New Matchmaking System

Everything dealing with the video game developed by Cyanide!
User avatar
TheSpydyr
Posts: 1226
Joined: 02 April 2012, 16:03

Re: New Matchmaking System

Postby TheSpydyr » 21 March 2016, 16:47

Also, 21 wins doesn't matter, 21 wins MORE than loses is what matters. Just incase you didn't miswrite :)
It was mis-typed and was meant to be a 21 win differential.
Spydyr

Image
The RNG is not broken, you are not "the unluckiest coach in the world" and the computer does not cheat, you just suck.

Twitch
YouTube

User avatar
Zelash
Posts: 8
Joined: 22 November 2015, 19:38
Location: Quebec City, Canada

Re: New Matchmaking System

Postby Zelash » 21 March 2016, 17:00

Great! Going to make a Chaos team, never pick up the ball. I'll lose first 20 games then I'll start wining from just pitch clears. Hurray for TV++ :twisted: In all seriousness, can we least have a hard cap, like ±400 from actual TV.

User avatar
SquirrelDude
Posts: 431
Joined: 18 September 2015, 00:30

Re: New Matchmaking System

Postby SquirrelDude » 21 March 2016, 17:06

So will inducements be based on team value or the TV+zSum?
The current system takes only TV into account for inducement, but we are discussing about taking into account TV++.
Will the 2 minutes only fail to matchup a coach if there is an odd number of coaches or will it also fail if there is an absurd TV++/TV difference?
It will fail if there is an odd number of coaches and if the TV difference exceed the TV limit set in the system.
Will there be any delays between activation periods of the pool? Will coaches need to reactivate for each period?
Probably not. No need to reactivate.
Sounds good.
Will coaches be able to activate multiple team to find a single match simultaneously?
I don't really understand the question. You can still look for a match with several teams in the same time.
I never did it because you have to babysit the matchmaker so much. Hopefully it will be a feature I take advantage of more going forward.
Makes sense.
Will coaches know how many other coaches are in the pool when they activate? Will they be able to see what teams or coaches are in the activation pool?
No.
That's a tad disappointing. Means coaches won't know if they're walking into a buzzsaw. Is there a minimum number of entered coaches required for the matchmaker create games?

Will private leagues have the option of using team value to match teams instead of TV++?
I'll ask to the devs. I don't think you can choose which system apply for each league.
Disappointing. Will every league's open matchmaking move to using 2 minute pools as well?
When will this update be implemented?
With the Norse team, in a few weeks.
A few weeks? I thought Norse would be here at the end of March, after we thought it would be mid march, after we thought it would be February. Now you're saying they'll be here in the middle of April?

How will concessions affect TV++/zSum?
A concession is a loss in the current system.
Thanks Netheos. Can the resulting numbers be negative (as my quick test seemed to show that they can)? Is this correct that you can potentially have someone with 7,000 and someone with -3,000? (although not matched against each other)
Yes, the R*(W-L) can be negative.
This seems like a poisonous combination. Pixelhugging murder chaos/orc/and dwarf teams already concede whenever they might have to play bash and don't get to go first. Now they're going to be rewarded for that noxious behavior.

Will the team be introducing the 51 SPP rule by which players with 51 or more SPP have a 50% chance to quit the team if a coach concedes to try and limit that behavior now that reconnections are possible?

licker34
Posts: 163
Joined: 09 March 2016, 17:40

Re: New Matchmaking System

Postby licker34 » 21 March 2016, 17:14

It would be nice if the team management/activation system were reworked somehow to make it easier.

Even just being able to sort (by league) or reorder your teams would be helpful.

Better still would be a real list that you could just select the teams you want to activate on.

Not huge really, but the system we have feels very clunky.

Also, concessions really cannot give you TV++ advantage, that's really broken.

Sephiroth
Posts: 77
Joined: 16 February 2016, 11:15

Re: New Matchmaking System

Postby Sephiroth » 21 March 2016, 17:24

Concessions should give you +50 TV++ the same as a win, this way you'll face even harder games and be discouraged to concess. A non-conceder coach wouldn't be affected by this if he only concedes rarely, a conceder though would be punished. It sounds crazy but seems better than the opposite that will probably be implemented.

User avatar
Mercy Flush
Posts: 562
Joined: 08 October 2011, 02:21

Re: New Matchmaking System

Postby Mercy Flush » 21 March 2016, 17:34

Concessions should give you +50 TV++ the same as a win, this way you'll face even harder games and be discouraged to concess. A non-conceder coach wouldn't be affected by this if he only concedes rarely, a conceder though would be punished. It sounds crazy but seems better than the opposite that will probably be implemented.
I'm surprised this hasn't been suggested before. Nice idea.
Image

Dreamerbg
Posts: 12
Joined: 02 March 2016, 13:33

Re: New Matchmaking System

Postby Dreamerbg » 21 March 2016, 18:00

Pls tell me again why are you changing the MM system ? cuz this new one sounds worse atm

User avatar
VoodooMike
Posts: 1614
Joined: 14 July 2009, 07:44
Location: Calgary, Alberta
Contact:

Re: New Matchmaking System

Postby VoodooMike » 21 March 2016, 18:11

There is a maximum TV difference applied to the matchmaking formula (2000). Two teams with a TV difference higher than this cannot be paired.
If you're using TV-only for inducements then 2000 is far too large a difference. The efficacy of inducements under TV matching hits rock bottom at around 500 difference, where the TV underdog mean win chance drops to around 35%. The most effective long-term strategy will remain to play a team that is easy to develop and maintain... specifically bash (and likely Chaos, until CDs and Nurgle come around to give bash a tiny bit of variety).
2 minutes? Not long enough. There are, on average, 3 matches played every 5 minutes, so 6 teams are paired. 2 minutes will work at peak times (EU evenings) when there are up to 9 matches every 5 minutes, but outside that it'll make little difference.
Even 5 minutes is fairly short, though I understand that mainstream gamers expect fast games... something tells me they aren't going to be happy with fast games that give them massive TV differences. That's ok, they'll just concede, right?
Means that you are considering implementing TV++ as just a pairings generator but NOT use it for actual inducement differential? Doesn't that basically defeat the entire purpose of it?
Sure does. Looks like we'll be waiting another few years before we see who is right about real TVPlus environments.
This system used for pairing looks good. I do really hope it will not be used for the inducements, because It will generate weird things like killer teams who do not care about winning rewarded with even more inducements.
An excellent continuation of your not getting it.. Without using TVPlus for inducements there's no real reason to bother... and including concessions as losses means that well-built kill teams have a way to be matched up with anyone without having to manage their TV to find victims. They've never cared about inducements because inducements don't bridge the gap in TV difference.
I think the 2000 number is crazy high for either TV or TV++, but it's just a starting point, so we will see how it works I guess.
It's crazy-high in any environment with normal inducement caps as it means the underdog basically has to buy every single inducement possible, and then some star players, to even be able to spend the inducement gold they get at the highest TV differences. We both know that entire pile of crap won't do much against a team that is 2000 TV higher than you - your best bet will be to declare you plan to lightning bolt their best player and hope they concede like 60% of everyone does.
The limit would be far less of an issue if the 2 minutes was 5....
You have data... what is the 5-minute TV difference distribution like across the times of day? If you build distributions based on different widths of time you should be able to figure out the needed pooling time. Really, a system could be built to base pooling time on exactly that data gathered from the last 2-5 poolings such that the window of time shifts to compensate for the time of day and availability of active teams.. but just giving them a number is far more likely to net results.
This seems like a poisonous combination. Pixelhugging murder chaos/orc/and dwarf teams already concede whenever they might have to play bash and don't get to go first. Now they're going to be rewarded for that noxious behavior.
It is a toxic combination in conjunction with a full-attrition environment. It gives teams built to mangle other teams a second method for preying on new, weak teams besides just minmaxing. I don't know how this escapes the people making the decisions at Cyanide.
Concessions should give you +50 TV++ the same as a win, this way you'll face even harder games and be discouraged to concess.
That'll just end up as a positive feedback loop - it won't make people concede less. Besides, if concession is counted as a win, and 60% of matches result in concession, why would we assume that higher ratings would mean harder teams? It'd just mean the ratings have no meaning (which, really, they already won't without TVPlus rating being used for inducements) and that the more you play a team the less likely you are to ever get a complete game.
Friendly Reminder: Correlation does not equal Causation - tattoo it on the inside of your eyelids if it'll help.

twitch/the_sage_bb
Posts: 616
Joined: 17 December 2015, 08:06
Facebook: https://www.facebook.com/theSagebb/
Twitter: https://twitter.com/the_Sage_BB

Re: New Matchmaking System

Postby twitch/the_sage_bb » 21 March 2016, 18:48

I like the suggestion of having concessions count as a win, giving conceders harder matches (or having them face eachother). I'd also like a hard cap of 50% of the lowest TV team (so 500 for a rookie, but a TV 2000 might be up vs 3000). I also think 5 mins would be better than 2, and that the use of TVplus should be optional for leagues.
Content: Twitch / Youtube ; Updates: Facebook / Twitter
(because people with big banners are just compensating)

User avatar
dode74
Posts: 7041
Joined: 11 December 2008, 11:18
Location: Nr. Reading, UK
Contact:

Re: New Matchmaking System

Postby dode74 » 21 March 2016, 19:16

You have data... what is the 5-minute TV difference distribution like across the times of day? If you build distributions based on different widths of time you should be able to figure out the needed pooling time. Really, a system could be built to base pooling time on exactly that data gathered from the last 2-5 poolings such that the window of time shifts to compensate for the time of day and availability of active teams.. but just giving them a number is far more likely to net results.
Absolutely, and I did suggest that pooling times increase at times of low activity. Simplicity of implementation was, I suspect, at least one reason. A good reason for not having variable pool times, though, is expectation management. I suggested a countdown clock until each pool so people knew how long they would be waiting when they activated and weren't sat there watching the screen - they could alt-tab out if there were a 10 second audio alert or flashing taskbar icon. Without that clock predictability is, I think, pretty important.
Image


Return to “General”

Who is online

Users browsing this forum: Trmbdunalt and 3 guests

cron