Elo Is Amazing.
#21
Posted 24 August 2014 - 10:58 AM
#22
Posted 24 August 2014 - 11:06 AM
MadTulip, on 24 August 2014 - 10:29 AM, said:
Actually no ... his ELO might not have changed at all. If the matchmaker randomly puts the player on the side that is predicted to lose every time then by losing every time he would have simply proved the matchmaker correct and his ELO would not have changed.
#23
Posted 24 August 2014 - 11:10 AM
Tor6, on 24 August 2014 - 10:58 AM, said:
Wow .. absolutely the opposite opinion from me.
Stomps in my game play were WAY WAY more common before the current iteration of the match maker. Also, as far as I know the matchmaker tries to get everyone in a small band of ELO initially and doesn't widen the search criteria until sometime later. This keeps the average ELO on both sides close to the same and the variance small.
They should start putting up the difference in ELO between two teams and the variance/range in ELO on each team so that folks can see that ELO doesn't cause stomps ... players do
#24
Posted 24 August 2014 - 12:40 PM
#25
Posted 24 August 2014 - 01:32 PM
What's your overall win/loss record - just curious?
#26
Posted 24 August 2014 - 01:47 PM
Tor6, on 24 August 2014 - 10:58 AM, said:
actually world of tanks has the same crappy matchmaker, and that was what made me quit. because you know that after a few good games, there will be games that you are meant to lose to achieve that bull.sh.it 50% winrate. games dont become more challangeing, they just become stomps that are decided by matchmaking before the match. that was the feeling in wot, and it gets similar more and more in this game. elo is for sologames (1 vs 1)...especially when only rated on winrate and no other factors, biggest fail ever. turn off elo, Balance by 3/3/3/3 - and i mean force it, make people wait if they only want to play heavy/assault, and maybe balance factors like ecm...i believe more close games will happen (just by randomnesss) than now.
Edited by Kutfroat, 24 August 2014 - 01:52 PM.
#27
Posted 24 August 2014 - 01:52 PM
Iron Frost, on 24 August 2014 - 01:32 PM, said:
I am also wondering that, my win ratio is stable at a hair over 1 and i never get bad streak loss ever. I think i can remember once having a streak loss of 6, 2 will happen often but then win streak of 2 happen too. PGI has to come clean about exactly how elo works. I can most always determine when the other team flat out played better or when we just made bad decision or even had bad luck, most of the time saying it's because your team is bad is overly simplistic and stupid. And that comes from a guy who like to write snarky sarcastic comment to his team every once in a while.
Kutfroat, on 24 August 2014 - 01:47 PM, said:
The matchmaker will put you versus players of the same caliber as you which is exactly why you don't win them all, how hard is that to understand?
Edited by DAYLEET, 24 August 2014 - 01:55 PM.
#28
Posted 24 August 2014 - 01:59 PM
#29
Posted 24 August 2014 - 02:09 PM
#30
Posted 24 August 2014 - 02:11 PM
I quit, you can keep your prizes. Before anyone says, "it's you", I'm doing the absolute best I can, I'm being as tryhardy as I can possibly get. There's just better tryhards out there right now.
Edited by Techorse, 24 August 2014 - 02:14 PM.
#31
Posted 24 August 2014 - 02:21 PM
The average is useful for a number of things, but balancing player ability is not among its uses. Due to the way the average is calculated, it is inherently vulnerable to outliers, which drag the end result down or up with them. In the context of this game, it can and does result in skilled/experienced players dropping with people who literally just downloaded the game being placed on the same team and same match. Such should only happen if skilled players choose to group up with unskilled players for some odd reason, or if they're playing in a custom match.
The median might be a better statistic for rating a team's elo, because it isn't nearly as weak to outliers. One or two really high or really low values won't drag the whole team up or down if a median was used instead of an average.
#32
Posted 24 August 2014 - 02:23 PM
Techorse, on 24 August 2014 - 02:11 PM, said:
I quit, you can keep your prizes. Before anyone says, "it's you", I'm doing the absolute best I can, I'm being as tryhardy as I can possibly get. There's just better tryhards out there right now.
I ran into that tryhardy moment last night, when started seeing names from the Mech Specs forums I knew my evening was over, was like seeing The Wild Hunt claim everyone on my team, only problem was none of us could run, hide, or join.
#35
Posted 24 August 2014 - 04:00 PM
#36
Posted 24 August 2014 - 04:09 PM
Tell them I said "hi" when you get there.
#37
Posted 24 August 2014 - 04:11 PM
#38
Posted 24 August 2014 - 04:53 PM
Edited by Dibster, 24 August 2014 - 04:53 PM.
#39
Posted 24 August 2014 - 05:07 PM
Team
Solo
Group
Loadouts
Poor play
discos
suiciders
etc.
That's just a very quick and short list of things that contribute to losses. TO blame any one singular thing is usually going to be a major over simplification.
#40
Posted 24 August 2014 - 05:15 PM
S13gtastic, on 24 August 2014 - 02:09 PM, said:
I am confused here. The only difference between Elo and a ranking system is that with the PGI Elo system you don't know what your rank is. How do you create "an honest to god ladder system" in a team game? Team work generally counts for more than damage output. A team of 12 will always defeat 12 solo heroes no matter how good they are ... add equivalent amounts of team work to both sides and the one with the better players will usually win. So how do you place players on a ladder when all there is to be evaluated are team outcomes?
You could try ranking on kills/assists and total damage ... then the highest ranked players could be those who have bad aim but get off a lot of shots while hiding behind the rest of their team mates ... they survive the match, get lots of kills and assists, do immense amounts of damage ... and may be useless team players. I'm not suggesting that the highest scoring players actually play that way ... but it is a playstyle that would lead to a high rank if performance is measured solely on damage/kills/assists.
On the other hand, the system PGI is applying calculates whether your team might be expected to win or lose based on the overall Elo score calculated for the team. If they meet expectations (win or lose) nothing changes ... if one side wins when expected to lose the Elo rank of the players on the winning side goes up while the losing side goes down. Over enough matches, the players individual contribution to winning or losing whether that is through doing damage, organizing the team, providing intel and motivation or whatever contribution that player makes to the team outcome ... gets included in the Elo score ... and hopefully results in more even matches.
29 user(s) are reading this topic
0 members, 29 guests, 0 anonymous users