Forums / Community / Matchmaking Feedback & Discussion

[Locked] Matchmaking Feedback Update – September 11

OP ZaedynFel

  1. 1
  2. ...
  3. 2
  4. 3
  5. ...
  6. 6
Weapon Tuning Changes to Matchmaking

Just a reminder that the new weapon tunings will go live today in Team Arena. For all the details, see THIS POST HERE.

For the curious, Team Arena population and engagement has roughly doubled with people trying the new settings, so if you haven’t, get in there and see what you think. The matchmaking in Team Arena right now is some of the best available in Ranked.


CSR (Rank) Boundaries

Some of you have noticed that Ranks feel a little “easier” this season. This is accurate. Before each new season, we evaluate the current Ranks distribution and update the boundaries. We found that the Rank boundaries going into the last season was too strict by the end of the season. The usual season update that we do has made it “feel” easier to get to some of the higher Ranks this season.

We haven’t made it “easy” to get into Onyx, per se, it was just too “hard” before, and now it’s more correct.

Also, keep in mind that this does NOT affect your matchmaking. You’re playing the same people as before, just with improved labels, so don’t “freak out” when you start playing more Onyx folks. Most of them are in the same boat as you, and are just as concerned as you are to play versus you.
Thank you! I'm excited to try the new weapons.

BR and Gunfighter Magnum starts was a great idea. It currently does not play too well because the BR is effectively a cross-map killer and once you were spotted it was likely you'd be dead in a few seconds.

But I found it useful to pop out my Gunfighter during close range encounters. It just feels nice to blast away at someone and get the kill.

Really hoping the new weapons create more dynamic differences between each one.
Just to continue on what was discussed in the last thread (was replying and got locked lol)

I'm sorry, but there's a big skill difference between say a 1550 onyx, someone who has just got there feet wet in onyx, so remember, there not far off from being diamond 6.... compared to say a 1900 onyx or champion 180[0] or something. [...]The skill gap is very real.
Quote:
ZaedynFel wrote I agree with you and we have a way to improve that in the pipeline. But we can't do it right now because it would prevent those Champs from being able to play the game at all.Ironically, I've had players tell me there's no difference between a 1500 Onyx and a Champ. The data, and your experience, say otherwise.
Just wanted to say that's good to hear and I understand. Myself and or my team will try not to be too mad the next time we're low Onyx and get stomped by champions lol ;) those players that think there's no difference in skill must either be really good or on drugs or something lol I'm glad you have data that backs what we think/say.

ZaedynFel wrote,
Quote:
For the curious, Team Arena population and engagement has roughly doubled with people trying the new settings, so if you haven’t, get in there and see what you think. The matchmaking in Team Arena right now is some of the best available in Ranked.
Really eh! Wow, that is good news and is kind of surprising to me. Glad to see people are still around/interested. I haven't yet but will be this week for sure!

Now, if you could only just show those population numbers of the playlists, that would be amazing ;)

EDIT: sorry for all the additional context, on my phone and hit submit by accident and I wasn't done yet.
Can you please update the SWAT playlist with more maps? It's only a semi-competitive mode anyway and I don't see why it can't have a lot more of the available maps.

The games are so quick (3-6 minutes) that it gets really repetitive with so few maps available.
FYI, I believe the new settings are live.
(Edit - Josh has corrected some erroneous assumptions I made below)

I've got a concern/question about the initial placement algorithm every season, which has bothered me for a while but I've finally got some evidence for my beliefs, by looking at my placement matches this season.

Does the method you guys you to "estimate" an updated MMR early in a season take kills and stats into account? This would then affect your initial CSR rank too (since it's based off your MMR after placement matches). I suspect it does, and obviously also considers wins/losses, opponent skill, and prior season MMR. BUT... there are a few big issues I find frustrating - summarized below.

  • I realize you have data evidence that kills are strongly correlated with winning games, so you use that as a predictor. I'm sorry, but that's not even close to the full story, especially in mixed-skill play. Here's a game I played recently in HCS (gold tier) where our team actually out-slayed the enemy but got absolutely destroyed 3-0 in CTF, because my teammates had no idea how to play objective, and weren't getting kills at important moments or map locations. I kept getting the other team 3-dead and immediately bringing the flag back 95% of the way back to our base, but my teammates couldn't protect our flag and convert it into a cap. The kill stats would suggest that we won the game or at least made it close, but no, we got completely spanked 3-0 in frustrating fashion. I see this kind of game often, I really do. Map positioning and smart play are more important than kills in objective gametypes. Making the enemy waste time while you make progress on the objective is the name of the game. Sometimes that means more kills, but not always.
  • My concern about a "kill bias" is that if you prefer to play with a certain friend (or fireteam), you get crushed on your initial CSR placement if you're an objective-focused player, while the main "slayers" get much higher ranks. This season I had a very odd placement. I played several of my HCS placement matches this season with friends,, and I placed into SILVER 6, despite being high gold recently. One friend placed into Plat 5. He's better, but not 2 full tiers better. I think the game is correct that our average skill is high gold / low plat, but it got the spread wrong. So in order for me to rank up to an appropriate CSR, I had to leave him and start running solo, which I did and easily ranked up to gold 4. The problem is that because of the bogus initial ranking placements, I was de-motivated to play with my buddy anymore because he was over-ranked and I was under-ranked! I would never have ranked up to my fair rank because our average MMR was correct. I'm not sure if this happened because of a "kill bias" or for some other reason.
  • Here was the second game of my placements, here was the last.
  • I often hear people complain "Why did my friend rank diamond and I ranked gold?!? We played all our placements together!!" It's frustrating.
  • I've posted about this before, but this experience still makes me very worried when you talk about incorporating more information about kill stats into MMR for future seasons or titles. I see so much evidence that it's an inadequate skill metric in many ways. I really hope you don't do this, or at least do it VERY carefully with an understanding that at some skill ranges, kills are not a sufficient skill indicator.
RzR J3ST3R wrote:
I've got a concern/question about the initial placement algorithm every season, which has bothered me for a while but I've finally got some evidence for my beliefs, by looking at my placement matches this season.

Does the method you guys you to "estimate" an updated MMR early in a season take kills and stats into account? This would then affect your initial CSR rank too (since it's based off your MMR after placement matches). I suspect it does, and obviously also considers wins/losses, opponent skill, and prior season MMR. BUT... there are a few big issues I find frustrating - summarized below.
No, kills weren't used at all in your placement.

They are only used if:
1) You already played in at least one other playlist anywhere in the system
2) You are playing in a new playlist you've never played in ever before.

In which case kills are only looked at in your first two matches, and only to make a rough adjustment so you aren't getting destroyed / destroying every match. So, like, if you get barely any kills, or if you get 20+, it'll move you a little back / forward, only once per playlist ever.

Regardless of season.
So what is looked at in determining CSR ranking? Hypothetically, if you play all ten placements with the same fireteam and win all of them, what determines whether one player gets a higher ranking than another? Is it strictly positional i.e. the person at the top of the leaderboard gets a higher MMR boost? Or are other factors involved (accuracy, damage, objective plays, etc)?
ZaedynFel wrote:
RzR J3ST3R wrote:
I've got a concern/question about the initial placement algorithm every season, which has bothered me for a while but I've finally got some evidence for my beliefs, by looking at my placement matches this season.

Does the method you guys you to "estimate" an updated MMR early in a season take kills and stats into account? This would then affect your initial CSR rank too (since it's based off your MMR after placement matches). I suspect it does, and obviously also considers wins/losses, opponent skill, and prior season MMR. BUT... there are a few big issues I find frustrating - summarized below.
No, kills weren't used at all in your placement.

They are only used if:
1) You already played in at least one other playlist anywhere in the system
2) You are playing in a new playlist you've never played in ever before.

In which case kills are only looked at in your first two matches, and only to make a rough adjustment so you aren't getting destroyed / destroying every match. So, like, if you get barely any kills, or if you get 20+, it'll move you a little back / forward, only once per playlist ever.

Regardless of season.
Ah, okay thanks that's very helpful - I misunderstood. The issue may have been that my teammates had erroneously high MMRs (maybe they had not played in a few seasons and their MMRs were out of date). That might have been a factor that screwed me into a lower rank than deserved.

As always, thanks for clarifying and for your constant attention to this thread.
yah0oit wrote:
So what is looked at in determining CSR ranking? Hypothetically, if you play all ten placements with the same fireteam and win all of them, what determines whether one player gets a higher ranking than another? Is it strictly positional i.e. the person at the top of the leaderboard gets a higher MMR boost? Or are other factors involved (accuracy, damage, objective plays, etc)?
Generally, only win / loss.

If everyone in the fire team is brand new and has never played any other playlist, then they will all have the same placements.

If everyone in the fire team is new to this playlist (no previous seasons) and have already played at least 10 matches in at least one other playlist, then the only thing that can result in different placement currently is if a player has an unusually high number of kills. If they do, the system bumps them up a little so that matchmaking is fairer next time.

In subsequent seasons, everything is then based purely off MMR and win / loss.
RzR J3ST3R wrote:
ZaedynFel wrote:
RzR J3ST3R wrote:
I've got a concern/question about the initial placement algorithm every season, which has bothered me for a while but I've finally got some evidence for my beliefs, by looking at my placement matches this season.

Does the method you guys you to "estimate" an updated MMR early in a season take kills and stats into account? This would then affect your initial CSR rank too (since it's based off your MMR after placement matches). I suspect it does, and obviously also considers wins/losses, opponent skill, and prior season MMR. BUT... there are a few big issues I find frustrating - summarized below.
No, kills weren't used at all in your placement.

They are only used if:
1) You already played in at least one other playlist anywhere in the system
2) You are playing in a new playlist you've never played in ever before.

In which case kills are only looked at in your first two matches, and only to make a rough adjustment so you aren't getting destroyed / destroying every match. So, like, if you get barely any kills, or if you get 20+, it'll move you a little back / forward, only once per playlist ever.

Regardless of season.
Ah, okay thanks that's very helpful - I misunderstood. The issue may have been that my teammates had erroneously high MMRs (maybe they had not played in a few seasons and their MMRs were out of date). That might have been a factor that screwed me into a lower rank than deserved.

As always, thanks for clarifying and for your constant attention to this thread.
The boundary changes probably also affected your friend a bit. The farther out you are from the middle, the stronger the effect.
yah0oit wrote:
So what is looked at in determining CSR ranking? Hypothetically, if you play all ten placements with the same fireteam and win all of them, what determines whether one player gets a higher ranking than another? Is it strictly positional i.e. the person at the top of the leaderboard gets a higher MMR boost? Or are other factors involved (accuracy, damage, objective plays, etc)?
He's posted about this before, but initial CSR is determined by setting it a little below your MMR determined by your placement matches and previous seasons. It's all about wins/losses and the skills of the players in your match.

CSR adjustments are made after every game by comparing the MMRs of the two teams, and then awarding / stripping CSR to the winners and losers, in an amount proportional to the average MMR difference between teams. i.e. you beat a "better" team, you'll get a lot of CSR, and vice versa.
I think my Silver 1 SWAT CSR is wrong... 50% W/L ratio, 1.761 K/D.

Here are my 10 matches, in order:

  1. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/d370daef-9442-4796-85d4-126323c3cc44/players/radar3301enigma?gameHistoryMatchIndex=51&gameHistoryGameModeFilter=Arena
  2. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/c314620c-00a7-4a2d-a40c-c24bf28dca63/players/radar3301enigma?gameHistoryMatchIndex=46&gameHistoryGameModeFilter=Arena
  3. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/8b5dd6c1-42b4-40c5-b71b-cb743800a1a5/players/radar3301enigma?gameHistoryMatchIndex=45&gameHistoryGameModeFilter=Arena
  4. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/d5c46d96-b4a0-48c5-961b-d9c46f2b1644/players/radar3301enigma?gameHistoryMatchIndex=44&gameHistoryGameModeFilter=Arena
  5. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/155ebb38-9825-4ca3-8563-118a052035a1/players/radar3301enigma?gameHistoryMatchIndex=36&gameHistoryGameModeFilter=Arena
  6. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/855e1d30-19b8-4109-9949-4fc0970d630d/players/radar3301enigma?gameHistoryMatchIndex=35&gameHistoryGameModeFilter=Arena
    Played with To3
  7. ...
  8. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/5a393ab1-a2c3-4acb-b353-4ec1d88e2f13/players/radar3301enigma?gameHistoryMatchIndex=33&gameHistoryGameModeFilter=Arena
    Played with To3
  9. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/b87995ef-71a5-44a0-a8af-a1faa618a181/players/radar3301enigma?gameHistoryMatchIndex=32&gameHistoryGameModeFilter=Arena
    Played with To3, 1 teammate lost connection
  10. https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/35776cd5-a075-43d7-a09c-6582b2db2a09/players/radar3301enigma?gameHistoryMatchIndex=31&gameHistoryGameModeFilter=Arena
    Played with To2
I know CSR isn't the same as MMR, but really? I was playing almost Golds, and frequently got #1 spot in the rankings.

Also, I thought searching Focused discouraged joining an in-progress Warzone match? I got 4 in a row...

Additionally, is the Slayer high-Silver/mid-Gold population so low, that it really takes an average right around 2 minutes searching solo focused? Have more people started teaming up now?
Changing weapon damage after two years is a bad idea. I'm not against change, if this had happened eighteen months ago I would be less resistent but, after thousands and thousands of games played, please don't mess with how I expect weapons to work. This is UX 101 stuff. Why anyone thinks this is a good idea is beyond me.
I think my Silver 1 SWAT CSR is wrong... 50% W/L ratio, 1.761 K/D.

You were beating bronze kids and losing to silver kids. So low silver sounds about right to me. K/D and win ratio mean nothing, it's all about HOW GOOD the teams you beat are. You want to be high silver, gotta beat teams of silvers.
I think my Silver 1 SWAT CSR is wrong... 50% W/L ratio, 1.761 K/D.

Here are my 10 matches, in order:

I know CSR isn't the same as MMR, but really? I was playing almost Golds, and frequently got #1 spot in the rankings.

Also, I thought searching Focused discouraged joining an in-progress Warzone match? I got 4 in a row...

Additionally, is the Slayer high-Silver/mid-Gold population so low, that it really takes an average right around 2 minutes searching solo focused? Have more people started teaming up now?
CSR placement only looks at where you were in the previous season, whether you won or lost placement matches, and how hard each match was. It doesn't care about your spot in the rankings or KD.

So, in your case, you went 50/50 and so it stayed where you were last season for the most part.
Man, you guys are getting good at answering these for me. I don't even have to post anymore.
ZaedynFel wrote:
Man, you guys are getting good at answering these for me. I don't even have to post anymore.
It's like we learn and pay attention. :D
Additionally, is the Slayer high-Silver/mid-Gold population so low, that it really takes an average right around 2 minutes searching solo focused? Have more people started teaming up now?
Well, searching focused limits the available of pool of players you're able to match with because the system is strictly trying to find players closed to your rank. If there's hardly anyone around your rank, then that drives up search times. That's the tradeoff of searching focused vs. Balanced/Expanded.

ZaedynFel wrote:
Man, you guys are getting good at answering these for me. I don't even have to post anymore.
Heh, now that some people have a pretty good general understanding of how the system works thanks to all your feedback threads up to this point, it's probably really only the super outlier matches that really need your input. But this being the internet, it'll still take you, for the most part, echoing someone else's input or adding to it for certain people to accept answers. Word of God, and all that jazz, ya know?
SGO SMACK wrote:
Changing weapon damage after two years is a bad idea. I'm not against change, if this had happened eighteen months ago I would be less resistent but, after thousands and thousands of games played, please don't mess with how I expect weapons to work. This is UX 101 stuff. Why anyone thinks this is a good idea is beyond me.
I complealy agree. I played for a good 5 hours today and the only changes I even slightly like is the sword, because it doesn't make flag runs so hard to stop anymore, and the beam rifle. Everything else is just bad and is going to kill the games population, expecialy in social playlists like btb.
  1. 1
  2. ...
  3. 2
  4. 3
  5. ...
  6. 6