Halo 3 Forum
This topic has moved here: Subject: In depth explanation of the Halo 3 skill ranking system.
  • Subject: In depth explanation of the Halo 3 skill ranking system.
Subject: In depth explanation of the Halo 3 skill ranking system.
  •  | 
  • Exalted Legendary Member

Iam enjoying Halo Reach and revisiting some past Bungie games, iam a big fan of Bungie and their online community.

Very imformative this means some people dont stand a chance.
At least ive had a little luck in team slayer now.

  • 08.22.2008 10:28 AM PDT

i r guy who gon t33ch u less0n

Good post

  • 08.22.2008 4:23 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

i was always wondering when i wasn't levelling up at the right time and this explained it! thanx! =P

  • 08.22.2008 9:10 PM PDT

Dead Mans Hand-Always Loaded.

I WAS SOCK

wow... i know this has been posted for a while but this is amazing... good work with all the research... very impressive

  • 08.24.2008 10:20 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

does quiting ruin anything?

  • 08.25.2008 11:17 AM PDT

i am a proud memeber of the BWO
i am a recruit and work extreamly hard as a member. come join bwo at
http://blackwaterops.tk/

good job just 3 words
wall o text =D
no im serious this helped alot

  • 08.26.2008 6:35 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Posted by: Luke 87777
does quiting ruin anything?


Well it causes a loss in most cases, so yes. It hurts your rank unless your team still wins...in which case it does nothing.

  • 08.26.2008 3:17 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Thank you man, this really helps a lot, I was wondering

  • 08.26.2008 3:26 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

HALO 3 IS TERRIBLE .. BUNGIE HAS ALOT OF MONEY N THEY FORCE PPL TO PLAY A HORRIBLE GAME .. y CANT BUNGIE PAY FOR A SERVER MAKING IT BETTER FOR EVERY1 .. NO THEY WANNA MAKE PPL PAY MORE N MORE FOR A GAME THEY BARELY UPGRADE, THE SHOT SYSTEM IS TERRIBLE, THE BEATDOWN SYSTEM MAKES NO SENSE, SOMETIMES U CAN JUMP AT PPL N HIT THEM N SOMETIMES U CANT, SAD COMING FOR A COMPANY THAT MAKES MILLIONS OF DOLLARS, NO WONDER UR LOSING CUSTOMERS TO CALL OF DUTY, THE GAME SUXS BUNGIE SUXS, HALO 1 AND 2 WAS GOOD THIS SHYT IS HORRID, N Y THE -blam!- DID U MAKE THE GAME BOXED OFF SO PPL GET STUCK ON STUPID AS CORNERS, MAKE A -blam!- BETTER GAME WORTH PLAYING, PPL ONLY PLAY THIS SHYT CUZ THEY HAVE NOTHING ELSE TO DO N THEY WANNA BE PRO SO BAD, YOU MAKE MILLIONS AND BILLIONS OF DOLLARS MAKE A -blam!- BETTER GAME N ET A SERVER FOR IT U CHEAP BASTARDS

  • 08.27.2008 9:46 PM PDT

To survive adaptation is necessary.

It helps but I still don't understand why I need to win 5-7 game to gain 1 level but only need to 1 game to derank.

  • 08.29.2008 1:11 PM PDT

You know, I'm starting to figure out why that was supposed to be a bad idea.

wow... thats interesting

  • 08.29.2008 2:39 PM PDT

3 years and only 1 ban, try harder next time noobs.

Very informative, trueskil makes more sense now. Thanks

  • 09.03.2008 9:40 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Yeah good stuff it's about time some peole did some service for the community.

  • 09.08.2008 7:39 AM PDT

Finished the Fight!

It's very interesting and well thought out. At the same time, If it truly works on such a statistical basis, it actually kinda discourages me from continuing playing ranked MM, since it seems that besides skill (or lack therefof), you also kinda play the odds in regards to how the people you're matched against with will positively or negatively afftect the values that can push you up or down the ranking ladders. Having that on my mind has truly choked me up on recent ranked matches. I'd rather not worry about that and play Social or Customs instead.

  • 09.08.2008 8:12 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Thank you so much for this dude

I was really puzzled trying to figure out why I wasn't ranking up and you cleared it all up,

thanks again ^.^

  • 09.09.2008 8:42 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Ok does negative exp boosting acutally make your Mu higher so you actually rank up faster and dont' de-rank when you lose?

  • 09.09.2008 12:51 PM PDT

Posted by: l KxS Lyle l
Ok does negative exp boosting acutally make your Mu higher so you actually rank up faster and dont' de-rank when you lose?
Quitting a bunch of games will make your sigma higher and lower your skill level (MU) in the playlist you quit in ONLY. As soon as you start winning your MU will go up and your sigma will go down.

Neg EXP boosting is not the "trick" everyone thinks it is. Quitting a bunch of social games to boost your friend in Team Doubles will not work. It's sigma and NOT EXP, that is at work here. You have a separate MU and sigma for each playlist. It is your sigma, MU and a couple other variables that make up your displayed skill level for each list.

  • 09.09.2008 2:42 PM PDT

I read this over the summer break and I had NO idea what the hell sigma and mu was until now. I am taking AP stats and I have had the class for about 4 weeks now. I think I have a basic understanding of bell curves but not completely. I am still pretty clueless.

The reason I mention this is so I don't get bashed when I make a comment about the O.P.'s equation.

He makes it sound like your sigma should be high. To my knowledge that is the total opposite to what is desirable. Your sigma is the standard deviation. A low standard deviation means the data is not spread out and a high standard deviation means the data is spread out. You want your data to be consistent and not all over the place.

Just look at the formula:

Ranking = mu - (K * Sigma)

If the Sigma was high your ranking would be lower.


I don't understand the conservative ranking and the meaning of that. BTW this http://research.microsoft.com/osa/apg/details.aspx goes into much more detail and is 10x more confusing.

  • 09.10.2008 5:42 PM PDT

Posted by: damn Juggernaut
I read this over the summer break and I had NO idea what the hell sigma and mu was until now. I am taking AP stats and I have had the class for about 4 weeks now. I think I have a basic understanding of bell curves but not completely. I am still pretty clueless.

The reason I mention this is so I don't get bashed when I make a comment about the O.P.'s equation.

He makes it sound like your sigma should be high. To my knowledge that is the total opposite to what is desirable. Your sigma is the standard deviation. A low standard deviation means the data is not spread out and a high standard deviation means the data is spread out. You want your data to be consistent and not all over the place.

Just look at the formula:

Ranking = mu - (K * Sigma)

If the Sigma was high your ranking would be lower.


I don't understand the conservative ranking and the meaning of that. BTW this http://research.microsoft.com/osa/apg/details.aspx goes into much more detail and is 10x more confusing.
Sigma, as it's used in TrueSkill is more than just a standard deviation. It also is used to "factor" the amount of MU gain or loss. So, if your sigma is high, you gain more MU for a win. Everyone starts out with a high sigma and it will normally be lowered after every game. As a players standard deviation gets smaller, the MU changes are smaller and basically means the system is more sure of your skill level so big MU changes are no longer necessary.

The reason the system is conservative is because of the K value. This is the amount of conservativeness and is used to factor down your MU, along with sigma to give you your displayed skill level. This basically is the degree of error so to speak and flattens out your skill level changes slightly.

  • 09.11.2008 6:25 AM PDT

thank you. It make a little more sense. Something I still don't get is if what you said was true, then the equation is not right.

Ranking = mu - (K * Sigma)

As Sigma gets smaller ranking does too (unless K is negative).





[Edited on 09.11.2008 6:19 PM PDT]

  • 09.11.2008 2:44 PM PDT

Posted by: damn Juggernaut
thank you. It make a little more sense. Something I still don't get is if what you said was true, then the equation is not right.

Ranking = mu - (K * Sigma)

As Sigma gets smaller ranking does too.

Try this:
mu=25
k=3
sigma = 8

25 - (3 * 8) = 1 your skill level during your first game

Play a game and win then you get
mu=26
k=3
sigma=7

26 - (3 * 7) = 5 your new skill level

Sigma has decreased after the win but you get MU for the win as well and your new skill level becomes 5 instead of the initial 1. Now these equations are over-simplified but it shows how it works without getting in to decimals.

  • 09.11.2008 6:27 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Under what circumstances would a player be able to raise, on average, their Sigma after a period of games rather than lower it? How would they create such uncertainty, especially since an individual's Sigma is expected to lower over time? I understand that a higher Sigma is desired, and I understand why that is so.

Is it possible to increase my rank at a faster rate, on an account that has played well over 1,000 games, by drastically increasing its Sigma? Is it even possible to drastically increase my Sigma on such an account?

SORRY if an answer to my question has already been posted and I didn't see it.

[Edited on 09.11.2008 6:55 PM PDT]

  • 09.11.2008 6:54 PM PDT

Posted by: WE ALREADY LOST
Under what circumstances would a player be able to raise, on average, their Sigma after a period of games rather than lower it? How would they create such uncertainty, especially since an individual's Sigma is expected to lower over time? I understand that a higher Sigma is desired, and I understand why that is so.

Is it possible to increase my rank at a faster rate, on an account that has played well over 1,000 games, by drastically increasing its Sigma? Is it even possible to drastically increase my Sigma on such an account?

SORRY if an answer to my question has already been posted and I didn't see it.
The only way for sigma to go up is through a series of unpredictable outcomes. A small amount of sigma is added back in after every game. If you can have enough unpredictable outcomes where the equations do not lower your sigma much then the amount added back in can actually increase it. This is the method MS states the system uses to detect skill changes. They go on to state it would take between 50-100 games for the system to detect the players skill change.

So, the short answer is there is no way to increase your sigma and help you level up quicker other than just winning.

  • 09.12.2008 6:12 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Since it takes around 50 to 100 games for Bungie to get enough information on your account to accurately detect skill change, does that mean it's easier to increase your skill level with a new account rather than an older account, or quite the opposite?

Thanks for answering my previous question.

  • 09.13.2008 10:36 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Let me rephrase my question:

If you play with someone that has negative exp, will that help you increase in rank faster or do you have to have the negative exp?

Simple yes or no will suffice. =]

  • 09.14.2008 3:46 AM PDT