Hence the Field and Mastery numbers.  

I do see your point about the MAstery number getting skewed by just playing against friends for the fun of it.  Got any ideas on how to mitigate that one?
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 6:42am

Cabin Boy periwink

Clan: Galconians
Joined: Mar 31, 2009
Posts: 416

Mastery = Number of Opponents / Games Played 
Field = Opponents Played Against / Total Opponents Possible  

Both of them help people who played a wide number of people. Is that normal?

And rank shouldn't be taken into account in the ranking.

Very nice attempt, though!
Keep up the good work.

(and did you check mewotwo's system?)
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:22am

Cabin Boy orphanmaker

Joined: Dec 31, 2010
Posts: 7

Do you think you could run the Fusion ranks through this and post it to?
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:27am

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

Mastery = Number of Opponents / Games Played 
Field = Opponents Played Against / Total Opponents Possible  

Both of them help people who played a wide number of people. Is that normal?


Absolutely.  Those are in there by design, for that very purpose.  The thing I absolutely can't stand about Phil's current system is that a GA could be attained by playing a VERY small number of people.  That completely does away with the competitive nature of rank.

I'm not surprised the currently high ranking people here dislike this new method.  As it actually forces them to play different people then they are used to.  Which....  is the whole point here!



And rank shouldn't be taken into account in the ranking.


Could you explain that one a bit more please?  I agree that basing a persons numbers off the in-game ranking introduces a little mischief into the numbers...  my long term goal here is to import the games and not even care about what that persons in-game rank was when calculating the scores.
post updated on Mar 5, 2011 @ 10:31am
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:28am

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

Do you think you could run the Fusion ranks through this and post it to?


Yup.  Was thinking of asking Phil for a Game Dump of all games, including Fusion.
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:34am

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

Just letting you all know....

The hosting company I'm using here appears to not like me running a 4 hour database process twice a day.  How surprising...  ;)  So until I can figure out a way to crunch all those numbers a little bit at a time, the stats will remain as they are right now.

Sorry about that.  :(
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:42am

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

I do see your point about the MAstery number getting skewed by just playing against friends for the fun of it.  Got any ideas on how to mitigate that one?


I really don't know haha.  Basically, your mastery % is higher based on playing opponents fewer and fewer games, basically meaning, you run in a game to get a few lucky wins and bail.  I'm starting to think mastery % is a measure of your 'rank whore-y-ness' haha.  

If this is the case, I'm almost thinking mastering % should just be removed completely.  Because, no matter how much you make mastery % insignificant, like mastery%/100 or mastery%/1000, it's still a measure of stealing cheap wins  from opponents and running to maintain a high win %. But I have no idea how that'd effect all the ranks.  You'd have to experiment with that I guess.
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 10:44am

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

Sorry I'm critiquing this so hard.  I just think you're on the verge of a huge breakthrough in a real measure of stats, and it can't be nothing short of perfection :)
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 12:28pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

Hmmm I could use the inverse of it...

(100 - (Opponents / Games))

Really however the Field number might be better to use then opponent count. As the field represents the overall amount of players possible. 

(100 - (Field / Games))

That should provide a % that reflects both diversity of players AND quantity of games played against them. So the more games played, the better that number will become. 

Hmmm but then a low field number would help you more then hurt... :(

But then again... Field is the basis if the Skill number so it's influence is already pretty high...

Tough to find the right balance here. 

Thoughts?
post updated on Mar 5, 2011 @ 12:35pm
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 1:02pm

Cabin Boy hollander

Clan: Galconians
Joined: Dec 20, 2009
Posts: 1711
Location: your pants

i think you just have to try out a lot. dont expect to come up with the perfect formula in nearly a week. i must say that im very impressed in what you got this far though.

i wonder if its possible for you to make a 2v2 ranking. but with the ingenious idea of ranking the teams instead of the individuals.
post updated on Mar 5, 2011 @ 1:02pm
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 2:31pm

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

Hmmm I could use the inverse of it...

(100 - (Opponents / Games))



I think this might just work.  But I suggest (1-(Opp/Games)+.01)*100

(The .01 is hard to explain, but it accounts for mega rank whores that play like 1 game per person, so the mastery skill doesn't = 0% so skill doesn't = 0.  It can be changed to compensate later.)

That should give you a good percent.


Experimentation:

Rank Whore vs Respectful Player

Rank Whore:
10 Opponents
10 Total Games (1 game per opponent)

(1-10/10+.01)*100=1% Mastery 

Respectful Player
10 Opponents
100 Total Games 

(1-10/100+.01)*100=91% Mastery


YES! I think this will work.  Cause more importantly, the more games you play, the Mastery % slopes off at a square root rate!  Meaning, more games help you, but the more you play, it helps you less and less toward your skill!

Ahhh Jazz read this quick! I think this will help a ton haha.
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 2:57pm

Cabin Boy ionizedfire

Clan: iElite
Joined: May 25, 2009
Posts: 178

More thoughts:


Mastery =(1-(Opponents Played/Total Games)+.01)*100
Field = Opponents Played Against / Total Opponents Possible
Skill = Field * Mastery * Win Percent
Score = Win Percent * Win Percent * Rank * Rank * Skill


For this random experiment, we'll say 50 possible total 3 stripes (used for Field %).

Ranking of best to worst based off new updated equation.  Random sample of players with different stats against 3 striped players.

Matty
Mastery 97%
Field 22%
Win 50%
Skill .1067

IonizedFire
Mastery 97%
Field 14%
Win 58%
Skill .078764  (.97*.14*.58)

Socratic
Mastery 84%
Field 20%
Win 37%
Skill .06216 

Pwn.Myself
Mastery 83%
Field 10%
Win 66%
Skill .05478 

Alienzed
Mastery 75%
Field 10%
Win 57%
Skill .04275

(All %'s are real, based off your stat website.)


This is just preliminary, and first thing I see right away is that Field % has too much affect on skill rank.  Because, for example, Socratic and pwn.myself have the same Mastery %, but field % carries Socratic skill quite a bit better than pwn.myself.  Maybe win% * mastery% * sqrt(field%)?  I've done enough math for today, so I'll let you take over from here until tomorrow.  Just wanted to show a rough idea of what your inverse idea would do to the ranks.
post updated on Mar 5, 2011 @ 3:07pm
Re: Official Stat Warriors Leader Board :: Mar 5, 2011 @ 5:52pm

Cabin Boy jazz

Joined: Feb 20, 2009
Posts: 534
Location: aka The_Musician

^ Coooooool!!  Nice job Ionized!!  

I'll plug those numbers in and see what it does!  I also have some ideas I came up with over the last few hours I'd like to try out as well.  

I'm also going to implement a different storing mechanism for all these numbers in an attempt to get that 4 hour process down to a few minutes.  

Might be a few days before I have anything to show.  sorry.  :(
post updated on Mar 5, 2011 @ 5:53pm

Page 1 / 22 1   2   3   4   5   6   7   8   9   10   11   12   13   14   15   16   17   18   19   20   21   22 

You must sign in to post.