Forums / Community / Matchmaking Feedback & Discussion

[Locked] Matchmaking Feedback Update – June 11

OP ZaedynFel

  1. 1
  2. ...
  3. 44
  4. 45
  5. 46
  6. 47
  7. 48
  8. ...
  9. 50
I think the new ranking system is good at sorting people by skill but that's where the positives end. It's started to change mine and my teammates attitude to playing, no longer are we just concerned with winning but it makes us play more selfish to preserve our personal stats. I'm now finding myself not wanting to capture the risky stronghold because I might die and that affects my ability to rank.

I think this system was implemented with good intentions to try and tackle problems with all ranking systems:
  1. weak players boosted to high ranks
  2. losing rank because of other players
  3. stopping smurfs
Unfortunately something happened today which has never happened before. I was playing with my doubles partner he was champ ~50 I was diamond 1. I had capped out and was getting 1 csr per game for a few games and he had just capped out. We lost the next game and both lost a large amount of CSR. We both agreed to quit playing, why bother grinding ranked when we aren't going to get anywhere. We are playing as hard as we can and the system says "you're ranked here" winning doesn't matter anymore so why bother to play if you know you're not going to play any better?

I think it's good to rank accurately but not of the cost of the experience being fun and addicting. I would like to see the stats of the amount of ranked games being played this season, I think there's a good chance they are down.

What needs to be changed?
Team games solely ranked on winning/losing even at the cost of returning to smurfs and people getting boosted.
Not getting "rank locked" although this may be your skill it removes the FUN and the motivation to play
The "expected to win or lose" doesnt seem to work. That means the game gave me about 15 games straight where i was expected to win since i gained no csr. Then the next game i lost and my csr was reduced tremendously. So that means i was expected to win that last game right? I can tell you those players were way more skilled then I. And also why am i getting 15 games where im expected to win anyways? I think it should resort back to the original system but with the addition of the better match ups of ts2. If someone gets carried then perhaps give them slightly less csr idk but right now the rank grind still feels very unpleasant
Sturbz wrote:
So that means i was expected to win that last game right? I can tell you those players were way more skilled then I.
Or maybe it is because your CSR is inflated over your MMR?
Sturbz wrote:
So that means i was expected to win that last game right? I can tell you those players were way more skilled then I.
Or maybe it is because your CSR is inflated over your MMR?
I dont see how. I started at platinum 3 then got shot up to diamond 3 over a few days then it just stopped giving me csr despite going on hugeee win streaks
Sturbz wrote:
Then the next game i lost and my csr was reduced tremendously. So that means i was expected to win that last game right? I can tell you those players were way more skilled then I.
No. The game you're referring to, where you lost CSR for losing, was not necessarily "expecting you to win". As you said, you qualified at Plat 3 and kept winning until you reached Diamond 3, then you stopped getting CSR. So that means Diamond 3 was your MMR (if you qualified at Plat 3, it means your MMR at the time of qualifying would have been Diamond 1, so your MMR also must have gone up a bit after your qualifying matches). So every win after getting to Diamond 3 gave you +1 CSR, and assuming your MMR stayed the same (because you were playing exactly like a Diamond 3 would in those games), your CSR would still be getting too high, just very slowly. So you win 15 games, get +15 CSR, but your MMR stayed put. Ergo, your CSR was still too high so when you finally lost, you got dinged with -15 CSR. In this loss, you could have either performed well enough that your MMR did not go down, or maybe your MMR went down with your CSR. Either way, the -15 CSR you got from the loss would still bring your rank closer to your skill after the slow divergence you had from your win streak. Of course, this is a possible explanation, not necessarily what happened to you. But it's an example that shows you weren't expected to win that game against stronger opponents. CSR drop does not necessarily mean MMR drop.
Sturbz wrote:
And also why am i getting 15 games where im expected to win anyways?
A mix of bad population numbers and your skill being above average, I'm guessing. If you're better than most players already, then most of the people searching at any given moment are probably weaker than you. And if the game can't find you an even match, it will put you in a slightly more uneven one to make sure you get to play. Could have also just been dumb luck.
Sturbz wrote:
but right now the rank grind still feels very unpleasant
Josh has said that he doesn't want ranks to be grindable. He wants the ranks to represent your actual skill. So if you play like a Diamond 3 regularly, then your rank will be Diamond 3. If you want to get up to the next rank, you've got to regularly play like a Diamond 4 would, even in the games where you are expected to win. Since that info isn't currently available, all you can do is play your best every game and hope it's enough. With enough games played, skill should increase, but gradually. If you play like a Diamond 3 today, I wouldn't expect you to play like a Diamond 4 or 5 tomorrow. I don't think any sort of training yields results that fast, even in games.
Chimera30 wrote:
Sturbz wrote:
Then the next game i lost and my csr was reduced tremendously. So that means i was expected to win that last game right? I can tell you those players were way more skilled then I.
No. The game you're referring to, where you lost CSR for losing, was not necessarily "expecting you to win". As you said, you qualified at Plat 3 and kept winning until you reached Diamond 3, then you stopped getting CSR. So that means Diamond 3 was your MMR (if you qualified at Plat 3, it means your MMR at the time of qualifying would have been Diamond 1, so your MMR also must have gone up a bit after your qualifying matches). So every win after getting to Diamond 3 gave you +1 CSR, and assuming your MMR stayed the same (because you were playing exactly like a Diamond 3 would in those games), your CSR would still be getting too high, just very slowly. So you win 15 games, get +15 CSR, but your MMR stayed put. Ergo, your CSR was still too high so when you finally lost, you got dinged with -15 CSR. In this loss, you could have either performed well enough that your MMR did not go down, or maybe your MMR went down with your CSR. Either way, the -15 CSR you got from the loss would still bring your rank closer to your skill after the slow divergence you had from your win streak. Of course, this is a possible explanation, not necessarily what happened to you. But it's an example that shows you weren't expected to win that game against stronger opponents. CSR drop does not necessarily mean MMR drop.
Sturbz wrote:
And also why am i getting 15 games where im expected to win anyways?
A mix of bad population numbers and your skill being above average, I'm guessing. If you're better than most players already, then most of the people searching at any given moment are probably weaker than you. And if the game can't find you an even match, it will put you in a slightly more uneven one to make sure you get to play. Could have also just been dumb luck.
Sturbz wrote:
but right now the rank grind still feels very unpleasant
Josh has said that he doesn't want ranks to be grindable. He wants the ranks to represent your actual skill. So if you play like a Diamond 3 regularly, then your rank will be Diamond 3. If you want to get up to the next rank, you've got to regularly play like a Diamond 4 would, even in the games where you are expected to win. Since that info isn't currently available, all you can do is play your best every game and hope it's enough. With enough games played, skill should increase, but gradually. If you play like a Diamond 3 today, I wouldn't expect you to play like a Diamond 4 or 5 tomorrow. I don't think any sort of training yields results that fast, even in games.
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
Still
QX wrote:
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
Still very flawed if one wants to determine ones "skill" Why would you even want to pick up the odd ball anymore?
Sturbz wrote:
Still
QX wrote:
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
Still very flawed if one wants to determine ones "skill" Why would you even want to pick up the odd ball anymore?
Agreed. We run slayer in a full team of onyx and champ. Most game we destroy the other team 50-20 on average. But because we are all good players it is difficult to get a high MMR on the players we are facing plat/diamonds. As we are dropping 12 kills each the system may be wanting us to drop 20 each which is impossible when we have 4 good players all trying to get a good kpm as well. The fix would be updating the MMR code to suit everyone. For example, if a player holds the oddball half the game then half their kpm goal that way it would work out normal still. If they hold the ball half the game they are never going to reach their goal unless they play like a god.
QX wrote:
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
That explains my slayer placing. I was running with my friends who are better than me, we were matching with onyx players and champs in a few games. I was always breaking even or going slightly negative although I had a total positive kd from the first easy games. I got placed gold 6, it made no sense how can I hold my own against champ level players and get gold.

i decided to solo search and then went 15-1 and 14-0....
Sturbz wrote:
Still
QX wrote:
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
Still very flawed if one wants to determine ones "skill" Why would you even want to pick up the odd ball anymore?
May be flawed, but it's less flawed than what came before. Where Trueskill1 correctly predicted the winner of a match only ~60% of the time, Trueskill2 is correct like 68% of the time. Perfect? No. But still better.
This post has been edited by a moderator. Please do not call out individuals. This includes forum members, moderators, administrators, and non-forum members.
*Original post. Click at your own discretion.
Spoiler:
Show
Boomy EU wrote:
Sturbz wrote:
Still
QX wrote:
Sturbz wrote:
Chimera30 wrote:
Sturbz wrote:
Chimera, thank you ever so much for that very detailed explanation. I feel much more comfortable with the system now. If you have the time could you possibly answer my two remaining questions? They are as follows:
1.) What exactly determines an increase or decrease in mmr? Is it kda, accuracy, damage, kpm, or dpm?
2.) How often does mmr fluctuate? If you have a really bad game does your mmr immediately drop or does it factor in consistency first?
  1. According to Josh, TS2 predicts your skill based on kills per minute. So what I take that to mean is that it defines a skill by the average kills per minute of all players at that skill level. The prediction will be different based on your opponents. E.g. a Diamond 3 could be expected to get 12 kills per minute against a Plat 6, but only 9 kills per minute against another Diamond 3 (not actual numbers, just example).
  2. MMR can change wildly from one game, or it can not change at all. Last season, someone went up 1200 points in MMR from one game. So if you have a really lucky or unlucky game, your MMR may go up/down alot. But it should still average at your normal skill level. CSR, on the other hand, does not move as wildly. It goes up either +/-15, or +/-1, depending on if you're inflated/deflated. If you quit, you get -30 CSR as punishment. This is one of the benefits of using CSR as public rank; because if MMR was your rank and you had one really bad game and went down from Diamond to Silver, you'd feel pretty bad.
Many thanks for another great response. Is mmr seriously just kpm? Seems very uninclusive especially regarding objective game types or players that play support roles or hold a spot with little action for the sake of controlling spawns
The reason for using kpm is that Josh and 343 ran the data and found that kpm and consequently DPM were the best indicators of which team would win the match. Basically those two data points would provide the system with the most accurate guess as to which team would win.

They tried other stats such as assists and obj work and accuracy but they either reduced the ability for the system to determine who would win or just make it take longer with no significant improvement.

So yes KPM is an important aspect of your rank and a second team rank only based on win/loss would be good to have but is unavailable in H5. However, I see a lot of players that now think that the only thing that matters is stealing kills and screwing over teammates. But winning still matters since your CSR can only go up from a win and a win will cause a net positive change to your team's combined MMR. The losing team will have their net MMR change be negative. So if you do really well and your teammates underperform you might go up in MMR even from a loss but that won't happen as many times as it goes up from a win.

The system also uses KPM which benefits from fast games where your team plays the obj and ends the game. After all a 3 minute game where you get 8 kills but 3-0 the other team is better than a 15 minute tie where you padded your stats to get 30 kills.
Still very flawed if one wants to determine ones "skill" Why would you even want to pick up the odd ball anymore?
Agreed. We run slayer in a full team of onyx and champ. Most game we destroy the other team 50-20 on average. But because we are all good players it is difficult to get a high MMR on the players we are facing plat/diamonds. As we are dropping 12 kills each the system may be wanting us to drop 20 each which is impossible when we have 4 good players all trying to get a good kpm as well. The fix would be updating the MMR code to suit everyone. For example, if a player holds the oddball half the game then half their kpm goal that way it would work out normal still. If they hold the ball half the game they are never going to reach their goal unless they play like a god.
Theoretically since it uses KPM instead of kills per game you playing with a bunch of higher level players should matter. In those blowout games you may only get 12 kills but you finish the game in 4 minutes instead of getting 20 kills in 8 minutes solo qing.

Also since the MMR curve is based on the population of everyone playing the gametype every ranks average kpm is lower in oddball than slayer. And even if someone never picks up the ball and only goes for kills I'm pretty sure that the system learns that and bases that players kpm on their game style (I think Josh has said this but I'm not sure and don't have confirmation). Now he has also said that everyone's mmr is lower in oddball than other gametypes in team arena. I don't know if that it because of the slower kpm or because it is a newer playlist and has less of a population to base ranks off of.

In a perfect world a lot of the concerns people have about the ranking system using kpm would be easily explain able but I can't say with confidence that there isn't an exploitable way to rank up. Or that only going for kills is 100% not the way to rank up. Again I think the system updates to understand when players play selfishly and ups their expected kpm compared to their obj players but I'm not sure.
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 still think kills per minute is and always will be a horrrrible way to determine mmr. Totally illogical
Weird. Been playing alot of Diamonds and lower ranked players but I still climb fast, and hardly lose CSR for a loss. Is FFA a grind? I feel like I don't belong at champ 41 (2044 CSR/MMR) Just look at the people I beat, yes I match champs ofen but to have get 1st 56% of my ffa games? Seems odd.
TraadeMerK wrote:
Weird. Been playing alot of Diamonds and lower ranked players but I still climb fast, and hardly lose CSR for a loss. Is FFA a grind? I feel like I don't belong at champ 41 (2044 CSR/MMR) Just look at the people I beat, yes I match champs ofen but to have get 1st 56% of my ffa games? Seems odd.
It is normal to meet lower ranked players, my friends (diamonds) meet 2 champion per game, i think you hardly lose CSR for a loss because you did at least 20 kills per game.
TraadeMerK wrote:
Weird. Been playing alot of Diamonds and lower ranked players but I still climb fast, and hardly lose CSR for a loss. Is FFA a grind? I feel like I don't belong at champ 41 (2044 CSR/MMR) Just look at the people I beat, yes I match champs ofen but to have get 1st 56% of my ffa games? Seems odd.
Merk you are near the top of population in terms of skill. Especially when FFA is a low population playlist. Since the clamp on FFA is going to be pretty low you will be considered a diamond even if you are onyx just so you can get games.
  1. 1
  2. ...
  3. 44
  4. 45
  5. 46
  6. 47
  7. 48
  8. ...
  9. 50