PDA

View Full Version : Hows matchmaking gonna work exactly?


violence.inc
03-09-2012, 03:23 PM
I ask because i think this could be crucial to getting new players to stick with the game and also potentially get more people to play 'competitive' games (scrimage best of 30 rds 5v5 or 6v6).

Will you be able to choose your game type (hostage rescue, defuse)? And then it finds a game for you?

I think if you made competitive match a game option more people would take a liking to the competitive side. If you added that option for a scrim match (call it something more coddish for the kiddies idk hardcore defusal) more people would learn to understand this is the main way the game was intended to be played. It would give the game more structure instead of playing for hours till the server time is up and the scores 256 rounds to 164.
Even when youre lookin at a list of servers today what do you see? hordes of 32+ player servers new players will see this and say 'is this what the games about? a massive deathmatch?' I think most non competitive players dont play scrim because maybe they dont even know the game type exists.

Valves trying to get the cod audience right? Or younger players? This would be a great way without ruining the game. If youve ever played a cod i have (guilty but only at my friends) youll see how you can choose your game type and then it finds a match. And really a scrimage game isnt a far cry from cod's search and destroy (except cs is actually good).

But im rambling i got that coffee in me and got pumped up and carried away sorry.:eek:

SWGEmu-cRush
03-09-2012, 03:29 PM
Let's hope it works better than League of Legends match making. Simply basing it on W/L ratio isn't good enough.

Squipe
03-09-2012, 03:32 PM
I think Valve should take some notes about League of Legend's matchmaking system. Whether or not they choose to use a different algorithm to determine a player's "skill" is a different topic. When discussing interface, LoL nailed it.

Ranked / Classic:

Maps should be random. It rewards players who are good at the game, not just good at de_dust2. Teams will win based off of experience, teamwork, and skill.

They should set some customizable parameters, being that you want to play bomb defusal only, hostage only, all of the above, etc.

5-man Premade

A group of friends can create a "clan" of 5-10 players. There should be a specific queue for just players who are a 5-man premade. In other words, it skips the entire process of IRC and just queues you up with a team who is already prepared and roughly your skill level.

Solo-queue (or duo-queue)

Queueing ranked by yourself. You get matchmade with 4 random players and you face another PUG of 5 players. This will give you the opportunity to play competitively if your entire team is not on. It will also allow you to be exposed to good players who you might want to recruit into your clan. Remember how hard it was to recruit players of your skill level in the old games? This should alleviate some time.

Casual Matchmaking:

Arsenal Matchmaking

Queues you with however many players you have in an aresnal game of 5v5.

Gungame

Queues you with however many players you have in a gungame round of 5v5.

Casual Classic

Queues you with however many players you have in a classic game of 5v5.

--------------------------------

I believe that anything else should be custom servers only. Maybe you can add a Big Team Battle queue like Halo 2 did in giant 15 vs 15 matchmakings. That could be a good time. Other than that, keep it 5v5 and quickplay. Anything else should be found in the server browser.

inputpow
03-09-2012, 03:42 PM
Let's hope it works better than League of Legends match making. Simply basing it on W/L ratio isn't good enough.

League match making is the best that is about for a team based game. CS could pretty much steal the whole algorithm.

The fact is solo rating will never be accurate, only teams can have a accurate rating, it is impossible to determine who got carried, but CS is a lot less role based so it should be even more accurate.

CS should always be played in 5s, even if you don't know the people, would be cool if they could have a "mix" rating of some sort for when you just play with random people you met in the past and it will try to match you as best as possible.

Any way its so much better than the current, ulow, low, lowmid, mid, midhigh, high, high+ system players try to attribute to them selves, so many mixes end up in either domination or sacrifice because people will lie about their skill level.

broadwey_again
03-09-2012, 03:45 PM
It matches you with players with similar stats and I beleive you can choose casual/competitive.

SpYd3r
03-09-2012, 03:56 PM
In the gg_lake video he said they're using the ELO system, but didn't say if that applies to individual players or teams or what.

SWGEmu-cRush
03-09-2012, 03:56 PM
An issue I readily have in LoL is solo queuing. I will end up getting matched up with 3-4 completely clueless teammates, one of which ends up going AFK for IP, while playing against 5 good players, with at least 3-4 of them knowing what they are doing.

This happens consistently for me during LoL solo queuing. It has driven me to the point where I won't queue up for normal games without 2-3 buddies.

I think the problem though is that LoL based the matchups on W:L ratio (or so I've been told), rather than a true skill algorithm.

Skeith
03-09-2012, 04:02 PM
ELO is probably the best way to rate a player because it's the only thing that truly matters. Any "skill" algorithm can be gamed to make people appear better than what they are.

Squipe
03-09-2012, 04:03 PM
I think the problem though is that LoL based the matchups on W:L ratio (or so I've been told), rather than a true skill algorithm.

It doesn't though. League of Legends has an Elo system named after the creator Arpad Elo when designing to pit two players of equal skill level in Chess.

In League of Legends ranked play, you get 10 placement games to put you in the ballpark of where the game thinks you are. The next 20 games will net you a large amount of Elo per win/loss (~30), with further intention of accurately placing you into your correct skill group.

Afterwards, when you win a game of league, you gain Elo. The amount of Elo you gained is dependent on your Elo relative to your opponent's average Elo. In other words, if you were first pick and beat a bunch of players 500 Elo under you, you will gain 9 Elo. If you were a player 500 Elo below the enemy average, you will gain around 13. Vice versa, you will lose 13 Elo if you lost to people "worse than you".

Normals:

You still have this Elo system, but it's hidden now. You used to be able to access the files on your computer to tell your hidden Elo, but they patched it up. Knowing your casual Elo is counter-intuitive to what casual is, right?

Instead of your 30 placement games, you choose your approximate skill level if you remember when you first create your account: beginner to RTS, average gamer, hardcore gamer, seasoned RTS veteran, etc. Your hidden elo is gained and lost like ranked.

You can, however, see your Dominion hidden Elo. It sounds sketchy, but send a friend request to JabeBot. You will be automatically accepted. Type "!pstats summoner" without quotes, with summoner being your in-game name. It will show you your ratios, hidden Dominion Elo, and Elo history in Season 1.

duki
03-09-2012, 04:05 PM
The major flaw in most modern match making systems is the inability to de-rank (i.e., rank is directly proportional to time played). SC2 has an amazing ranking system - they could potentially learn a lot from that.