Suggested changes for matchmaking rank system

Games Halo Wars Series
1

There have been a lot of complaints about the matchmaking rank system, so I wanted to create a post to suggest some changes that might hopefully improve the experience.

As with all complaints about matchmaking this is spurred on by a slow grind of wins followed by a dramatic fall from a few games. My team mate and I were playing in 2v2 and we were getting consistent small wins from playing people of similar or lesser ranks. We then faced off against an Onyx ranked player (his max rank obtained being champion #16), ended up loosing (twice), and because his team mate was unranked in the 2v2 playlist (he was also a max rank of champ) I ended up loosing all the progress I had made and much more. I should add that I’m not basing this post off of just one bad instance, but a rather an accumulation of frustration, with this as the most recent example.

There are two issues I think here:

  1. The player population is too small such that players are being matched against too broad skill levels. [This is the main point]
  2. Unranked players playing with their high level partners are skewing the gains/penalties too much.

Suggested changes:

  1. Cap the % / points you can loose in a single game. This will prevent a bad or unfairly matched game from completely demoralising players. Random disconnects are still a thing as well, so this would make that less frustrating.

  2. Introduce a minimum % / points gain for a game won. Again, this is would be to recognise that the player pop is too low for the total spread of available ranks and help keep players motivated even after a bad loss.

  3. If the devs are reluctant to put in place caps, and I don’t know the ins and outs of how the system works - so this may not even be feasible, then I would say that the overall number of ranks needs to be drastically reduced. So instead of 6 levels per rank it should be cut down to 3 (or even 2). The fact that it’s (from what I can tell) nearly impossible to place in bronze is suggestive that there are simply too many ranks for the player pop.

  4. Global team deathmatch rank i.e. One single rank for 2v2 and 3v3 playlists. This is a more tentative suggestion that I’m not completely sold on, but for most players if you’re good in a 2v2, then chances are you’re gonna be just as good in a 3v3. I left out 1v1 because it is quite a different style, and a lot of the high-skill players will likely want to keep that separate from other ranks. This suggestion is meant to target instances where you loose against good players who just happen to not have played much of a given playlist. This is a more tentative suggestion mainly because of differences in the maps etc …

Without knowing exactly how csr gains and losses are calculated and what ‘% changes’ below Onyx mean it’s difficult to say what the caps should be. Psychologically a +10% or 10 point gain for a win is enough to keep me happy, and any more than a -30% or 20 point loss will demoralise me - but only under the current system of meagre gains for consistently being matched against lower ranked players. However without knowing exactly how this would work ‘under the hood’ that’s a totally subjective point.

This is a topic many will have views on, and many people have raised issue with, so hopefully something can be changed to make the experience of playing what is a fantastic game even more enjoyable.

2

We had this discussion quite a few times already.
EiTeNeR and I have tried to derive a model for the CSR adjustments based on the MMR of the players.
I think we developed a good understanding of what is happening behind the scenes even though there’s still a tiny detail missing.
Join us in the #data-lab channel of the The Banished discord if you want to join our discussion.
We have some nice plots there that show how MMR variance, rating, delta affect the CSR adjustments for specific playlists.

Your points:

1) Correct.
It’s very important to understand that the matchmaking itself does a decent job given the current playerbase.
What needs to be fixed is the CSR adjustment AFTER a match, NOT the matchmaking!
So we cannot fix the situation that a player is repeatedly paired with much worse opponents as that is simply an issue with the current size of the community.
What we could fix, though, is the situation where you lose a significant amount of CSR progress against a TEAM with WRONG MMRs (which can be either smurfs or inactive but good players).

3) While you are correct that we probably have too many ranks, I think the proposal is unrealistic.
Speaking of the Halo Wars API it seems that they built everything around the CSR designations & tiers (Diamond 4 etc).
Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.

4) I don’t like that idea.
As you said 1v1 is quite different from 2v2 and so is 3v3.
Having a single MMR across (a subset of) playlists does not sound very reasonable to me especially as this alone won’t really solve the issue.
In the problematic matches an inactive player or smurf pairs up with a stronger player.
Inactive players as well as smurfs will also have a wrong MMR across playlists.

Regarding 2) this also has been proposed a few times already.
Capping the CSR adjustments alone is not sufficient, but would only shift the problem.
Very heterogeneous teams will still cause higher, now capped, CSR adjustments.
As the problem still remains unsolved, you will have a hard life finding min/max adjustments that work and don’t destroy the rank progression.
It’s good that you lose CSR points on a loss, it’s just the amount that is sometimes wrong.

Having said that, I think there is an elegant way to solve this problem:
For teams where a strong player (that is, a player with a high mmr) pairs up with a much worse player or smurf (with a low mmr), we need to estimate the team strength based on the mmr of the dominant player.
If we added weight to higher mmr values the attributed strength of a team with an equal strength distribution would remain untouched while the problematic teams would appear slightly stronger.
This would lead to smaller CSR gains for the dominant players (=> smurfing becomes less effective) and to smaller CSR losses for normal teams (=> less frustration).

3

> 2533274933478052;2:
> Having said that, I think there is an elegant way to solve this problem:
> For teams where a strong player (that is, a player with a high mmr) pairs up with a much worse player or smurf (with a low mmr), we need to estimate the team strength based on the mmr of the **dominant player.**If we added weight to higher mmr values the attributed strength of a team with an equal strength distribution would remain untouched while the problematic teams would appear slightly stronger.
> This would lead to smaller CSR gains for the dominant players (=> smurfing becomes less effective) and to smaller CSR losses for normal teams (=> less frustration).

I like this idea but it becomes a problem for non - premades (although I would personally argue you shouldn’t be playing ranked if you aren’t a premade) I’m not sure there could be two different solutions in place for premades and non - premades. I suppose it could just be ignored for the hope of a better csr system but otherwise I’m not sure if there is a way to overcome it.

4

> 2533274933478052;2:
> Having said that, I think there is an elegant way to solve this problem:
> For teams where a strong player (that is, a player with a high mmr) pairs up with a much worse player or smurf (with a low mmr), we need to estimate the team strength based on the mmr of the **dominant player.**If we added weight to higher mmr values the attributed strength of a team with an equal strength distribution would remain untouched while the problematic teams would appear slightly stronger.
> This would lead to smaller CSR gains for the dominant players (=> smurfing becomes less effective) and to smaller CSR losses for normal teams (=> less frustration).

I like this suggestion - it would go a long way to fixing the situation I described above.

I admit the caps idea is really just treating the symptoms rather than the underlying problems, which you get at in your reply.

Although it wouldn’t necessarily solve the issue, reducing the overall number of ranks I think should happen anyway. There’s just too many for the current player base, and it just seems meaningless.

@Postums has made a few posts in other forums on this topic that suggest he and others are keen to solve matchmaking related problems, so I’m not sure I’d assume anything regarding resources before making suggestions.

The fourth point was really just a wild suggestion to address the issue of going up against players who are very good, but just inactive in a given playlist. I haven’t played that much of 3v3 war recently, but if I jumped in a game it would be totally unfair if platinum players or below loose a boat load of progress against me.

5

> 2533274823396923;1:
> There have been a lot of complaints about the matchmaking rank system, so I wanted to create a post to suggest some changes that might hopefully improve the experience.
>
> As with all complaints about matchmaking this is spurred on by a slow grind of wins followed by a dramatic fall from a few games. My team mate and I were playing in 2v2 and we were getting consistent small wins from playing people of similar or lesser ranks. We then faced off against an Onyx ranked player (his max rank obtained being champion #16), ended up loosing (twice), and because his team mate was unranked in the 2v2 playlist (he was also a max rank of champ) I ended up loosing all the progress I had made and much more. I should add that I’m not basing this post off of just one bad instance, but a rather an accumulation of frustration, with this as the most recent example.
>
> There are two issues I think here:
> 1) The player population is too small such that players are being matched against too broad skill levels. [This is the main point]2) Unranked players playing with their high level partners are skewing the gains/penalties too much.
>
> Suggested changes:
> 1) Cap the % / points you can loose in a single game. This will prevent a bad or unfairly matched game from completely demoralising players. Random disconnects are still a thing as well, so this would make that less frustrating.
>
> 2) Introduce a minimum % / points gain for a game won. Again, this is would be to recognise that the player pop is too low for the total spread of available ranks and help keep players motivated even after a bad loss.
>
> 3) If the devs are reluctant to put in place caps, and I don’t know the ins and outs of how the system works - so this may not even be feasible, then I would say that the overall number of ranks needs to be drastically reduced. So instead of 6 levels per rank it should be cut down to 3 (or even 2). The fact that it’s (from what I can tell) nearly impossible to place in bronze is suggestive that there are simply too many ranks for the player pop.
>
> 4) Global team deathmatch rank i.e. One single rank for 2v2 and 3v3 playlists. This is a more tentative suggestion that I’m not completely sold on, but for most players if you’re good in a 2v2, then chances are you’re gonna be just as good in a 3v3. I left out 1v1 because it is quite a different style, and a lot of the high-skill players will likely want to keep that separate from other ranks. This suggestion is meant to target instances where you loose against good players who just happen to not have played much of a given playlist. This is a more tentative suggestion mainly because of differences in the maps etc …
>
> Without knowing exactly how csr gains and losses are calculated and what ‘% changes’ below Onyx mean it’s difficult to say what the caps should be. Psychologically a +10% or 10 point gain for a win is enough to keep me happy, and any more than a -30% or 20 point loss will demoralise me - but only under the current system of meagre gains for consistently being matched against lower ranked players. However without knowing exactly how this would work ‘under the hood’ that’s a totally subjective point.
>
> This is a topic many will have views on, and many people have raised issue with, so hopefully something can be changed to make the experience of playing what is a fantastic game even more enjoyable.

How about rank doesn’t matter because even a casual player can become champion and since the rank doesn’t matter your not going to fix it with these adjustments. You could have diamonds that kill champions. Then you put them up against legit diamonds and there’s a problem.

6

> 2533274933478052;2:
> Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.

We have said that it would be difficult to do so, but at current time we are still investigating the system and are in no way promising that we will make changes to it at this time. As well, we have said that there are no plans to directly change how it functions at this time. We are looking into it.

As an update on the investigation, we have spoken with the team that built the entire system and in this research it is doing what it is designed to do at this time. The complaints which I do here don’t boil down to just one thing being the problem. There are a lot of factors that must be taken into account and could potentially be doing what players believe is a negative impact on the player experience. Such things are as follows.

  • Player population at time of searching. - Smurf accounts that have an unfair win to loss ratio due to nefarious behavior (boosters). - A player simply losing to someone who the system believes they shouldn’t have based on previous win/loss records (possibly a smurf or simply because they got out played).There are many more factors that go into it when a player matches and plays against another player and a very complex algorithm that is in place to calculate what a player will receive should they win or lose. An idea that has been floated which a lot of people are gravitating towards is that “players should get more points from a win.” However, in investigating the times where this happens - it should not be so in many cases.

When you lose at high levels of play or to people who are of different ranks than you - you lose and gain rank accordingly. If an onyx player beats a gold which they should based on how placements and season records work - you don’t receive a lot towards your next rank. Hence a lot of very small increments of points. If an onyx player beats another onyx player - they will and do gain more points than the win over the gold player. The same is done in reverse where if an onyx player loses to a gold - you will lose a lot of rank in comparison to if you lose to another onyx player.

Personally, I do not believe this is a problem with how the ranking system is working in itself based on the investigation that we have done and the conversations that we have had. I believe that the frustrations people are discussing is being perceived due to other things happening and how the matchmaking is choosing matches and when an unfortunate loss happens.

This all meaning that I believe this comes to a head when high level players lose to lower level players and end up losing a lot of rank as well as lower level players being matched against higher level players and having a bad matchmaking experience.

In Halo Wars 2 the system that chooses how to create a match is based on many different things which are not in this order of importance but include and are not limited to skill level of players being matched, the time players are searching and working to match, network statistics of the players searching, and many more elements in a complex algorithm.

There is no simple answer to these systems and how they interact, but know that we are investigating and looking into ways to better improve the experience for all players that are playing.

Cheers,
Postums

7

> 2533274792014876;6:
> > 2533274933478052;2:
> > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
>
> We have said that it would be difficult to do so, but at current time we are still investigating the system and are in no way promising that we will make changes to it at this time. As well, we have said that there are no plans to directly change how it functions at this time. We are looking into it.
>
> As an update on the investigation, we have spoken with the team that built the entire system and in this research it is doing what it is designed to do at this time. The complaints which I do here don’t boil down to just one thing being the problem. There are a lot of factors that must be taken into account and could potentially be doing what players believe is a negative impact on the player experience. Such things are as follows.
> - Player population at time of searching. - Smurf accounts that have an unfair win to loss ratio due to nefarious behavior (boosters). - A player simply losing to someone who the system believes they shouldn’t have based on previous win/loss records (possibly a smurf or simply because they got out played).There are many more factors that go into it when a player matches and plays against another player and a very complex algorithm that is in place to calculate what a player will receive should they win or lose. An idea that has been floated which a lot of people are gravitating towards is that “players should get more points from a win.” However, in investigating the times where this happens - it should not be so in many cases.
>
> When you lose at high levels of play or to people who are of different ranks than you - you lose and gain rank accordingly. If an onyx player beats a gold which they should based on how placements and season records work - you don’t receive a lot towards your next rank. Hence a lot of very small increments of points. If an onyx player beats another onyx player - they will and do gain more points than the win over the gold player. The same is done in reverse where if an onyx player loses to a gold - you will lose a lot of rank in comparison to if you lose to another onyx player.
>
> Personally, I do not believe this is a problem with how the ranking system is working in itself based on the investigation that we have done and the conversations that we have had. I believe that the frustrations people are discussing is being perceived due to other things happening and how the matchmaking is choosing matches and when an unfortunate loss happens.
>
> This all meaning that I believe this comes to a head when high level players lose to lower level players and end up losing a lot of rank as well as lower level players being matched against higher level players and having a bad matchmaking experience.
>
> In Halo Wars 2 the system that chooses how to create a match is based on many different things which are not in this order of importance but include and are not limited to skill level of players being matched, the time players are searching and working to match, network statistics of the players searching, and many more elements in a complex algorithm.
>
> There is no simple answer to these systems and how they interact, but know that we are investigating and looking into ways to better improve the experience for all players that are playing.
>
> Cheers,
> Postums

The only big complaint is I should NEVER have to play anyone below diamond if I’m in champion (unless they are in placements, then I can accept it). I would rather sit and wait then get a game that is just going to end in a win or if my xbox crashes/ game lags out (for whatever reason) then I lose a ton of csr normally 30+. Considering I get 0-1 csr from these games I would rather just not play them at all.

8

> 2533274792014876;6:
> > 2533274933478052;2:
> >

Postums First of all, thank you very much for this detailed response.
You apparently spent quite some time with the issue which I appreciate a lot!
You talked about not touching the game code at the moment so I thought this would apply here as well.
I have a few questions and comments on what you just said so I’ll just respond with inline quotes here.

> - Player population at time of searching…
> In Halo Wars 2 the system that chooses how to create a match is based on many different things which are not in this order of importance but include and are not limited to skill level of players being matched, the time players are searching and working to match, network statistics of the players searching, and many more elements in a complex algorithm.

Ok, this I understand.
Just for clarification, I was not complaining about the matchmaking system at all!
Referring to, for example, the TrueSkill paper from Microsoft, I’m pretty sure the matchmaking system itself has a higher complexity and is nothing we should argue about without knowing the internals.
But is the matchmaking system really that tied to the CSR adjustments?

> There are many more factors that go into it when a player matches and plays against another player and a very complex algorithm that is in place to calculate what a player will receive should they win or lose. An idea that has been floated which a lot of people are gravitating towards is that “players should get more points from a win.” However, in investigating the times where this happens - it should not be so in many cases.
>
> When you lose at high levels of play or to people who are of different ranks than you - you lose and gain rank accordingly. If an onyx player beats a gold which they should based on how placements and season records work - you don’t receive a lot towards your next rank. Hence a lot of very small increments of points. If an onyx player beats another onyx player - they will and do gain more points than the win over the gold player. The same is done in reverse where if an onyx player loses to a gold - you will lose a lot of rank in comparison to if you lose to another onyx player.

These two paragraphs really surprise me.
Are the CSR adjustments really that complex?
Referring to the second paragraph you’re saying that the CSR difference (Onyx vs Gold etc.) is a factor that determines the CSR adjustments?
I always thought that the MMR would be the primary driver for the adjustment computation because it’s a much more accurate metric for the player strength.
There are quite a few players that have a high MMR but a lower-than-usual CSR rank as they do not play very often (e.g. MMR > 3.5, CSR ~ 1600).
So using the CSR here instead of the MMR to compute the CSR adjustment just feels really wrong, but I’ll plot that later.

> Personally, I do not believe this is a problem with how the ranking system is working in itself based on the investigation that we have done and the conversations that we have had. I believe that the frustrations people are discussing is being perceived due to other things happening and how the matchmaking is choosing matches and when an unfortunate loss happens.

No, this is definitely a misperception!

The average players are not so much complaining about the matchmaking.
What frustrates the players primarily are the CSR adjustments, and there primarily those that come with the losses.
I and many others don’t mind playing against much better or much worse opponents.
I beat the worse opponents quite fast and the better players beat me quite fast, so that is not really terrible.
(Except for very active streamers like Nakamura who are bored to death)
Far too frustrating however, are these 15-30 CSR drops that you get from time when you lose against specific teams.
(And this is what you can find in all these smurf threads)
People complain about playing against teams that have a very unequal strength distribution which then leads to a very high CSR loss while all other games increase/decrease the CSR only slightly.
It happened often enough that me and team-mate play 10 2v2 games an evening with a plus of perhaps 10-15 CSR points just to lose almost all of the progress to a single team that contained a low rank, but good player.

So we’re not asking to touch the matchmaking system. This I really like to put emphasis on.
I’m questioning instead, whether the adjustments of the CSR can’t be tweaked a bit to prevent these sudden CSR drops against teams where the system is apparently not sure about the player strengths due to smurfs or inactivity.

9

I agree that the matchmaking system itself is quite good and only falters when the population in a playlist bottoms out. Additionally matchmaking and CSR adjustments in 1v1 is just much better in general. As an Onyx player I feel adequately punished for losing to a diamond player, and adequately rewarded for winning against a peer or superior opponent. The issue becomes team games. Last season, while my MMR was out of whack with my CSR placement (Placed gold 5 and got to diamond 3 with like a 90% W/L including DCs, placing me at Onyx 1658 this season), I was getting zero reward and extreme punishment. Win vs a team of platinums when my team is platinum and 1 diamond, 4-8% increase. Lose to a team of 1 plat, 1 diamond, 1 onyx… 50% decrease. Beat the SAME EXACT diamond and Onyx player with a different plat player… 15% increase. There seems to be far too much weighting on the lower CSR member of a team, actively punishing you out of proportion when losing to a team otherwise your equal or better.

10

> 2533274843669931;7:
> > 2533274792014876;6:
> > > 2533274933478052;2:
> > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
> >
> > …
>
> The only big complaint is I should NEVER have to play anyone below diamond if I’m in champion (unless they are in placements, then I can accept it). I would rather sit and wait then get a game that is just going to end in a win or if my xbox crashes/ game lags out (for whatever reason) then I lose a ton of csr normally 30+. Considering I get 0-1 csr from these games I would rather just not play them at all.

I think this represents most of the frustration with Matchmaking. Take this game:

https://www.halowaypoint.com/en-us/games/halo-wars-2/matches/f06c0f36-d855-437a-9aea-5c93b35ec4dd/players/zildjians?gameHistoryMatchIndex=1&gameHistoryGameModeFilter=All

The guy on our team was lvl 1 playing his 3rd game – being matched with 4 diamonds, and a gold (but with ~250 games played) in ranked. This game couldn’t of been fun for him, and certainty wasn’t for his two teammates. I don’t really care much about CSR, but it can be tough not to back out when you see the “lvl 1”…I always hope its a smurf…

11

> 2582868000586318;10:
> > 2533274843669931;7:
> > > 2533274792014876;6:
> > > > 2533274933478052;2:
> > > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
> > >
> > > …

Well, this is exactly not what we are talking about.
At the point where you searched a game the matchmaking system very likely just had no other choice to team you up with him.
As the community is not that large, these situations will continue to happen and cannot be prevented easily.

12

> 2533274933478052;11:
> > 2582868000586318;10:
> > > 2533274843669931;7:
> > > > 2533274792014876;6:
> > > > > 2533274933478052;2:
> > > > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
> > > >
> > > > …
>
> Well, this is exactly not what we are talking about.
> At the point where you searched a game the matchmaking system very likely just had no other choice to team you up with him.
> As the community is not that large, these situations will continue to happen and cannot be prevented easily.

the point for me (not sure if you are responding to me at all) is that I don’t care if there was no one of even diamond rank, me playing them is just always a bad result, either I get a bad quick uniteresting game or I lose a ton of csr cause of a lag out or crash, just tell me there aren’t any players or throw me vs some placements and test their skill. (that has at least some value)

13

> 2533274843669931;12:
> > 2533274933478052;11:
> > > 2582868000586318;10:
> > >
>
> the point for me (not sure if you are responding to me at all) is that I don’t care if there was no one of even diamond rank, me playing them is just always a bad result, either I get a bad quick uniteresting game or I lose a ton of csr cause of a lag out or crash, just tell me there aren’t any players or throw me vs some placements and test their skill. (that has at least some value)

Sure, that was what I understood and it is a good point.
I only referred to Zildjians response.

14

> 2533274933478052;11:
> > 2582868000586318;10:
> > > 2533274843669931;7:
> > > > 2533274792014876;6:
> > > > > 2533274933478052;2:
> > > > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
> > > >
> > > > …
>
> Well, this is exactly not what we are talking about.
> At the point where you searched a game the matchmaking system very likely just had no other choice to team you up with him.
> As the community is not that large, these situations will continue to happen and cannot be prevented easily.

This was 3v3X, the queue time is pretty short – the system can wait longer and get a bigger sample size.

In games like I showed holding them off can take 15-20 minutes. I would much rather spend that 15-20 minutes in the queue finding a better match. For the sake of the new-comer as well, it would also be more enjoyable for them to be placed closer to their level.

15

All I can say is, the matchmaking in Blitz is absolutely dreadful. Well over half my games, it groups me up with people who have no idea what they are doing or should be doing to win the match and its quite frustrating.

For anyone who says I should friend up or whatever, that’s not the issue. The issue is, matchmaking is suppose to group you up with people of your skill level or above and against similar others too. Not put random low level noobs or people who have low skill level on your team who have no idea what they are doing.

I’m quite sure Halo Wars 2 uses pretty much the same ranking system Gears of War 4 does, because they have the exact same problems everyone talks about and of course I’ve played it a lot.

16

I stopped playing rank when the disconnect and out of sync issues were never addressed properly. Until then you’ll never have an accurate ranked playlist. Thanks.

17

> 2533274792014876;6:
> > 2533274933478052;2:
> > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
>
> We have said that it would be difficult to do so, but at current time we are still investigating the system and are in no way promising that we will make changes to it at this time. As well, we have said that there are no plans to directly change how it functions at this time. We are looking into it.
>
> As an update on the investigation, we have spoken with the team that built the entire system and in this research it is doing what it is designed to do at this time. The complaints which I do here don’t boil down to just one thing being the problem. There are a lot of factors that must be taken into account and could potentially be doing what players believe is a negative impact on the player experience. Such things are as follows.
> - Player population at time of searching. - Smurf accounts that have an unfair win to loss ratio due to nefarious behavior (boosters). - A player simply losing to someone who the system believes they shouldn’t have based on previous win/loss records (possibly a smurf or simply because they got out played).There are many more factors that go into it when a player matches and plays against another player and a very complex algorithm that is in place to calculate what a player will receive should they win or lose. An idea that has been floated which a lot of people are gravitating towards is that “players should get more points from a win.” However, in investigating the times where this happens - it should not be so in many cases.
>
> When you lose at high levels of play or to people who are of different ranks than you - you lose and gain rank accordingly. If an onyx player beats a gold which they should based on how placements and season records work - you don’t receive a lot towards your next rank. Hence a lot of very small increments of points. If an onyx player beats another onyx player - they will and do gain more points than the win over the gold player. The same is done in reverse where if an onyx player loses to a gold - you will lose a lot of rank in comparison to if you lose to another onyx player.
>
> Personally, I do not believe this is a problem with how the ranking system is working in itself based on the investigation that we have done and the conversations that we have had. I believe that the frustrations people are discussing is being perceived due to other things happening and how the matchmaking is choosing matches and when an unfortunate loss happens.
>
> This all meaning that I believe this comes to a head when high level players lose to lower level players and end up losing a lot of rank as well as lower level players being matched against higher level players and having a bad matchmaking experience.
>
> In Halo Wars 2 the system that chooses how to create a match is based on many different things which are not in this order of importance but include and are not limited to skill level of players being matched, the time players are searching and working to match, network statistics of the players searching, and many more elements in a complex algorithm.
>
> There is no simple answer to these systems and how they interact, but know that we are investigating and looking into ways to better improve the experience for all players that are playing.
>
> Cheers,
> Postums

TL:DR looking at it not changing it.

18

> 2533274998443218;17:
> > 2533274792014876;6:
> > > 2533274933478052;2:
> > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
> >
> > We have said that it would be difficult to do so, but at current time we are still investigating the system and are in no way promising that we will make changes to it at this time. As well, we have said that there are no plans to directly change how it functions at this time. We are looking into it.
> >
> > As an update on the investigation, we have spoken with the team that built the entire system and in this research it is doing what it is designed to do at this time. The complaints which I do here don’t boil down to just one thing being the problem. There are a lot of factors that must be taken into account and could potentially be doing what players believe is a negative impact on the player experience. Such things are as follows.
> > - Player population at time of searching. - Smurf accounts that have an unfair win to loss ratio due to nefarious behavior (boosters). - A player simply losing to someone who the system believes they shouldn’t have based on previous win/loss records (possibly a smurf or simply because they got out played).There are many more factors that go into it when a player matches and plays against another player and a very complex algorithm that is in place to calculate what a player will receive should they win or lose. An idea that has been floated which a lot of people are gravitating towards is that “players should get more points from a win.” However, in investigating the times where this happens - it should not be so in many cases.
> >
> > When you lose at high levels of play or to people who are of different ranks than you - you lose and gain rank accordingly. If an onyx player beats a gold which they should based on how placements and season records work - you don’t receive a lot towards your next rank. Hence a lot of very small increments of points. If an onyx player beats another onyx player - they will and do gain more points than the win over the gold player. The same is done in reverse where if an onyx player loses to a gold - you will lose a lot of rank in comparison to if you lose to another onyx player.
> >
> > Personally, I do not believe this is a problem with how the ranking system is working in itself based on the investigation that we have done and the conversations that we have had. I believe that the frustrations people are discussing is being perceived due to other things happening and how the matchmaking is choosing matches and when an unfortunate loss happens.
> >
> > This all meaning that I believe this comes to a head when high level players lose to lower level players and end up losing a lot of rank as well as lower level players being matched against higher level players and having a bad matchmaking experience.
> >
> > In Halo Wars 2 the system that chooses how to create a match is based on many different things which are not in this order of importance but include and are not limited to skill level of players being matched, the time players are searching and working to match, network statistics of the players searching, and many more elements in a complex algorithm.
> >
> > There is no simple answer to these systems and how they interact, but know that we are investigating and looking into ways to better improve the experience for all players that are playing.
> >
> > Cheers,
> > Postums
>
> TL:DR looking at it not changing it.

Yes, we are currently looking into it. Nothing at this time is being changed because a better solution has not been found to resolve the various complaints players are having. I understand the frustrations, but it’s not as simple as many think. There are lots of moving parts that come into play that need to be taken into account.

19

> 2533274808527133;16:
> I stopped playing rank when the disconnect and out of sync issues were never addressed properly. Until then you’ll never have an accurate ranked playlist. Thanks.

Yeah these out of sync issues are ridiculous, the last two games of Blitz I played, I received the out of sync error. Weirdly enough, 90% of the time it happens, I’ve just had a great start. But yeah I switched my Xbox off and stopped playing when it happened today.

Why hasn’t this been addressed and fixed yet? I see posts from last year about this problem.

20

> 2533274792014876;18:
> > 2533274998443218;17:
> > > 2533274792014876;6:
> > > > 2533274933478052;2:
> > > > Postums and Wrensi emphasised quite a few times already, that they don’t have any resources to touch the code at the moment so this is probably not an option.
>
> Yes, we are currently looking into it. Nothing at this time is being changed because a better solution has not been found to resolve the various complaints players are having. I understand the frustrations, but it’s not as simple as many think. There are lots of moving parts that come into play that need to be taken into account.

Thanks for your response. I appreciate this is not a trivial problem. My specific problem comes from the losses incurred by facing up against teams that have unbalanced CSR and MMR. Though its unfortunate, I can live with small gains from crushing gold players - that’s just borne out of the matchmaking system eventually saying “well I can’t find anyone, so here’s some cannon fodder”. Its just when there’s team with a good player, with a team mate who’s a good but inactive player when the huge losses become very annoying.