Forums / Community / Matchmaking Feedback & Discussion

[Locked] Downranking too harsh

OP ProMarmot477250

  1. 1
  2. ...
  3. 2
  4. ...
  5. 3
Lose a game? Lose nearly half a rank regardless of team makeup. Win a game with 3 unranked teamates against an onyx and 2 diamonds? Here's less than a quarter rank for ya.
Lose a game? Lose nearly half a rank regardless of team makeup. Win a game with 3 unranked teamates against an onyx and 2 diamonds? Here's less than a quarter rank for ya.
You most definitely do not lose half a rank regardless of team makeup. Link the games please.
https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/13e0f5f7-4915-48b7-9fa4-e7f856a96e0c/players/phat%20pat0101?gameHistoryMatchIndex=13&gameHistoryGameModeFilter=All

first game here both teams has a diamond 3 and a diamond 4. The other team had an onyx and plat 6 and my team had a diamond 5 and a diamond 3. Relatively similar ranks on both sides yet I lost almost half a rank from the loss

https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/eb5ba298-2510-453a-8500-6fcb211ccfcf/players/phat%20pat0101?gameHistoryMatchIndex=12&gameHistoryGameModeFilter=All

Second one here my team had 3 unranked players and me (diamond 4) and the other team had an onyx, a diamond 3, a diamond 2 and one unranked player yet that victory netted less than a quarter of a rank.

I've read enough posts on these forums to mostly understand how the ranking system works. I am just voicing my displeasure and frustration with it as I don't think it is very good.
Game 1:
You lose 20 CSR, which is close to what you would lose in an even match (15). So you lost slightly more than an even match. You got only 7 kills on an expected 11. Your in-match Performance was also 1320, so bringing your CSR from 1385 (too high) to 1365 (still too high but closer to your 1320 Performance) is reasonable. 20 is also not even half a rank (25) let a lone a full rank (50).

Game 2:
Two of the unranked players on your team were Onyx level. Which explains why two of them outscored everyone on your opponents' team. Remember, just because they haven't been ranked yet this season, doesn't mean we don't know exactly how good they are. Also, everyone but you on your team was in a to3, which the system also knows gives your team a higher advantage than usual. Your opponents were all solo queued. You underperformed in that game (11 kills where someone of your expected skill should have had more like 14), leaving your overall performance at 1338. So even 9 CSR (what you gained) is generous in this case given it moved you to a 1365, which is inflated. So you were basically carried in that match by the Onyx players and the fact they were in a to3. Your team's average MMR in that match was 1396, which was well above your own MMR (1324), so of course the system is going to expect the win to come more from the Onyx players than from you, and reward them accordingly --- and it was right.
I appreciate the explanation. However I don't agree with gaining or losing csr based on kills. Wouldn't KDA or KD make more sense?
I appreciate the explanation. However I don't agree with gaining or losing csr based on kills. Wouldn't KDA or KD make more sense?
No, that promotes lone-wolfing and prompts people to go for stats over teamwork.

Reach's Arena did this. it was terrible.
I appreciate the explanation. However I don't agree with gaining or losing csr based on kills. Wouldn't KDA or KD make more sense?
No, that promotes lone-wolfing and prompts people to go for stats over teamwork.

Reach's Arena did this. it was terrible.
Kills are the most cherished stat. How does kpm promote teamwork over kills, while KD wouldn't? In Team Arena, selfish play can easily get you enough wins to take advantage of the MMR boost from being selfish. I've played against and with many of them. And in regards to teamwork, I've been playing online since Halo 2 and basically there isn't any teamwork worth noting when solo queing

@Phat Pat0101
Csr isn't gained or loss for kills(per minute), but mmr is. Csr is only effected by winning and losing. The amount of csr lost or gained is determined by the relationship between your csr and mmr(which of course is greatly influenced by your kpm)
Kills are the most cherished stat. How does kpm promote teamwork over kills, while KD wouldn't?
K/D is a stat that isn't super representative of your performance in a game. A player can get 2-3 kills and then go hide the rest of the game to end up with a high K/D. Get a few kills and going to hide will end up making your kpm terrible, though, especially if the game drags out longer because of your absence.
Going for a good kpm means you'll always be looking for a fight, but still gotta be careful because if you aren't your dpm goes up. The best way to keep your kpm up and your dpm down is to work with your team so they can help you, and by extension you help them. Measuring kpm and dpm still puts kills as the most important factor in determining who will win or lose.
Chimera30 wrote:
Kills are the most cherished stat. How does kpm promote teamwork over kills, while KD wouldn't?
K/D is a stat that isn't super representative of your performance in a game. A player can get 2-3 kills and then go hide the rest of the game to end up with a high K/D. Get a few kills and going to hide will end up making your kpm terrible, though, especially if the game drags out longer because of your absence.
Going for a good kpm means you'll always be looking for a fight, but still gotta be careful because if you aren't your dpm goes up. The best way to keep your kpm up and your dpm down is to work with your team so they can help you, and by extension you help them. Measuring kpm and dpm still puts kills as the most important factor in determining who will win or lose.
Who does that? Isn't the miracle worker that does that consistently going to lose a lot matches? Essentially, everyone wants kills, including the other team. Which means hiding and doing what you propose is very difficult. Isn't this system supposed to be smart? Why would the system reward someone for going 2 and 0?

My understanding is that dpm has minimal influence compared to kpm
Chimera30 wrote:
Kills are the most cherished stat. How does kpm promote teamwork over kills, while KD wouldn't?
K/D is a stat that isn't super representative of your performance in a game. A player can get 2-3 kills and then go hide the rest of the game to end up with a high K/D. Get a few kills and going to hide will end up making your kpm terrible, though, especially if the game drags out longer because of your absence.
Going for a good kpm means you'll always be looking for a fight, but still gotta be careful because if you aren't your dpm goes up. The best way to keep your kpm up and your dpm down is to work with your team so they can help you, and by extension you help them. Measuring kpm and dpm still puts kills as the most important factor in determining who will win or lose.
Who does that? Isn't the miracle worker that does that consistently going to lose a lot matches? Essentially, everyone wants kills, including the other team. Which means hiding and doing what you propose is very difficult. Isn't this system supposed to be smart? Why would the system reward someone for going 2 and 0?

My understanding is that dpm has minimal influence compared to kpm
Exactly, someone who got 2 or 3 kills then went to hide would lose a ton of games based on their team always being in a 3 v 4 scenario so they would have a very difficult time ranking up. Also if the system weighted kpm and dpm equally then it would just be a straight K/D. You eliminate the time factor when you consider them both.

I understand the flaws of using KD as a indicator of performance but personally I think that not taking a ton of deaths is almost as important as kills in a game of slayer. I've been in many a game where the team leader in kills has a negative KD and I don't think that should be rewarded more than someone who kept their deaths down while still contributing to the team.

Maybe using a system that took into account KD and damage dealt could strike the balance between the two. That way you wouldn't be able to just stay out of battle but you would also have to be conscious of your deaths.
Chimera30 wrote:
Kills are the most cherished stat. How does kpm promote teamwork over kills, while KD wouldn't?
K/D is a stat that isn't super representative of your performance in a game. A player can get 2-3 kills and then go hide the rest of the game to end up with a high K/D. Get a few kills and going to hide will end up making your kpm terrible, though, especially if the game drags out longer because of your absence.
Going for a good kpm means you'll always be looking for a fight, but still gotta be careful because if you aren't your dpm goes up. The best way to keep your kpm up and your dpm down is to work with your team so they can help you, and by extension you help them. Measuring kpm and dpm still puts kills as the most important factor in determining who will win or lose.
Who does that? Isn't the miracle worker that does that consistently going to lose a lot matches? Essentially, everyone wants kills, including the other team. Which means hiding and doing what you propose is very difficult. Isn't this system supposed to be smart? Why would the system reward someone for going 2 and 0?

My understanding is that dpm has minimal influence compared to kpm
Exactly, someone who got 2 or 3 kills then went to hide would lose a ton of games based on their team always being in a 3 v 4 scenario so they would have a very difficult time ranking up. Also if the system weighted kpm and dpm equally then it would just be a straight K/D. You eliminate the time factor when you consider them both.

I understand the flaws of using KD as a indicator of performance but personally I think that not taking a ton of deaths is almost as important as kills in a game of slayer. I've been in many a game where the team leader in kills has a negative KD and I don't think that should be rewarded more than someone who kept their deaths down while still contributing to the team.

Maybe using a system that took into account KD and damage dealt could strike the balance between the two. That way you wouldn't be able to just stay out of battle but you would also have to be conscious of your deaths.
We've analyzed both K/D and Damage dealt and neither improve accuracy at predicting wins when added to the model.

The value the system places on kpm vs. dpm is also inferred directly from the data. It turns out that players who focus on dpm at the expense of kpm also lose more games.

This means getting a higher kpm yields more wins than an equally lower dpm across a ridiculously huge amount of matches, even if that goes against intuition.

Also, keep in mind that the balance between kpm and dpm is adapted per game mode per playlist. In objective modes, for example, we see a wider expectation on kills than in Slayer.

Also, to be clear, the model does not work by weighting kills and weighting deaths towards an MMR score. Instead, it forces your MMR to be indicative of who wins, and then uses that MMR to predict how many kills and deaths you will have. So the model accommodates kpm and dpm predictions within a win priority.

The model says: given your MMR and your teams MMR vs. your opponent teams MMR, will you win? how many kills and deaths will you have?

Then, if that prediction is wrong, it updates your MMR accordingly.
It would be cool if the system could use some logic when punishing "quitters". Like if my team is winning 47 to 39 and I'm 10 and 7 I think it's pretty obvious that I was dropped from that game and didn't quit. I think quitting should be punished very harshly because it ruins the game but in cases like that it seems a little harsh.
It would be cool if the system could use some logic when punishing "quitters". Like if my team is winning 47 to 39 and I'm 10 and 7 I think it's pretty obvious that I was dropped from that game and didn't quit. I think quitting should be punished very harshly because it ruins the game but in cases like that it seems a little harsh.
TrueSkill2 detects that and doesn't hit your MMR as hard.

The ban system is not currently aware of that though. Maybe someday.
ZaedynFel wrote:
It would be cool if the system could use some logic when punishing "quitters". Like if my team is winning 47 to 39 and I'm 10 and 7 I think it's pretty obvious that I was dropped from that game and didn't quit. I think quitting should be punished very harshly because it ruins the game but in cases like that it seems a little harsh.
TrueSkill2 detects that and doesn't hit your MMR as hard.

The ban system is not currently aware of that though. Maybe someday.
Thanks for the info. I was always curious about this as someone who gets disconnected from time to time. Not that I play ranked much anymore, but it's still nice to know that if you're doing fine and winning, then get disconnected, it doesn't hurt your rank as much.
ZaedynFel wrote:
It would be cool if the system could use some logic when punishing "quitters". Like if my team is winning 47 to 39 and I'm 10 and 7 I think it's pretty obvious that I was dropped from that game and didn't quit. I think quitting should be punished very harshly because it ruins the game but in cases like that it seems a little harsh.
TrueSkill2 detects that and doesn't hit your MMR as hard.

The ban system is not currently aware of that though. Maybe someday.
What is the current prediction success rate of the ranking system for this ranked season, if possible, or for last season?

I imagine that the matchmaker attempts to achieve matches as close to 50/50 as possible. What is the average that the matchmaker has obtained this season and/or last season in ranked playlists?
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
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
343. I love Halo with all my heart... I've recently been trying to go for the Champion rank in Slayer. But this is to no avail as Ill always get matched with just terrible teammates.
I mean let me ask y'all a question. Have you guys ever worked soooo hard, and put your all into something just for you to lose and get reprimanded, and their was nothing for you to do about it? The last game I played on Halo 5 I played was pain painstakingly dreadful. It was on the map "The Rig" a beautiful map that I absolutely love playing. Well unfortunately for me on my team we had good ol' redacted and redacted. redacted literally had a 23.7 accuracy. I dealt a total of 3,812 DMG that game. A whopping 1,600 more damage then the next person my team.
I have a multitude of Screenshots of me doing amazing and my teammates doing less then mediocre to say the least. ill literally have a 2.0 KD, not KDA, but KD and still lose. You know what the algorithm thinks? "Reprimand this player for doing extremely well and his random teammates that dont give a hoot about their CSR." Ill go 25 and 12 with 3,000 plus damage and still lose. Sometimes it feels like im going down by like 20-30 CSR.
I am posting this with so much emotion and headache. Please do something I was Onyx 1875 and SOOOO close to champ, but people like redacted get placed on my team and now im 1742.Pls respond soon... Wow that is disheartening... cant believe I had to edit out some gamer tags, who were never in their life going to see this post...
(I also didnt even get a response)
The system accounts for how bad your teammates are.

If you link a match I can break it down, but in every case so far that people have linked matches, the CSR has been correct.
It sucks losing CSR, it really does, but you'll get there dude! I got to 2k MMR/CSR mostly solo, I had games where my teammates were much less skilled and cost ke the game. But that's ok, it happens! Your CSR will catchup to your MMR assuming you aren't already near your true MMR.

But yes, you will have games where you drop 30 CSR, and sometimes a losing streak, but keep playing, keep getting better and your CSR will increase accordingly! The system is terrific in my opinion.
As awful as the ranking system is bro its unfortunately ( almost exclusively) calculated with W/L! I'm swinging between platinum 5 and 6 on ranked team arena because I can't get a dedicated team! My KDA is good but my W/D doesn't reflect that in the eyes of the system :( Your best bet is to either only play ranked team games with a FULL party or go it alone in FFA which is a bit different as you can place less than first and still rank up, not to mention YOU are team.. The current ranking system is really bad but its all we got right now. I did terrible in my qualifying 10 FFA matches and placed Gold 5 but right now I'm plat 2. I think its harder onced initially ranked but I'm still ranking up! Good luck man :D
  1. 1
  2. ...
  3. 2
  4. ...
  5. 3