Forums / Community / Matchmaking Feedback & Discussion

[Locked] Matchmaking Feedback Update – June 11

OP ZaedynFel

  1. 1
  2. ...
  3. 45
  4. 46
  5. 47
  6. 48
  7. 49
  8. ...
  9. 50
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
Sturbz wrote:
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
I get that you are angry about kills per minute being used as a metric for determining which team will win with the best accuracy. Just to spark discussion though I have a few questions.

What would you like the matchmaker to use instead of KPM even though every other trackable stat is worse or doesn't improve upon the determination made just based on KPM. Remember 343 has never said that the matchmaker is always right. In fact they proudly state that it is only correct 68% of the time, which is up from the high 50s and low 60s of trueskill 1.

What would you like the skill ranking system do when a player wins? Grant a set amount of CSR gain and loss? I know the matchmaker is frustrating but should the ranks in this game be based on who plays the most? It's a hard decision and I wish h5 had both including a separate team rank that is only based on win/loss.

I see a lot of anger coming from people in this thread about how broken the matchmaker is but there is little discussion or points made on how they would improve it.

And in terms of the OP: I think the first game the matchmaker was wrong since it thought your team would do better. In terms of the players on the enemy team I know two of them that are honestly barely diamonds and are bad. The other two are good SWAT players however. In the second game it looks like you performed as expected of a high diamond with the champion level teammates you had and against the same team with two good players and two not so great ones. Since you performed as expected in that scenario why should you go up in rank? (I don't actually know MMR data so I could be wrong but those are just my theories for the two games).
I really hate the rank up in this. I have played countless of breakout games and won a lot but i can get passed plat 6, the rank up system is trash and broken. And i want to keep on playing but no point if i cant even get to champ.

https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/service-records/arena/players/f35demonator117

These are my games and as you can see, ive won a lot of games in a row and i cant move foward. 343 your rank up sucks.
Just take a look at my games and how many of them i won in a row....and i havent moved at all

http://halotracker.com/h5/games/F35DEMONATOR117
This post has been edited by a moderator. Please do not repost content a moderator has removed, repost a topic that has been locked, or post about forum moderation decisions. If you have a question or concern about a forum moderation decision, please private message the applicable moderator.
*Original post. Click at your own discretion.
Spoiler:
Show
I was previously an Onyx in FFA. I decided to jump back on this season and boy oh boy was I suprised. I ranked up to Onyx from Platinum 5, as I won a few games and played decently. However during my placements I would come up against Champs, and usually they would stomp the other 4 and me and win with a 20KDR. This needs to stop.
Anyways, I ranked up to Onyx 1500 and placed in the top 3 twice and got up to Onyx 1530 - Not bad right? - Wrong. Next two games I came 4th and dropped back down to 1500.

However, the best is yet to come.
The very next game I won and I got placed 1501, then played another game and came 2nd and got placed 1502. What?? What happened to 15 CSR when you come in the top 3??? I won 2 other games and got to 1532. I played a few more until my rank was at 1502 again - then came in the top 3 three times - and got to 1505 which is ridiculous. A 3 CSR jump for 3 top placements? Wow!

It gets better - I came 4th in my most recent game and dropped all the way down to Diamond 6!!!!!!

My main point is - Why is it when I place either 1st, 2nd or 3rd I get 1 CSR, but when I come 4th I lose 15 CSR? And how do other Onyx's get away with it?

Please explain.
QX wrote:
Sturbz wrote:
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
I get that you are angry about kills per minute being used as a metric for determining which team will win with the best accuracy. Just to spark discussion though I have a few questions.

What would you like the matchmaker to use instead of KPM even though every other trackable stat is worse or doesn't improve upon the determination made just based on KPM. Remember 343 has never said that the matchmaker is always right. In fact they proudly state that it is only correct 68% of the time, which is up from the high 50s and low 60s of trueskill 1.

What would you like the skill ranking system do when a player wins? Grant a set amount of CSR gain and loss? I know the matchmaker is frustrating but should the ranks in this game be based on who plays the most? It's a hard decision and I wish h5 had both including a separate team rank that is only based on win/loss.

I see a lot of anger coming from people in this thread about how broken the matchmaker is but there is little discussion or points made on how they would improve it.

And in terms of the OP: I think the first game the matchmaker was wrong since it thought your team would do better. In terms of the players on the enemy team I know two of them that are honestly barely diamonds and are bad. The other two are good SWAT players however. In the second game it looks like you performed as expected of a high diamond with the champion level teammates you had and against the same team with two good players and two not so great ones. Since you performed as expected in that scenario why should you go up in rank? (I don't actually know MMR data so I could be wrong but those are just my theories for the two games).
TS1 was less accurate but more consistent so you can not find platinums with K/D of 3 and win ratio over 70% or player with 2 accounts where the main is bronze 1 and the other account is champion 1. (smurf tend to have at least 1 or 2 rank above the main)
TS2 is rewarding lucky players not better players so at this point i prefer a system where you can grind.
-15/-1/+1/+15 is not working because when you begin to lose 15 points per lost and to earn 1 per win, you don't encourage players to play.
QX wrote:
Sturbz wrote:
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
I get that you are angry about kills per minute being used as a metric for determining which team will win with the best accuracy. Just to spark discussion though I have a few questions.

What would you like the matchmaker to use instead of KPM even though every other trackable stat is worse or doesn't improve upon the determination made just based on KPM. Remember 343 has never said that the matchmaker is always right. In fact they proudly state that it is only correct 68% of the time, which is up from the high 50s and low 60s of trueskill 1.

What would you like the skill ranking system do when a player wins? Grant a set amount of CSR gain and loss? I know the matchmaker is frustrating but should the ranks in this game be based on who plays the most? It's a hard decision and I wish h5 had both including a separate team rank that is only based on win/loss.

I see a lot of anger coming from people in this thread about how broken the matchmaker is but there is little discussion or points made on how they would improve it.

And in terms of the OP: I think the first game the matchmaker was wrong since it thought your team would do better. In terms of the players on the enemy team I know two of them that are honestly barely diamonds and are bad. The other two are good SWAT players however. In the second game it looks like you performed as expected of a high diamond with the champion level teammates you had and against the same team with two good players and two not so great ones. Since you performed as expected in that scenario why should you go up in rank? (I don't actually know MMR data so I could be wrong but those are just my theories for the two games).
TS1 was less accurate but more consistent so you can not find platinums with K/D of 3 and win ratio over 70% or player with 2 accounts where the main is bronze 1 and the other account is champion 1. (smurf tend to have at least 1 or 2 rank above the main)
TS2 is rewarding lucky players not better players so at this point i prefer a system where you can grind.
-15/-1/+1/+15 is not working because when you begin to lose 15 points per lost and to earn 1 per win, you don't encourage players to play.
I perfer the new system. I feel that it shows MY skill. I have a group that I play with and we win far more than we lose. My rank would be much higher with the old ranking system, but it would not be a true reflection of my skill. You want your rank to get higher? Play better. Maybe improve how well you read the map, help weaker players, be more agressive, or die less. Ask people you play with what they think is your weakness. Play with players better than you and learn what they are doing better than you. Your rank isn't going up because you are NOT improving.
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
QX wrote:
Sturbz wrote:
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
I get that you are angry about kills per minute being used as a metric for determining which team will win with the best accuracy. Just to spark discussion though I have a few questions.

What would you like the matchmaker to use instead of KPM even though every other trackable stat is worse or doesn't improve upon the determination made just based on KPM. Remember 343 has never said that the matchmaker is always right. In fact they proudly state that it is only correct 68% of the time, which is up from the high 50s and low 60s of trueskill 1.

What would you like the skill ranking system do when a player wins? Grant a set amount of CSR gain and loss? I know the matchmaker is frustrating but should the ranks in this game be based on who plays the most? It's a hard decision and I wish h5 had both including a separate team rank that is only based on win/loss.

I see a lot of anger coming from people in this thread about how broken the matchmaker is but there is little discussion or points made on how they would improve it.

And in terms of the OP: I think the first game the matchmaker was wrong since it thought your team would do better. In terms of the players on the enemy team I know two of them that are honestly barely diamonds and are bad. The other two are good SWAT players however. In the second game it looks like you performed as expected of a high diamond with the champion level teammates you had and against the same team with two good players and two not so great ones. Since you performed as expected in that scenario why should you go up in rank? (I don't actually know MMR data so I could be wrong but those are just my theories for the two games).
TS1 was less accurate but more consistent so you can not find platinums with K/D of 3 and win ratio over 70% or player with 2 accounts where the main is bronze 1 and the other account is champion 1. (smurf tend to have at least 1 or 2 rank above the main)
TS2 is rewarding lucky players not better players so at this point i prefer a system where you can grind.
-15/-1/+1/+15 is not working because when you begin to lose 15 points per lost and to earn 1 per win, you don't encourage players to play.
I perfer the new system. I feel that it shows MY skill. I have a group that I play with and we win far more than we lose. My rank would be much higher with the old ranking system, but it would not be a true reflection of my skill. You want your rank to get higher? Play better. Maybe improve how well you read the map, help weaker players, be more agressive, or die less. Ask people you play with what they think is your weakness. Play with players better than you and learn what they are doing better than you. Your rank isn't going up because you are NOT improving.
Well that's what me and a few others are talking about. You get to that higher rank, then you win 4 games in the higher tier, you only get 4 points. Then if you lose, it's -15. So then you have to grind, again to level up. So I'll ask you as I've asked the monitors. How can you play better players, when you don't even get to play them if you lose one game in the higher tier?
QX wrote:
Sturbz wrote:
I just had 2 swat games in a row against the same to4 searching solo.
Game 1 was a blowout loss and I went -3 and got knocked down to d5 loosing a 1/4 rank.
Game 2 was an evenly matched win and I went +5 and got a minuscule csr update.

Why didn’t I get back the same csr that I lost ??
cause true skill 2.0 doesn't work, that's why. Your kills per minute is that of a d5 so the game's trying to bring your csr down to where your mmr is. Your confusion in the situation most likely stems from the fact that kills per minute cannot accurately represent skill
I get that you are angry about kills per minute being used as a metric for determining which team will win with the best accuracy. Just to spark discussion though I have a few questions.

What would you like the matchmaker to use instead of KPM even though every other trackable stat is worse or doesn't improve upon the determination made just based on KPM. Remember 343 has never said that the matchmaker is always right. In fact they proudly state that it is only correct 68% of the time, which is up from the high 50s and low 60s of trueskill 1.

What would you like the skill ranking system do when a player wins? Grant a set amount of CSR gain and loss? I know the matchmaker is frustrating but should the ranks in this game be based on who plays the most? It's a hard decision and I wish h5 had both including a separate team rank that is only based on win/loss.

I see a lot of anger coming from people in this thread about how broken the matchmaker is but there is little discussion or points made on how they would improve it.

And in terms of the OP: I think the first game the matchmaker was wrong since it thought your team would do better. In terms of the players on the enemy team I know two of them that are honestly barely diamonds and are bad. The other two are good SWAT players however. In the second game it looks like you performed as expected of a high diamond with the champion level teammates you had and against the same team with two good players and two not so great ones. Since you performed as expected in that scenario why should you go up in rank? (I don't actually know MMR data so I could be wrong but those are just my theories for the two games).
TS1 was less accurate but more consistent so you can not find platinums with K/D of 3 and win ratio over 70% or player with 2 accounts where the main is bronze 1 and the other account is champion 1. (smurf tend to have at least 1 or 2 rank above the main)
TS2 is rewarding lucky players not better players so at this point i prefer a system where you can grind.
-15/-1/+1/+15 is not working because when you begin to lose 15 points per lost and to earn 1 per win, you don't encourage players to play.
I perfer the new system. I feel that it shows MY skill. I have a group that I play with and we win far more than we lose. My rank would be much higher with the old ranking system, but it would not be a true reflection of my skill. You want your rank to get higher? Play better. Maybe improve how well you read the map, help weaker players, be more agressive, or die less. Ask people you play with what they think is your weakness. Play with players better than you and learn what they are doing better than you. Your rank isn't going up because you are NOT improving.
how can a champion improving?
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
maybe use the placement for example i am the first of the winning team i get a plus 5 on my CSR gain and so on
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
maybe use the placement for example i am the first of the winning team i get a plus 5 on my CSR gain and so on
Unfortunately neither of these approaches will get you accurate measures of skill unless you account for how strong your opponents are compared to the rest of the population. Having high stats against the worst players doesn't mean anything.

Likewise, having average stats against great players also doesn't mean you're average.

Those methods will do terrible at actually predicting who will win the match, and predicting your kills and deaths.

TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
maybe use the placement for example i am the first of the winning team i get a plus 5 on my CSR gain and so on
I think that doesn't represent your skill good enough. Cause even if you got the most kills but went negative, why should you gain the most CSR ?
ZaedynFel wrote:
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
maybe use the placement for example i am the first of the winning team i get a plus 5 on my CSR gain and so on
Unfortunately neither of these approaches will get you accurate measures of skill unless you account for how strong your opponents are compared to the rest of the population. Having high stats against the worst players doesn't mean anything.Likewise, having average stats against great players also doesn't mean you're average.

Those methods will do terrible at actually predicting who will win the match, and predicting your kills and deaths.

TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
Unfortunately TS2 is terrible to measure skill because many smurf are one tier above the main
ZaedynFel wrote:
TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
I honestly would like to see some raw data to support that because on one of my gamertags I had back to back 20 kill games. Yet, I am not seeing myself facing challenging people on that tag.
tmaker502 wrote:
ZaedynFel wrote:
TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
I honestly would like to see some raw data to support that because on one of my gamertags I had back to back 20 kill games. Yet, I am not seeing myself facing challenging people on that tag.
He said TS2 is good at predicting winners and performance. He didn’t say that TS2 is going to guarantee you evenly-matched opponents. You getting easy matches doesn’t mean TS2 isn’t correctly predicting the outcome. It just means that a better match couldn’t be found in the time you took searching for a game. That’s generally a consequence of low population.
ZaedynFel wrote:
Ok I have an idea. Let's keep CSR to a points system. So if you win, add up total kills. So if you went for example 15-7, thats 8 CSR points. Then lets add .5 points for every assist. So say you went 15-7 with 4 assists, you go up 10 CSR. Then, we can add .1 for all medals obtained. And if you are on the losing team, the kills you get can help you lose less points. With a cap 20 CSR points gained for a win, and the lowest being 2 gained if you played a horrible game. Like if you went 3-14. And same for if you lose. Maximum of 20CSR loss for a completley bad game and 2 CSR loss if you had a good game. If anyone could elaborate or other ideas to throw in, let's hear em!! Or if you think it sucks, let's hear that too!!
maybe use the placement for example i am the first of the winning team i get a plus 5 on my CSR gain and so on
Unfortunately neither of these approaches will get you accurate measures of skill unless you account for how strong your opponents are compared to the rest of the population. Having high stats against the worst players doesn't mean anything.Likewise, having average stats against great players also doesn't mean you're average.

Those methods will do terrible at actually predicting who will win the match, and predicting your kills and deaths.

TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
Unfortunately TS2 is terrible to measure skill because many smurf are one tier above the main
I have not seen this. Show me an example. Every case I've seen the smurf has been within an expected gap from the main.
tmaker502 wrote:
ZaedynFel wrote:
TrueSkill2, on the other hand, is doing an amazing job at predicting who will win each match, and how many kills and deaths each player will have. So we are already accurately measuring how good each player is.
I honestly would like to see some raw data to support that because on one of my gamertags I had back to back 20 kill games. Yet, I am not seeing myself facing challenging people on that tag.
Plenty of raw data in the TrueSkill2 paper.
  1. 1
  2. ...
  3. 45
  4. 46
  5. 47
  6. 48
  7. 49
  8. ...
  9. 50