An investigation into the TrueSkill system of Halo Reach's Matchmaking.

Games Other Halo
1

(Keep in mind this applies to the competitive player of MLG. So if your reading this from the Halo Waypoint forums this might apply to you.)

So I was playing Reach last night with my buddy Chris (MLG Playlist) and we played around 11 games. Out of these 11 games we lost 9 of them because of teammates. Particularly bad teammates, or laggy teammates, and I used to be able to tolerate this but I can’t anymore. It’s just simply unacceptable that Bungie decided to not use the TrueSkill searching system in this game. I mean you used it in Halo 2 and 3, why not Reach? I understand that some people might not know what MLG is and are just trying it out for the first time but getting these bad and inexperienced teammates every game just makes me depressed at how this happened. It seems to me that Bungie just crapped out Reach and then handed it off to 343i to sort out. “Hey we made this game and it’s really flawed but now you guys can sort it out! Have fun 343!” That’s literally what they did. You would of never got these kinds of teammates if you were searching for a game in Halo 3. Most of the time you would match up against like minded people who wanted to win and have mics. But now you just get these kids basically who run for flag and walk it back to their base and go -17 and don’t even ATTEMPT to shoot anybody or get any kills! It’s pathetic!

I’m not here to break down people’s gameplay either and I’m not here to boast about how good I am. (I’m really not.) But all I want to do is play the game and win, and I want to have teammates who have the same competitive drive as I do. I play this game because I love the competition, not because I like the game. Reach stripped down to its basic’s is VERY boring and un-entertaining. Now I understand there are people out there who always play matchmaking in a team of four and this doesn’t apply to you. But if you’re like me and you search alone or in ones, twos, or threes you’ll know what I’m talking about. It’s incredibly annoying when you get SneakySniper13 on your team who just grabs the sniper and camps at your flag all game and never pushes up. Or when you get FireLord7000 who’s -18 every game with 3 kills. Now this test is no way of me trying to prove how great I am because I’m just an average player compared to most people. But what I’m going to prove is how ****** up basically the TrueSkill is in Reach.

What I’m going to do is, is go into matchmaking and play 5 games of MLG on my own without warming up or anything and I’m going to come back here and edit the thread with a picture of the scoreboard of each game and a Bungie.net link to the game.

I’m also going to add up my K/D of the 5 games and my teammates K/D’s.

Game 1: Sanctuary CTF:
Blue team:
NOX Lemon -7
ThatLuckyKiid -10
OctoSin +2
BossTh****** -17

B.net link: http://www.bungie.ne…yer=NOX%20Lemon

We lost that game 3-1. Guess how much flag time BossTh****** had? 1 MINUTE AND EIGHTEEN SECONDS. He went 12-0-29. Proves my point.

Game 2: Sanctuary CTF
Red team:
NOX Lemon 0
xo i SNiPR i ox -4
w The Game w -2 (He was the yellow bar host and quit half way through.)
l AlexTheKid l -4

This game wasn’t as bad, it was a 3v4 Sanc CTF and we managed to hold them off for the 0-0 draw. Let’s call this game an anomaly.

B.net link: http://www.bungie.ne…yer=NOX%20Lemon

Game 3: Sanc TS 4v3 (I was on the team of 4.)
Red team:
NOX Lemon +9
xXBrudu BXx -5
xDaRkW1ZarDx -7
OysterMeister -3

This game was golden. One of the blue team quit out right at the beginning of the game. We held the lead until the 30 kill mark and we lost a 4v3 50-44.

B.net link: http://www.bungie.ne…yer=NOX%20Lemon

Game 4: Sanc CTF (Again.)
NOX Lemon +12
iCr0ssBones +18
TOMiCiDE L +18
Thiz iz Spar7a -6

Not going to lie this was a pretty boss game. Even though we did have a bad teamate the other 2 made up for it. Now i know what it is like to have Totz on your team…

B.net Link: http://www.bungie.ne…yer=NOX%20Lemon

Game 5 Zealot TS
NOX Lemon +2
Fat OfThe La7Vd +1
WookieBeGone -6
ASAINCHONNY -7

Yet again proves my point. My teamates were also very annoying this game. Just the icing on top of a ****** cake.

B.net link: http://www.bungie.ne…yer=NOX%20Lemon

I hope you all understand where I’m coming from and I hope you support this thread to get 343i to put some sort of TrueSkill or the 1-50 ranking system back in Reach with the upcoming title update.

Thanks for Reading -Tom

tl;dr: TrueSkill pl0x

2

I think you just misunderstand how trueskill works.

You’re getting these bad guys on your team because of how good you are.

Lets say you get in a game, and you’re Blue team.

If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.

Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.

So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.

3

> I think you just misunderstand how trueskill works.
>
> You’re getting these bad guys on your team because of how good you are.
>
> Lets say you get in a game, and you’re Blue team.
>
> If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.
>
> Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.
>
> So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.

It works a little but no where near as good as Halo 2 or Halo 3’s ranking system did. Why you ask? Because there are gameplays on Youtube of Pros beating 4 random kids in the playlist 50-0. How the hell do 4 of the best players in the world get matched up with 4 random terrible players? By the way it was on their main accounts, the ones they apparently “already built trueskill in” in the playlist. And no, it’s not just because the pros are “that good”. Pros would lose many times in matchmaking in previous games.

4

I never or rarely got good team mate… I hate it all I saw are incompetent player that cannot even kill someone when they are 3 againts one… Even my sister could do better :smiley:

5

> It works a little but no where near as good as Halo 2 or Halo 3’s ranking system did. Why you ask? Because there are gameplays on Youtube of Pros beating 4 random kids in the playlist 50-0. How the hell do 4 of the best players in the world get matched up with 4 random terrible players? By the way it was on their main accounts, the ones they apparently “already built trueskill in” in the playlist. And no, it’s not just because the pros are “that good”. Pros would lose many times in matchmaking in previous games.

Lets dissect it some eh?

Your argument is that Trueskill doesn’t work in Halo Reach because you saw a YouTube of 4 pro’s beating 4 random kids 50-0.

You don’t have an actual game played in someone’s game history to show us the playlist/players/stats or ANYTHING about that ONE game you found on YOUTUBE.

But instantly you’re converted. You know better than everyone that Reach’s Trueskill is worse than H2 or H3 BECAUSE YOUTUBE TOLD YOU SO.

6

Ok everyone i have played my 5 games and guess what? Almost all of them were loses because of bad teamates. If you look at the stats i was pretty much the average guy on my team for most of the games so you can’t say we lost because of me. Also another thing that i want to point out is that in the games someone almost always quits out which again is very frustrating for whichever team the quitter is on. Here are the results.

Game 1: Sanctuary CTF:
Blue team:
NOX Lemon -7
ThatLuckyKiid -10
OctoSin +2
BossTh****** -17

We lost that game 3-1. Guess how much flag time BossTh****** had? 1 MINUTE AND EIGHTEEN SECONDS. He went 12-0-29. Proves my point.

Game 2: Sanctuary CTF
Red team:
NOX Lemon 0
xo i SNiPR i ox -4
w The Game w -2 (He was the yellow bar host and quit half way through.)
l AlexTheKid l -4

This game wasn’t as bad, it was a 3v4 Sanc CTF and we managed to hold them off for the 0-0 draw. Let’s call this game an anomaly.

Game 3: Sanc TS 4v3 (I was on the team of 4.)
Red team:
NOX Lemon +9
xXBrudu BXx -5
xDaRkW1ZarDx -7
OysterMeister -3

This game was golden. One of the blue team quit out right at the beginning of the game. We held the lead until the 30 kill mark and we lost a 4v3 50-44.

Game 4: Sanc CTF (Again.)
NOX Lemon +12
iCr0ssBones +18
TOMiCiDE L +18
Thiz iz Spar7a -6

Not going to lie this was a pretty boss game. Even though we did have a bad teamate the other 2 made up for it. Now i know what it is like to have Totz on your team…

Game 5 Zealot TS
NOX Lemon +2
Fat OfThe La7Vd +1
WookieBeGone -6
ASAINCHONNY -7

Yet again proves my point. My teamates were also very annoying this game. Just the icing on top of a cake.

7

> I think you just misunderstand how trueskill works.
>
> You’re getting these bad guys on your team because of how good you are.
>
> Lets say you get in a game, and you’re Blue team.
>
> If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.
>
> Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.
>
> So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.

If TrueSkill works like that, it works wrong. It isn’t fair to get bad teammates just because you’re so good. That’s like lowering your damage every time you get a killed, getting punished from being good. The TrueSkill should always match everyone in the game from a skill pool where the biggest differences between players’ skills are no more than four levels.

The system should never throw a team of unorganized randoms against a team of good and organized players. It also shouldn’t put bad players into the same team with good players, not even into the same match.

Tighter matches aren’t a real argument because if every player in the match was similarly skilled, the matches would be even tighter.

After reading your posts in a few threads, I have got the impression that you’re trying to make excuses for every flawed thing in Reach. But you know what, those excuses don’t help anything, you’re just fooling yourself.

8

> > I think you just misunderstand how trueskill works.
> >
> > You’re getting these bad guys on your team because of how good you are.
> >
> > Lets say you get in a game, and you’re Blue team.
> >
> > If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.
> >
> > Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.
> >
> > So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.
>
> It works a little but no where near as good as Halo 2 or Halo 3’s ranking system did. Why you ask? Because there are gameplays on Youtube of Pros beating 4 random kids in the playlist 50-0. How the hell do 4 of the best players in the world get matched up with 4 random terrible players? By the way it was on their main accounts, the ones they apparently “already built trueskill in” in the playlist. And no, it’s not just because the pros are “that good”. Pros would lose many times in matchmaking in previous games.

You mean this one?

http://www.youtube.com/watch?v=cTTxcdJIkkE

9

> > I think you just misunderstand how trueskill works.
> >
> > You’re getting these bad guys on your team because of how good you are.
> >
> > Lets say you get in a game, and you’re Blue team.
> >
> > If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.
> >
> > Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.
> >
> > So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.
>
> If TrueSkill works like that, it works wrong. It isn’t fair to get bad teammates just because you’re so good. That’s like lowering your damage every time you get a killed, getting punished from being good. The TrueSkill should always match everyone in the game from a skill pool where the biggest differences between players’ skills are no more than four levels.
>
> The system should never throw a team of unorganized randoms against a team of good and organized players. It also shouldn’t put bad players into the same team with good players, not even into the same match.
>
> Tighter matches aren’t a real argument because if every player in the match was similarly skilled, the matches would be even tighter.
>
> After reading your posts in a few threads, I have got the impression that you’re trying to make excuses for every flawed thing in Reach. But you know what, those excuses don’t help anything, you’re just fooling yourself.

So it’s more fair to take the bad guys and put them all on one team to be stomped on than it is to balance the skill levels of both teams to create a more competitive game?

Lets get something straight here… the MLG playlist isn’t struggling to find good matches because TrueSkill is broken, or works differently than it did in H2 or H3. It’s struggling because NO ONE PLAYS IT, or those that do play it ARE fooling themselves with how good/bad they are when paired with match making randoms.

It’s a playlist that SHOULD have big party restrictions, but with such a craptasticly small margin of players even wanting the MLG experience there’s no way that it could support such a restriction.

bash on my posts all you want, but when I’m wrong I admit it. In this case, I refuse to believe Trueskill is broken based on a YouTube video and a guy who plays the MLG playlist by himself.

10

Honestly dude, you can’t blame 9 losses on teammates

11

All the playlists use Trueskill for matchmaking, your skill number is hidden. Just because you can’t see your Trueskill level, doesn’t mean that you aren’t being matched accordingly.

Also, understand that there are many variables that go into matching players in matchmaking, your trueskill level isn’t the sole factor. Although, you can choose to search option to search for a game with closer matched skill levels if you wish for teams with closer skill levels. This will enable the skill variable to have a stronger effect in the search for your next game.

Furthermore, if the population is somewhat low in the playlist, you may end up with a substandard matchmaking game simply because the pool of players to choose from is very limited at that time. It is more difficult to find a closely suited match when the population severely limits the number of possible options for matching players.

I hope this explanation clarifies any issues you have with the matchmaking system in Halo: Reach.

12

> So it’s more fair to take the bad guys and put them all on one team to be stomped on than it is to balance the skill levels of both teams to create a more competitive game?
>
> Lets get something straight here… the MLG playlist isn’t struggling to find good matches because TrueSkill is broken, or works differently than it did in H2 or H3. It’s struggling because NO ONE PLAYS IT, or those that do play it ARE fooling themselves with how good/bad they are when paired with match making randoms.
>
> It’s a playlist that SHOULD have big party restrictions, but with such a craptasticly small margin of players even wanting the MLG experience there’s no way that it could support such a restriction.
>
> bash on my posts all you want, but when I’m wrong I admit it. In this case, I refuse to believe Trueskill is broken based on a YouTube video and a guy who plays the MLG playlist by himself.

I never said it’s fair to put all the bad players into one team against good players. i said the bad players should be put into the same team against other bad players. The good players should be put into the same team with other good players against other good players.

Although the low player count may be a partial reason that TrueSkill doesn’t find good enough matches, it’s not the only one. TrueSkill isn’t strict enough in Halo Reach. In Halo 3 I remember a playlist with usually very low population since 2009: Squad Battle. It’s a wonderful example of how well Halo 3 TrueSkill worked. Before Halo 3 lost a huge chunk of it’s population in 2009, I had attained a level 30 in Squad Battle. After Halo 3 lost majority of it’s population, it took me long time to find a match because the playlist had about twenty players with skill as high as mine.

In Halo Reach, I can find a game in playlists with half the population of Squad Battle in few seconds. This is because the matchmaking system doesn’t even try to find an even match if the playlist has a low population. Most of the time it doesn’t even find good matches in playlists with relatively high populations.

You don’t need a Youtube video to see this, you only need to go to play few games of matchmaking and see that the players you play against/with are absolutely horrible.

13

> > So it’s more fair to take the bad guys and put them all on one team to be stomped on than it is to balance the skill levels of both teams to create a more competitive game?
> >
> > Lets get something straight here… the MLG playlist isn’t struggling to find good matches because TrueSkill is broken, or works differently than it did in H2 or H3. It’s struggling because NO ONE PLAYS IT, or those that do play it ARE fooling themselves with how good/bad they are when paired with match making randoms.
> >
> > It’s a playlist that SHOULD have big party restrictions, but with such a craptasticly small margin of players even wanting the MLG experience there’s no way that it could support such a restriction.
> >
> > bash on my posts all you want, but when I’m wrong I admit it. In this case, I refuse to believe Trueskill is broken based on a YouTube video and a guy who plays the MLG playlist by himself.
>
> I never said it’s fair to put all the bad players into one team against good players. i said the bad players should be put into the same team against other bad players. The good players should be put into the same team with other good players against other good players.
>
> Although the low player count may be a partial reason that TrueSkill doesn’t find good enough matches, it’s not the only one. TrueSkill isn’t strict enough in Halo Reach. In Halo 3 I remember a playlist with usually very low population since 2009: Squad Battle. It’s a wonderful example of how well Halo 3 TrueSkill worked. Before Halo 3 lost a huge chunk of it’s population in 2009, I had attained a level 30 in Squad Battle. After Halo 3 lost majority of it’s population, it took me long time to find a match because the playlist had about twenty players with skill as high as mine.
>
> In Halo Reach, I can find a game in playlists with half the population of Squad Battle in few seconds. This is because the matchmaking system doesn’t even try to find an even match if the playlist has a low population. Most of the time it doesn’t even find good matches in playlists with relatively high populations.
>
> You don’t need a Youtube video to see this, you only need to go to play few games of matchmaking and see that the players you play against/with are absolutely horrible.

You’re just not getting it.

Nothing has changed with true skill.

The problem is the MLG playlist.

And YOUR observation (severely lacking any empirical data to support it) has NOTHING to do with what the OP is whining about, which is getting teamed up with bad players, NOT getting matched up against ANYTHING.

Tighten YOUR search parameters.
Play a playlist people want to play, and not MLG.
Don’t go it alone, get a group.

14

> tl;dr: TrueSkill pl0x

Reach uses TrueSkill.

The problem you are having is with the population of the playlists your playing.

Without a sufficent number of players TrueSkill can’t function properly.

15

> the MLG playlist isn’t struggling to find good matches because TrueSkill is broken, or works differently than it did in H2 or H3. It’s struggling because NO ONE PLAYS IT, or those that do play it ARE fooling themselves with how good/bad they are when paired with match making randoms.
>
> It’s a playlist that SHOULD have big party restrictions, but with such a craptasticly small margin of players even wanting the MLG experience there’s no way that it could support such a restriction.

I pretty much agree with this. There’s just not enough players playing the MLG playlist for it to truely be accountable as a measure of how well the system works.
There’s like 1000 players on at anytime when I log on roughly, the only mode I see with less is campaign. Most people seem to play Team Slayer, Firefight, or Infection.

16

would it kill you to put a TL:DR? i’m pretty tired of you walls of text. Plus, i believe that Reach DOES use trueskill, but in a more limited, non-advetised way. Plus, it’s because there’s a smaller player pool that the system isn’t working too well. I also don’t care about your game history, so put the games in a link to your stats in waypoint. It’s a pretty deep stats system you know.

17

Bad teammates are everywhere. Whether it’s in Reach MLG or Halo 3 Team Slayer. The best way to avoid them is by playing with friends that you know are good. One thing that I’ve learned from searching with randoms is that they will never fail to let you down.

18

> > > I think you just misunderstand how trueskill works.
> > >
> > > You’re getting these bad guys on your team because of how good you are.
> > >
> > > Lets say you get in a game, and you’re Blue team.
> > >
> > > If we took those two blues you say suck so much, and put them on the red team, instead of your team, then the match will end in a blow out.
> > >
> > > Giving you those bad players might cost you the game, but it’s going to result in tighter matches every time.
> > >
> > > So it is working, in spite of how you interpret what you are seeing, and what you are believing Bungie’s intention with Halo Reach is/was.
> >
> > It works a little but no where near as good as Halo 2 or Halo 3’s ranking system did. Why you ask? Because there are gameplays on Youtube of Pros beating 4 random kids in the playlist 50-0. How the hell do 4 of the best players in the world get matched up with 4 random terrible players? By the way it was on their main accounts, the ones they apparently “already built trueskill in” in the playlist. And no, it’s not just because the pros are “that good”. Pros would lose many times in matchmaking in previous games.
>
> You mean this one?
>
> http://www.youtube.com/watch?v=cTTxcdJIkkE

This is the Second 50-0 , this is the first 50-0 http://www.youtube.com/watch?v=rZO2CE8bqPo

Halo Reach needs 1-50 ranking system , not only MLG , I saw a 25-0 in Doubles Arena , 75-0 in Squad Slayer , 100-0 in BTB

19

Why would you believe any 50 to 0 score Youtube game is real?

Or a better question is, if it’s happening so often, why can’t anyone post a link to the actual game?

It’s not happening that often, if at all.

Trueskill is fine, no 1-50 needed.

20

You lost credibility when you claimed that Halo 2 used Trueskill and that Reach doesn’t.

Trueskill wasn’t around during halo 2’s time; Halo 2 used a modified ELO system. As for Trueskill, since it was created Microsoft require all multiplayer games using Xbox LIVE to use Trueskill or a modified variant of it to find matches.

Reach simply has very loose Trueskill restrictions outside of the Arena playlist.