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

Posted by: The First Legend
Why do you keep saying that they are my buddies? They arent. I dont know them. Also, I love the troll card. If someone presents a valid argument, and defends it, they arent yeilding to the OP, and they are trolls. Yes, I am a troll for defending my poitns. Get over yourself.


This thread is supposed to be about HOW the Trueskill system works, not whether or not you like the way it works. Valid or not, your argument does not belong in here. Please stay on topic.

  • 11.26.2007 1:02 PM PDT

Posted by: The First Legend
Posted by: Xbot
Posted by: El Kafungus
Posted by: The First Legend
Why do you keep saying that they are my buddies? They arent. I dont know them. Also, I love the troll card. If someone presents a valid argument, and defends it, they arent yeilding to the OP, and they are trolls. Yes, I am a troll for defending my poitns. Get over yourself.


This thread is supposed to be about HOW the Trueskill system works, not whether or not you like the way it works. Valid or not, your argument does not belong in here. Please stay on topic.

Translation: You're right, the system does suck but stop posting. I don't want my thread to get unpinned.


Normally I wouldnt say this, but if you go ahead and look at how many other people talk about their ranks in this thread, you will see that he never called any of them off topic or pulled the troll card. I am the only one to defend my stance, and I am the only one he tried that with. Is there a connection? I think so.
Maybe I can calm the tension here. Is this system perfect? No. Does that mean it's completly broken? No. There are issues at present and Bungie is looking into them. You have reasonable creavences. Can something be done to fix the problems? That's what Bungie is working on now. For the time being those that experience issues need to work around them.


Posted By: Lukems - Source

Issues like Anti Jalox and the vD0 guys are issues we're looking into and at how we can address them. In rare cases, players that the system thinks are "good" are getting matched in matches that are producing "low match quality" -- this isn't quality related to player or system performance, but rather in the system's assessment of how reasonable the match is for assessing skill -- Mixed Party scenarios are often "low match quality" to prevent boosting. Like I said, we're looking at addressing it, we're sorry for the inconvenience and we're sorry.


[Edited on 11.26.2007 3:17 PM PST]

  • 11.26.2007 3:02 PM PDT

My Problem:

Usually I'm not one to rant about these types of things but after reading through the explanation on the ranking system I still have questions and complaints. To me the system favors a player with a high Sigma rating. A player who consistantly plays well shouldn't be the person on the short end of the stick. I've been hovering around level 35 in Team Slayer for about a month now. I have lost games and won games but always played well. I have come to the conclusion that basing the player's Trueskill on wins and losses is the fatal flaw in the system. If Player A performs well (in comparison to the Player B-H) but is on the losing team, Player A shouldn't be penalized. It is obvious that Player A is playing at a higher skill level than the opposing team. The way I see it is you're punishing one person because of the play of the rest of the team. Then I question why Player A is placed on a poor team. Is it because of this ranking system? Yes and no. Playing alone, the single player falls victom to the ranking system, being paired with people who aren't up to par to their level. The ideal solution is to party up--you can guarantee getting a decent matchup if you play with a group that you know will play well. The downfall to this is that you can abuse the party system by including a person with a lower rank (not too much lower), because you'll start playing people slightly under your skill range but close enough where every player can still gain rank.

My Solution:

I think the best scenerio would be to take actual game performance into account when selecting a team and when deciding whether or not a player should level up. I'm no mathmatical genius but I'd image it'd work something like this: mu and sigma would retain their values but the introduction of a third variable would be introduced to keep track of the map goal and then compared to your ability to complete the map goal. So for example you're playing CTF and you score twice, the variable 'y' would track this in some way and in the final formula add it to your final score. For Slayer games I'd expect it to keep track of kills but that could be difficult because a player could get 18 kills and 20 deaths, which isn't that good. It could keep track of your kill/death spread (not to be confused with ratio). Again, the variable 'y' would be applied to your final score. I'd expect the formula for objective games to be different then Slayer games.

So yeah I kind of went off track, with my point, but ultimately I want you to see what I'm talking about. The system should place more emphasis on a single player's ability to complete a specific goal, not the team's. I don't know how many people will get this comparison but I compare this to the yesteryears of the NHL. Lets say the '97 Red Wings is Red Team and Blue Team is the '07 Penguins. The Red Wings have all the star players while the Penguins have all mediocre players for the exception of one. The Red Wings inevitablely win 5-3 but the Penguin's star player was able to get all 3 goals for their team. In my system this player gets credited with a place on the 3 stars post game. In Bungie's system, the player gets no recognition whatsoever.

  • 11.26.2007 3:31 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

This is not true... i made a new account named "nub booster"...i won 27 games in a row on team slayer with 567 kills 230 deaths and 260 head shots...i am a lvl 16 after all of that....thats some bull

  • 11.26.2007 3:40 PM PDT

Posted by: II THING 2 II
This is not true... i made a new account named "nub booster"...i won 27 games in a row on team slayer with 567 kills 230 deaths and 260 head shots...i am a lvl 16 after all of that....thats some bull
What's not true? The OP's original post? We have a very good understanding of the system. Pretty much what we can know is known and there hasn't been an unexplainable circumstance yet based on our understanding of the system.

First the pretty numbers you threw out there about your performance don't mean anything to the system. So don't expect to rank faster because of it. You ranked up to 16 in only 27 games. That's pretty fast, but here is the thing. Your undefeated. I checked your stats. That's going to cause issues. You are in fact going to rank slower if you continue to play undefeated. This issue is discussed in this thread. So if you want to know more about it and why it happens look at page 5 where it's explained. Bungie may find a way to fix this issue but that won't happen until the next patch at the earliest. For now to rank fast you can't win to much. If you are winning 90% of the time or better you are going to slow your progress.

[Edited on 11.26.2007 7:56 PM PST]

  • 11.26.2007 7:54 PM PDT

Posted by: maximus_asinus
My Problem:

Usually I'm not one to rant about these types of things but after reading through the explanation on the ranking system I still have questions and complaints. To me the system favors a player with a high Sigma rating. A player who consistantly plays well shouldn't be the person on the short end of the stick. I've been hovering around level 35 in Team Slayer for about a month now. I have lost games and won games but always played well. I have come to the conclusion that basing the player's Trueskill on wins and losses is the fatal flaw in the system. If Player A performs well (in comparison to the Player B-H) but is on the losing team, Player A shouldn't be penalized. It is obvious that Player A is playing at a higher skill level than the opposing team. The way I see it is you're punishing one person because of the play of the rest of the team. Then I question why Player A is placed on a poor team. Is it because of this ranking system? Yes and no. Playing alone, the single player falls victom to the ranking system, being paired with people who aren't up to par to their level. The ideal solution is to party up--you can guarantee getting a decent matchup if you play with a group that you know will play well. The downfall to this is that you can abuse the party system by including a person with a lower rank (not too much lower), because you'll start playing people slightly under your skill range but close enough where every player can still gain rank.

My Solution:

I think the best scenerio would be to take actual game performance into account when selecting a team and when deciding whether or not a player should level up. I'm no mathmatical genius but I'd image it'd work something like this: mu and sigma would retain their values but the introduction of a third variable would be introduced to keep track of the map goal and then compared to your ability to complete the map goal. So for example you're playing CTF and you score twice, the variable 'y' would track this in some way and in the final formula add it to your final score. For Slayer games I'd expect it to keep track of kills but that could be difficult because a player could get 18 kills and 20 deaths, which isn't that good. It could keep track of your kill/death spread (not to be confused with ratio). Again, the variable 'y' would be applied to your final score. I'd expect the formula for objective games to be different then Slayer games.

So yeah I kind of went off track, with my point, but ultimately I want you to see what I'm talking about. The system should place more emphasis on a single player's ability to complete a specific goal, not the team's. I don't know how many people will get this comparison but I compare this to the yesteryears of the NHL. Lets say the '97 Red Wings is Red Team and Blue Team is the '07 Penguins. The Red Wings have all the star players while the Penguins have all mediocre players for the exception of one. The Red Wings inevitablely win 5-3 but the Penguin's star player was able to get all 3 goals for their team. In my system this player gets credited with a place on the 3 stars post game. In Bungie's system, the player gets no recognition whatsoever.
This is my take on personal performance: You can't use it. It can't be part of the equation because the evaluation is not uniform across all gametypes and playlists.

Take Lone Wolves for instance. Assists are bad. Assists are kills that got away so the lower the better in that playlist. K/D ratio is meaningless. The goal is to get the most kills, not to have the best K/D ratio. I've played games of rumble pit (back in halo 2) and gone 20/10, but I came in 3rd or 4th because other people got out there and made the kills. Even though they ended up going 22/24 they knew how to get more kills then me.

How about Objective game types. Who should be rewarded more the guy who scored the flag? The guy who stole the flag? or the guy who covered the flag carrier? Do you give credit for a great K/D ratio. Should the flag scorer get more credit or less credit then someone who sat back and slayed all game allowing you to get that flag and score it. What about the person who lead the team and created the plan to get that flag? How do you quantify that?

In my opinino personal performance cannot be used. It just can't be quantified and just about every gametype has a different and debatable measure of good performance.

As far as a player being penalized for playing well but losing: They really aren't. The system can't be judged on a per game basis. Think about a person playing many games. Sometimes you'll be matched with players who do well. Other times you'll be matched with players that don't. If you go into matchmaking alone this will even out. The only variable that's left is the individual. Thus if you do well you will rank up because that positive contribution to the team will allow your teams to win more. As you move up your contribution slowly drops to nuetral. That's when you are at your level.

So you see the skills of your random teammates has to average out. The only thing left that effects whether you go up or down is the individual. If you team up you might be able to get a little higher, but that's expected. If you play with people you know you are going to be familiar with them and there skills/strategies. Communication will generally be better. It's natural for a team that plays together to perform better then a team that is randomly put together. Even if they are all the same skill level. Random teams will always be worse off then teams that practice together.

**EDIT**
Okay, I looked over your last 51 games (one more then I planned :) ) and I noted all your K/D spreads. When you total it all up you are positive 61 kills. Since you played 51 games take 61/51 and on average you go positive 1.19 kills per game. It looks like on average you are contributing positivly, but not by very much. Based on that number I would conclude you are near your level. Will you ultimatly move up to a 36 or a 37...maybe, but you probably aren't heading up much further. From what I can see you aren't being held back by the random people you play with. The reason you aren't moving much has more to do with you then them. As I said in time you may inch up a bit further but by the looks of it not very far.

"Playing alone, the single player falls victom to the ranking system, being paired with people who aren't up to par to their level."

Almost everyone is up to par with there level. Only individuals who play with a team could possibly make it higher then they "should". (I'm talking by more then a couple of ranks here) So those random people you get paired with are playing at there level. The other team is paired up using the same system. This means it has to average out. There is no way you can suffer more then the next person. If you get "bad" teammates then the other team also has to get "bad" teammates from time to time. The same holds true for "good" teammates. The only variable that's left is the individuals skill.

So my conclusion is that based on your data you appear to be around your level. Not because you are being held back, but because of your own performance. I don't think using just wins/losses is a problem of the system. Individual performance can't be used because it's just impossible to quantify with any degree of accuracy and each gametype will change what a good performance is.

[Edited on 11.26.2007 8:57 PM PST]

  • 11.26.2007 8:12 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Nice explanation man

  • 11.26.2007 8:32 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Do you lose a 50 after 1 loss or does the experience accumulate like it does with all other number levels?

I want to know because I am near getting a 50 but do not want to stop playing team slayer until my friend gets his as well (he has a 47).

  • 11.26.2007 10:59 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

I kind of understand your explanation and I know you say its not perfect. In fact, this ranking system is very unfair. I have a roommate that played 19 consecutive games and won all the way to game 18, he ranked up 1 time. His teammate (they were in team doubles) ranked up all the way to 50 from 46 in the exact same games. They lost there last game and my friend went from 48 to 47 with one loss, his teammate stayed at 50. If this system is in fact "working as intended" it is a stupid system and does not reward a gamer for skill. Bungie, please choose another system or make some tweaks to the existing one.

  • 11.27.2007 2:33 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

well that is does not make any sense at all...you are telling me i have to lose to rank up faster?? Thats absolutely ridiculous! I have now won 54 in a row and i'm only a 31....this system needs to be fixed so that i don't have to embarrass the casual players and finally get a challenge!

  • 11.27.2007 2:49 AM PDT

Posted by: II THING 2 II
well that is does not make any sense at all...you are telling me i have to lose to rank up faster?? Thats absolutely ridiculous! I have now won 54 in a row and i'm only a 31....this system needs to be fixed so that i don't have to embarrass the casual players and finally get a challenge!
Honestly I don't really understand what you are complaing about. You are ranking up at a rate of about 1 level every 1.7 games. That's a very reasonable speed and nothing worth complaining about. I don't see the problem. It would be nice if the system would just put you at 50 and see how well you do but it's not going to do that. It is going to take a little time for it to find your level, but it seems to be going along smoothly.

  • 11.27.2007 3:11 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Wow. Instead of all that crap, Bungie should use a system where:
If a player is consistently winning games, well, he must be doing something right for his team, therefore he levels up easily.
Levelling up is also related to positivity in a game, as well as assists (but to a lesser extent). Don't know how many games I've lost while playing really well, and gone down levels.

p.s. Is it just me or is MVP simply given to the player with most kills, regardless of deaths?

  • 11.27.2007 3:27 AM PDT

Posted by: The Trunk Munky
Wow. Instead of all that crap, Bungie should use a system where:
If a player is consistently winning games, well, he must be doing something right for his team, therefore he levels up easily.
Levelling up is also related to positivity in a game, as well as assists (but to a lesser extent). Don't know how many games I've lost while playing really well, and gone down levels.

p.s. Is it just me or is MVP simply given to the player with most kills, regardless of deaths?


So people can simply go into Team Slayer and just run around on their own and get kills and go up?

It is called Team Slayer for a reason, and that reason isn't so people just run around without their team.

  • 11.27.2007 3:42 AM PDT

Posted by: The Trunk Munky
Wow. Instead of all that crap, Bungie should use a system where:
If a player is consistently winning games, well, he must be doing something right for his team, therefore he levels up easily.

That's how the system currently works. Well, depending on your definition of easily. Yes there are a few exceptions to this regarding mixed skilled parties and teams that only play together but these specific issues are being investigated. They do not affect the majority of players.

Posted by: The Trunk Munky
Levelling up is also related to positivity in a game, as well as assists (but to a lesser extent). Don't know how many games I've lost while playing really well, and gone down levels.

It's been stated numerous times in this thread and in others that individual performance does not matter in a team game. Being rewarded for loosing is not and should not happen. Your team lost.

  • 11.27.2007 6:45 AM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

very nice. thanks.

  • 11.27.2007 11:41 AM PDT

We have no Mercy
We shall show no Fear
We do not boast our Strength
We have Honour
We are the Elites!!!

That was a very in-depth explanation of the ranking systems,it helped me out alot,and i'm sure you helped out alot more people who didn't quite understand the ranking system either.

  • 11.27.2007 1:01 PM PDT

I've read through this thread and the ranking system. It's not much different than H2. But, none of this explains why after 3 or 4 wins, you would not level up, in say Lone Wolves, even against player 2-3 levels above you, and then suddenly de-level after 1 loss to players who are your same level. According to the system, you should have leveled up in those first 2-3 matches. I mean just now, I won 3 games in a row where I was in the middle of the levels (38 to 42), and I didn't level up. Then I just lost a game against people all my level and I de-leveled. My point is that I think the system itself says one thing, but in actual practice, something else is happening.

In teams, I agree - leveling should be based on performance of the player, not the team. In understand that the system rewards you more for winning the match, but does that mean the player who made 25 kills on the losing team shouldn't level up, or isn't at a higher skill level? I can't count how many times I've carried a team just to lose in the end and then de-level. If I am in last place on the team, I understand, but if you get more single kills that anyone else in the whole match? Come on...

  • 11.27.2007 1:35 PM PDT

Posted by: somdowns
I've read through this thread and the ranking system. It's not much different than H2. But, none of this explains why after 3 or 4 wins, you would not level up, in say Lone Wolves, even against player 2-3 levels above you, and then suddenly de-level after 1 loss to players who are your same level. According to the system, you should have leveled up in those first 2-3 matches. I mean just now, I won 3 games in a row where I was in the middle of the levels (38 to 42), and I didn't level up. Then I just lost a game against people all my level and I de-leveled. My point is that I think the system itself says one thing, but in actual practice, something else is happening.
If you would like an explanation please list the range of games you are questioning and I'm sure someone will try to help explain it. I just looked at your games of Lone wolves. I don't see anything that looks unexpected. On the first page you pretty much alternate between 4/5/6 finish and 1/2/3 finish. Your rank through that period also was alternating between 38/39.

Assuming 3rd or better counts as a win you have 20 wins and 18 losses in lone wolves on the first two pages. That's occured mostly at skill level 38 but some of those games you were a 39. Looking at it from a broader perspective does that really seem like you should be moving up. Just looking at a small number of games doesn't always give you the real picture.

Posted by: somdowns
In teams, I agree - leveling should be based on performance of the player, not the team. In understand that the system rewards you more for winning the match, but does that mean the player who made 25 kills on the losing team shouldn't level up, or isn't at a higher skill level? I can't count how many times I've carried a team just to lose in the end and then de-level. If I am in last place on the team, I understand, but if you get more single kills that anyone else in the whole match? Come on...
On page 13 of this thread I posted a semi long post discussing why I think it's impossible to use individual performance's as part of the equation. I also discuss why only using wins/losses works perfectly fine for determining your skill level and why those games you carried your team but lost shouldn't be a concern. Check out my post and see if it makes sense to you. I really don't think it's possible to use individual performance in a ranking system.

[Edited on 11.27.2007 2:44 PM PST]

  • 11.27.2007 2:33 PM PDT

i have a Q. lets say i raise my level up to 40 for just a game or 2 and then i go down to 39 for the rest of eternity. once i gain the EXP to become a colonel will my highest level reached (40) allow me to be promoted?

in laimens terms... is the true skill part of rank based on highest level achieved or current level?

thank you very much in advance.
(im sure this question has been asked b4 but i never use the forums and am not sure how to find my answer in any other way other than asking it myself)

  • 11.27.2007 2:45 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

Ok don't the words streakily and consistently mean the same thing if I'm winning a lot? Unless I'm consistent at losing. What's your definition of streakily?

  • 11.27.2007 3:01 PM PDT
  • gamertag: [none]
  • user homepage:

Thx for the help =P

[Edited on 11.27.2007 3:14 PM PST]

  • 11.27.2007 3:11 PM PDT
  • gamertag:
  • user homepage:
  • last post: 01.01.0001 12:00 AM PDT

I was a level 21 undefeated in about 15 games... I played with my friend who was a level 35. I got to level 43 by the time he got to 37. I went undefeated with him for another 15 games or so in less than 2 hours; I understand how trueskill works but it seems slightly unfair that it started giving me 3 levels a game and he only leveled twice after I more than doubled my level! Currently this name is probably just a couple away from 50.

My other name got to 50(zosu) after I went on two 5 game win streaks from 48 to 50.. seems fair. My brothers name is trapped at 48 (chobofied). He got both my names from 47 to 49/50. We won 8 games twice in a row against players ranging from 46 to 50 and he did not level up. I like how the ranking system is, but it seems like it unfairly traps people at a level if they end up playing too many games at it. At this rate it looks like it will take his him 20 games in a row to get from 48 to 50, even though we are constantly beatting 49s and 50s

We are just going to make new names, it only takes like 35-40 wins to get to 45 if we play with mid 30s and dont lose any games. I am sure he will level from 47-50 faster than it is leveling him from 48-49.

BYT I do like the explanation of the original post, however he omitted a lot of the details from the official trueskill ranking system, about how it calculates their "trueskill" and depending on the trueskill of both your teammates, oppononents, and everyone thats played online... how it ranks you up compared to that. I guess I really had nothing to say, just wanted to comment on how the ranking system is good, but unfair in penalizing people at high levels who can get "trapped"

  • 11.27.2007 3:21 PM PDT