Concerning Matchmaking and Queues

[Rojan]

As it stands right now, custom matches seem to be the preferable method of finding matches. I would like to present an alternative to this in hopes that streamlining a matchmaking process might make it easier (and more enjoyable overall) to find a match. The three things that I am hoping to improve upon are as follows:
i. Shorter queue/matchmaking times
ii. Better matching of opponents of similar skill
iii. Keeps a unified playerbase

The way in which I would like to achieve these three goals is by essentially building a matchmaking system with a hidden ELO rating that will be used to match players into a random match. They could select the game mode, their deck, and the size of the match (1v1 all the way to 10v10). More or less it would be the ranked matchmaker with a hidden ELO (so people don't get ranked anxiety or whatever). This implementation would keep the playerbase together with no more matchmaker vs. custom game split. It would make games actually fair by using the hidden ELO to match players of similar skill. Lastly, it would shorten queue times by having everyone using the matchmaker. I obviously understand that people would still like to play custom matches to test things, play inhouses, and for other things of similar intent. There could still be custom matches but simply done through an invite system.

I'd like to see if anyone here had some ideas to build upon this idea and also I wanted to get some feed back in the form of posts and through this strawpoll that I put up: http://www.strawpoll.me/13063202

Thank you in advance for taking the time to answer the poll and give some feedback regarding my proposal.

 

[Sarin]

Matchmaking will live or die off how fast it finds matches.
Simply, if it takes too long to find a match people will not use it.

I am not disagreeing with anything you wrote above...
In fact. Ideally I would PREFER matchmaking as the only way to find a game, and for skill/stat matchup to happen. That is what I WOULD WANT.

HOWEVER

If the matchmaking takes too long. It simply doesn't work. Everything has to be based around how long it takes to get a game.

Currently you have players split into a ton of different things.
10v10, Conquest Lobbies, Destruction Lobbies, Ranked, QM 1v1, 2v2, 3v3 and 4v4.
SD doesn't have the most enormous of populations as it is, so each of these 'pools' is not very large. Combined with the fact that matches can take up to 45 mins and people fiddling with decks and such. Means the amount of people looking for specific game type at any time isn't very large.

So if you want to play 3v3, and over a 10 minute time period only 5 people are searching. When that 6th person joins the game is going to start.
It isn't going to have time available to do any ELO checks. It doesn't have enough people to skill balance. It just goes (crap its been 10 mins. GO)

If you wanted to play 3v3 and over a 10 minute time period you had 40 players searching.
Well thats a different ball game. You can have an ELO system figuring out based between ELO scores currently available and TIME spent currently waiting on how to balance and when to start games.



This makes it easy to see that matchmaking just needs a larger pool, and the smaller the pools. The worse it is.
Taking an already small pool and trying to split it up into even smaller ELO pools just makes it slower, and kills matchmaking.
Where as making the pool larger, allows ELO to happen.


Pushing everyone into Matchmaking will increase the pools size.
However half the problem is how many pools SD matchmaking has.

They could select the game mode, their deck, and the size of the match (1v1 all the way to 10v10).

This is so many pools of people looking for games. So many options. This is the problem.

I really think that if an ELO system, or skill balancing is going to happen, with matchmaking in this system. The amount of options you can join has to drastically decrease.
Such as scrapping 10v10 except for custom invites. Scrapping Destruction, and even limiting what you want to join to 2 options.

1v1 Ranked or Group Combat (which would be 2v2, 3v3, 4v4)

The ELO system will then put people together. So you might end up in a 2v2 match, you might end up in a 4v4 match.

I just don't see how matchmaking will work with so many different queue options.
I don't want to see 10v10, Destruction or anything else scrapped. I would feel bad for those players.
Just personally, I don't see matchmaking being successful and having enough players without it being put in two large pools for the ELO to pick games from.
Matchmaking speed is what is important, no one wants to wait 20+ minutes for a game.

 

[Rojan]

This is so many pools of people looking for games. So many options. This is the problem.

I really think that if an ELO system, or skill balancing is going to happen, with matchmaking in this system. The amount of options you can join has to drastically decrease.
Such as scrapping 10v10 except for custom invites. Scrapping Destruction, and even limiting what you want to join to 2 options.

1v1 Ranked or Group Combat (which would be 2v2, 3v3, 4v4)

The ELO system will then put people together. So you might end up in a 2v2 match, you might end up in a 4v4 match.

I just don't see how matchmaking will work with so many different queue options.
I don't want to see 10v10, Destruction or anything else scrapped. I would feel bad for those players.
Just personally, I don't see matchmaking being successful and having enough players without it being put in two large pools for the ELO to pick games from.
Matchmaking speed is what is important, no one wants to wait 20+ minutes for a game.

I agree with you are saying entirely. I think we need to cut pools. It is why I advocated for turning custom lobbies into invite only. I think once that gets cut down there will be a boost in matchmaking though, don't you think? ELO pools don't need to be restrictive either. It can be a funnel that widens from restrictive to around the same skill, to broad search, and then finally to no restriction. I don't think ELO matchmaking would work with a hard restriction on it.

 

[Sarin]

I think once that gets cut down there will be a boost in matchmaking though, don't you think?

There will be a bump I am sure. Still too many pools of players though, each game size split into 2 by Conquest and Destruction, and 10v10 being a large waiting time. That not only causes waiting in its own mode, but would have a disproportionate delaying effect on the smaller sided games.


1v1 ranked, 2v2 Conquest, 3v3 Conquest, 4v4 Conquest, 2v2 Destruction, 3v3 Destruction, 4v4 Destruction, 10v10 Conquest, 10v10 Destruction.
9 different matchmaking options.

Now think about the 10v10 Destruction and 10v10 Conquest waiting players. Say over a 10 minute time you had 8 for 10D and 7 for 10C waiting.
If you were to take those 15 players and throw them into smaller sided games. That is going to have a big impact on matchmaking speed.

Now scrap Destruction and suddenly the Conquest pools are a bit bigger. Scrap individual game sizes down to 1v1 and Group, where the ELO decides how many get into a game.
Suddenly matchmaking is flying.



Simply making Lobby harder to find isn't going to give the matchmaking we are talking about.
It will give a bump to numbers in the pools, but the problem is. Too many pools.

 

[IS-2]

Well IMO if they just copy paste CoH 2's matchmaker it will work excellently. You can't select between different game rules, it is all standardized. What you can do is select if you want to queue for 1v1, 2v2 .etc, and allows you to queue for any combination of player count either solo or with friends. It also has map vetos to let you turn off a few maps you don't want to play. AFAIK the matchmaker tries to match you with those close in skill, but as time goes on the matchmaking process will widen the criteria to an extent, allowing you to get a game if there is nobody of exact same skill online. In addition to this it has an ELO system that is quite subtle and so won't scare away the carebear crowd and allows fair matches. Custom battles in CoH 2 were barely played and this did not kill the carebear community, in fact they all got their little play ground to battle in known as any rank below 10.

 

[Rojan]

There will be a bump I am sure. Still too many pools of players though, each game size split into 2 by Conquest and Destruction, and 10v10 being a large waiting time. That not only causes waiting in its own mode, but would have a disproportionate delaying effect on the smaller sided games.


1v1 ranked, 2v2 Conquest, 3v3 Conquest, 4v4 Conquest, 2v2 Destruction, 3v3 Destruction, 4v4 Destruction, 10v10 Conquest, 10v10 Destruction.
9 different matchmaking options.

Now think about the 10v10 Destruction and 10v10 Conquest waiting players. Say over a 10 minute time you had 8 for 10D and 7 for 10C waiting.
If you were to take those 15 players and throw them into smaller sided games. That is going to have a big impact on matchmaking speed.

Now scrap Destruction and suddenly the Conquest pools are a bit bigger. Scrap individual game sizes down to 1v1 and Group, where the ELO decides how many get into a game.
Suddenly matchmaking is flying.



Simply making Lobby harder to find isn't going to give the matchmaking we are talking about.
It will give a bump to numbers in the pools, but the problem is. Too many pools.

How do you cut pools without pissing everyone off? Destruction and 10v10 seem to be the huge player sink. I guess just keep the 10v10 lobby and then let players make invite only lobbies while making matchmaking 1v1-4v4 Conquest.

 

[Sarin]

Well IMO if they just copy paste CoH 2's matchmaker it will work excellently.

Problem is COH 2 had a much larger player base.
Not only that, but as you already pointed out, they had less options, fewer pools to put those players in.

COH peak players at launch 20,800. Average about 10k.
WRD peak players at launch 4,800. Average about 2k
SD44 peak players at launch 4,606. Average about 1k (still in its first month)

So the difference in the amount of people using matckmaking is VERY different.
SD44 is taking a much smaller player base and spreading into a much larger amount of pools.

 

[IS-2]

Problem is COH 2 had a much larger player base.
Not only that, but as you already pointed out, they had less options, fewer pools to put those players in.

COH peak players at launch 20,800. Average about 10k.
WRD peak players at launch 4,800. Average about 2k
SD44 peak players at launch 4,606. Average about 1k (still in its first month)

So the difference in the amount of people using matckmaking is VERY different.
SD44 is taking a much smaller player base and spreading into a much larger amount of pools.

CoH 2 averages about 4-5k players these days. I see what you're saying though, but if matchmaking is taking too long it can be adjusted to have more lax requirements for acquiring a game.

 

[Sarin]

How do you cut pools without pissing everyone off? Destruction and 10v10 seem to be the huge player sink. I guess just keep the 10v10 lobby and then let players make invite only lobbies while making matchmaking 1v1-4v4 Conquest.

You don't. That is the problem.

That is I believe why we currently have the system we currently have.

SD44 either needs an incredibly large jump in sales so that every gametype pool gets a big rise in population.
Or it needs to drastically cut down on how many different gametype pools people are spread into.

Everyone keeps pointing at COH2 and saying (see matchmaking works). But it was at least 5x the amount of people spread into far fewer gametypes.

It is either worry about matchmaking and scrap 10v10 and Destruction.
Or keep 10v10 and Destruction but don't worry about matchmaking.

Even now, at launch when the most people that are going to play the game are around and try matchmaking.
There simply isn't enough people for it to be working at any sort of speed let alone skill matching.

I think they will just continue as is, and more people will move away from matchmaking because it takes too long.
The more people leave it because it takes too long, the longer it takes, so the more people leave it, and so on.

 

[Sarin]

CoH 2 averages about 4-5k players these days. I see what you're saying though, but if matchmaking is taking too long it can be adjusted to have more lax requirements for acquiring a game.

Yea I worded it a little awkwardly. Those were peak numbers I quoted. I just meant the AVERAGE of the PEAK over time.
COH2 was around 9k but jumping up to 16k at times. It currently peaks around 6.8 k.

Its AVERAGE users is about 5k per month the last year.
Red Dragon was between 700 and 1000.
SD44 you cannot tell yet really. Currently says 1050 but hard to tell with first month.

Roughly the pop of COH2 is about 5x the size of SD44 it looks like.

So if COH2 was splitting say 5000 people into say (5) game modes for example.
That would be 1000 people split into each pool. (not every pool would actually be equal)

IF SD44 has 1000 people and is splitting them into (9) game modes.
That is 111 per pool on average (not every pool would be equal)

So you have 1/10th the amount of people in an SD44 Matchmaking pool compared to a COH2 matchmaking pool and so on.
Starts to show why its slower.

 

[Rojan]

Maybe it is blind optimism but I really do think that if you limited destruction to lobby only, closed lobby to invite only, left the 10v10 as is, and pushed everyone else into a single matchmaker it would seriously improve the quality and ease of finding a match.

 

[1776ZOOMSNIPE1911]

As it stands right now, custom matches seem to be the preferable method of finding matches. I would like to present an alternative to this in hopes that streamlining a matchmaking process might make it easier (and more enjoyable overall) to find a match. The three things that I am hoping to improve upon are as follows:
i. Shorter queue/matchmaking times
ii. Better matching of opponents of similar skill
iii. Keeps a unified playerbase

The way in which I would like to achieve these three goals is by essentially building a matchmaking system with a hidden ELO rating that will be used to match players into a random match. They could select the game mode, their deck, and the size of the match (1v1 all the way to 10v10). More or less it would be the ranked matchmaker with a hidden ELO (so people don't get ranked anxiety or whatever). This implementation would keep the playerbase together with no more matchmaker vs. custom game split. It would make games actually fair by using the hidden ELO to match players of similar skill. Lastly, it would shorten queue times by having everyone using the matchmaker. I obviously understand that people would still like to play custom matches to test things, play inhouses, and for other things of similar intent. There could still be custom matches but simply done through an invite system.

I'd like to see if anyone here had some ideas to build upon this idea and also I wanted to get some feed back in the form of posts and through this strawpoll that I put up: http://www.strawpoll.me/13063202

Thank you in advance for taking the time to answer the poll and give some feedback regarding my proposal.

I like your ideas!

I would like to see rewards for the top 100 players in every mode, 1v1 --> 4v4. Aces skins for example

 

[Stoffen]

To be fair there is also an element of balance discussion in this whole automatch debate.
I see a lot of lobbies where the host (+ teammates) has chosen both division and the map for synergy.
For instance, 3rd Falls is amazing in close quater maps but struggles more on open maps.

Having a random map pool with limited veto options would really present a more balanced game.
Even if its only a "Random" option from the mappool dropdown menu.

 

[Rojan]

To be fair there is also an element of balance discussion in this whole automatch debate.
I see a lot of lobbies where the host (+ teammates) has chosen both division and the map for synergy.
For instance, 3rd Falls is amazing in close quater maps but struggles more on open maps.

Having a random map pool with limited veto options would really present a more balanced game.
Even if its only a "Random" option from the mappool dropdown menu.

I really enjoy the random aspect. It makes me think of new ways to fight on maps I may not be familiar with.

 

[sam hamwitch]

How do you cut pools without pissing everyone off? Destruction and 10v10 seem to be the huge player sink. I guess just keep the 10v10 lobby and then let players make invite only lobbies while making matchmaking 1v1-4v4 Conquest.

Why not just make it so that rather than making the various different options (1v1 vs 2v2 and 3v3, destruction vs conquest etc) either/ or choices, you just have a checkbox system where you can choose to queue for either (for example) destruction, conquest, or both. You might need a preference system so you can pick a first or second choice, but the idea is that players wouldn't be isolating themselves by queuing for unpopular options, since if there aren't enough players online for those options it'll stick them in with something there are players for instead. Then you avoid the situation where nobody queues for 4v4 quickplay because nobody plays 4v4 quickplay, because nobody queues for 4v4 quickplay because nobody plays 4v4 quick play etc etc.

 

[Max_Damage]

I havent played custom game for like a week. Matchmaking is a go to solution to have any reasonable games for me.

The only thing i would like to add is randomize side selection at the beginning of ranked game. I dont believe axis and allies are balanced and will be balanced.

 

[Rojan]

I havent played custom game for like a week. Matchmaking is a go to solution to have any reasonable games for me.

The only thing i would like to add is randomize side selection at the beginning of ranked game. I dont believe axis and allies are balanced and will be balanced.

It is my go to as well. It's just... the time can take a while.

 

[Nalydix]

It has probably been already suggested above but IMO :

-Merge Quickmatch and Ranked into a single "Matchmaking".
-Match people with the closest ELO against each others, how much ELO you earn/lose depend on the difference between yours and his.
-Restrict the Custom lobby to invite only and for 10vs10.
-Allow players to choose between Destruction or Conquest in the merged Matchmaking.
-Allow players to select multiple queue at the same time (both Axis/allies) ; (1vs1/2vs2/3vs3/4vs4).
-Allow players to select their deck after knowing on which map they are going to play.

Obviously, it's going to piss off pro lobby players. But this has to be done if you want the Matchmaking to work, which is clearly the main system the devs want us to use from the many changes they did during the early access.

This game just doesn't have enough players to afford to divide its players base across several game finding mechanics.

Depending on how many players actually play your game, the priority should be this : Lobby system -> Matchmaking -> Ranked + Quickmatch -> Anything else