#121
Posted 04 September 2014 - 09:17 AM
#122
Posted 04 September 2014 - 10:24 AM
Duke Nedo, on 04 September 2014 - 09:17 AM, said:
That's just it, most on here are to short-sighted to understand that it's not just one factor. They latch on to one thing and want it "fixed". The problem is, especially when dealing with something like the MM, no single thing is going to "fix" it. Just like 3/3/3/3 didn't "fix" anything except seeing 2 more lights and 2 more mediums on the field than usual. While that's an improvement it didn't touch the underlying issues with the MM and no singular fix for that ever will.
#123
Posted 04 September 2014 - 10:46 AM
#124
Posted 04 September 2014 - 10:50 AM
Kutfroat, on 04 September 2014 - 10:46 AM, said:
Join a Merc command Silliness like that happens way less often and usually only after much beer consumption!
#125
Posted 04 September 2014 - 10:51 AM
Kutfroat, on 04 September 2014 - 10:46 AM, said:
and then we get "gems" like this
Let me translate for everyone
I was in a game, we were winning! Then my team does something stupid and we lose so i'm going to jump on a bandwagon and blame PGI/MM
#126
Posted 04 September 2014 - 11:25 AM
Joseph Mallan, on 04 September 2014 - 10:50 AM, said:
s
Joseph Mallan, on 04 September 2014 - 10:50 AM, said:
sorry, i really had to lol ...I MYSELF had 3 beers at this point, that is the real sad part of this story i guess...or makes it even more sad...but don´t you go and use that 3 beers "agianst" me now, there was nothing i could do. they decided to "rush" them 1 by 1 from threre on...just like in a tower defense game. and i was in an xl long range jagermech on river city...
Sandpit, on 04 September 2014 - 10:51 AM, said:
Let me translate for everyone
I was in a game, we were winning! Then my team does something stupid and we lose so i'm going to jump on a bandwagon and blame PGI/MM
would be true if it was a "gem"...lately it´s quite common. don´t get me wrong, but it is more like...you tell people in chat, what and where the enemy is most likely to attack...it turns out the way you tell them is right...but they go " hey, **** that dude giving advices, i will just go in my 100 ton atlas and "rush" the entire team." game after game after game...don´t tell me this would´nt get you upset and thinking "hey, something is wrong".
Edited by Kutfroat, 04 September 2014 - 11:29 AM.
#127
Posted 04 September 2014 - 11:28 AM
Kutfroat, on 04 September 2014 - 11:25 AM, said:
s
sorry, i really had to lol ...I MYSELF had 3 beers at this point, that is the real sad part of this story i guess...or makes it even more sad...but don´t you go and use that 3 beers "agianst" me now, there was nothing i could do. they decided to "rush" them 1 by 1 from threre on...just like in a tower defense game. and i was in an xl long range jagermech on river city...
would be true if it was a "gem"...lately it´s quite common.
That is not a rush. That is a trickle. A proper rush would have ended the game in your favor.
Edited by Bilbo, 04 September 2014 - 11:29 AM.
#128
Posted 04 September 2014 - 11:43 AM
Sandpit, on 04 September 2014 - 10:51 AM, said:
Let me translate for everyone
I was in a game, we were winning! Then my team does something stupid and we lose so i'm going to jump on a bandwagon and blame PGI/MM
I suspect the poster forgot the /sacasm and /end sarcasm tags wrt blaming that on matchmaker.
That said, it's awesome when you get dug in and accepting a loss with a 1-6 disadvantage, and the other team pulls a brainfart and looses.
#130
Posted 04 September 2014 - 12:05 PM
Bilbo, on 04 September 2014 - 11:28 AM, said:
Bilbo, on 04 September 2014 - 11:28 AM, said:
that is why i said "rush" i mean...i really dont care that much about my stats, but...come on...when all of a sudden you have losing streaks of 5+ games and every win is/feels like hard work it get´s a little odd, or does´nt it? it just more and more feels like world of tanks, and the wot matchmaker was what made me quit.
Edited by Kutfroat, 04 September 2014 - 12:09 PM.
#131
Posted 04 September 2014 - 12:06 PM
no big deal, but I was running my 3AC2 mech, so, we're talking a steady, river, of flame and tracers......and they walked right in front.
one guy turned around, STILL in front of me and let me have an alpha strike. LOL
matchmaker is wonky
#133
Posted 04 September 2014 - 01:17 PM
Almeras, on 04 September 2014 - 01:39 AM, said:
until eventually you're carry more and more teammates. Its worth noting however your opponents may have a better arrogate ELO but chances are the same superstar+handicap set up (then good luck if your the uller and he's the madcat).
This is completely wrong, and not how elo is used anywhere, nor does ANY evidence convincingly point to this.
Almeras, on 04 September 2014 - 08:19 AM, said:
Nope.
Duke Nedo, on 04 September 2014 - 04:58 AM, said:
It is not how it works nor how it should ever work, it tires to match teams of a same elo while trying to minimize the variance (square distance from average) of each team member and then tries to match that team to another team with a close elo average.
People keep saying that it matches your with low skilled players because of confirmation bias (they see someone on their team doing bad so of course it must have matches them with a low elo teammate) and unwillingness to accept their own role in losses. The easiest way to see this is just look at League of Legends, even people in diamond+ ***** at their teammates calling them unskilled etc even though they are the same rank. Fact of the matter is even high elo players have bad games and I will straight up call anyone a lier if they say they never have bad games because one team always has to lose.
#134
Posted 04 September 2014 - 03:41 PM
StandingInFire, on 04 September 2014 - 01:17 PM, said:
People keep saying that it matches your with low skilled players because of confirmation bias (they see someone on their team doing bad so of course it must have matches them with a low elo teammate) and unwillingness to accept their own role in losses. The easiest way to see this is just look at League of Legends, even people in diamond+ ***** at their teammates calling them unskilled etc even though they are the same rank. Fact of the matter is even high elo players have bad games and I will straight up call anyone a lier if they say they never have bad games because one team always has to lose.
Granted that's some of them, but that's not entirely accurate either. I've seen it happen. Whether it's the release valve mechanics or something screwy with the MM system in general, new(ish) players are dropping into matches out of their league with more frequency than to be considered an outlier or anomaly
#135
Posted 05 September 2014 - 01:33 AM
This happens a lot and doesn't make sense! And there seems to be a systematic aspect to it because loss streaks are just too common...
#136
Posted 05 September 2014 - 03:02 AM
Kutfroat, on 04 September 2014 - 11:25 AM, said:
Edited by Joseph Mallan, 05 September 2014 - 03:02 AM.
#137
Posted 05 September 2014 - 06:13 AM
Sandpit, on 04 September 2014 - 03:41 PM, said:
Duke Nedo, on 05 September 2014 - 01:33 AM, said:
As I outlined in this post
StandingInFire, on 03 September 2014 - 05:35 PM, said:
that happens not because of a matchmaker problem but bad elo adjustment problem where win/loss streaks can catapult people up and down ranks while potentially taking a very long time to climb/fall back to their appropriate elos (if you are accurately predicted to win/lose your elo doesn't change much).
For example a high elo player on a 5 game loss streak can lose up to 150 elo and similarly a low elo player on win streak of 5 games can gain up to 150 elo. If then they are also on the high/low ends on each team you can get an effective elo difference of ~400 elo, with the matchmaker favoring the team of the high elo player if the low elo players team then looses he won't loose much elo and the high elo player won't gain much elo meaning they can be matched against each other again neither having fun games.
Edited by StandingInFire, 05 September 2014 - 06:15 AM.
#138
Posted 05 September 2014 - 07:19 AM
StandingInFire, on 05 September 2014 - 06:13 AM, said:
For example a high elo player on a 5 game loss streak can lose up to 150 elo and similarly a low elo player on win streak of 5 games can gain up to 150 elo. If then they are also on the high/low ends on each team you can get an effective elo difference of ~400 elo, with the matchmaker favoring the team of the high elo player if the low elo players team then looses he won't loose much elo and the high elo player won't gain much elo meaning they can be matched against each other again neither having fun games.
I can't understand how you would get streaks because of ELO unless the MM is broken as well.
For example, assume that ELO is completely inaccurate, then that will be true for all players and all we get is random match-making = no abnormal streaks.
Assume that only a select few gets inflated ELO for some reason, if the MM was working these persons would end up with good teammates and good opponenst = only 1/12 players would be underperforming relative to elo = minimal effect on streaks.
You would have to combine this with something like the MM pairing you up with several low elo players to compensate (the handicap theory some suggested). Perhaps you could reach the same effects with a really small playerbase and a REALLY inflated ELO, but in that case the waiting times should be long, and they really aren't as long as you dont play the most popular class.
Perhaps abnormal ELO + that the MM tries to force 3-3-3-3 too hard relative to the ELO acceptance... or something like that. I have no clue. I just can't get it to make sense without dead hamsters in the MM...
#139
Posted 05 September 2014 - 07:37 AM
my novas collectively are about a .7 w/l ratio wich is way under my averages. my average including novas of mechs ive been leveling since the clans have come out is .94 . My historic averages are 1.1 wich includes 7k+ games going back to my newb days of getting pugstomped.
example, my nova S, 6 ermlas, 4mg, ams with ams range and ams overload, UAV/coolshot and seismic.
.45 w/l ratio, 1.25 k/d ratio 313 damage per game.
Edited by Fuggles, 05 September 2014 - 07:41 AM.
#140
Posted 06 September 2014 - 09:06 AM
Duke Nedo, on 05 September 2014 - 07:19 AM, said:
Even a perfect MM with an infinite player base would not prevent streaking, liken a match to a coin flip it has a 50/50 of landing heads or tails (similar to a perfect match having a win/loss chance) but every now and then you will get a string of heads or tails the ELO will after a few matches result in the % being weighted more and more in favor against the low elo player on a win streak as he in essence would be "weighing down" the rest of his team or conversely for a high elo player on a loss streak. But even so its not that hard to go win on 50%, 49%, 47%, 45%, 40% for a 5 win streak.
Duke Nedo, on 05 September 2014 - 07:19 AM, said:
Having no elo would mean your win loss chance is completely random, if anything it would be more prone to streaking as elo wont eventually kick to put you back to the correct place. If you played before the MM/elo was introduced you would know that by enlarge matches have been a lot closer since they where introduced, 12-0 stomps where very common back then.
Duke Nedo, on 05 September 2014 - 07:19 AM, said:
You would have to combine this with something like the MM pairing you up with several low elo players to compensate (the handicap theory some suggested). Perhaps you could reach the same effects with a really small playerbase and a REALLY inflated ELO, but in that case the waiting times should be long, and they really aren't as long as you dont play the most popular class.
There is no reason the inflated/deflated elo example I gave needs to be limited to just 1 player on each side, it is technically possible it could apply to every single player in a match (just statistically unlikely). Coupled with the slow elo adjustment on accurate prediction it would take a long time for the inflation/deflation effect to be removed from a players previous streak and can result in a reverse streak of nearly no elo adjustment. This is fixed by adding the elo clamping mechanism.
About the reverse streak after inflation say in the previous example of the low elo player, he is playing at ~150 elo above where he should be, he then looses the match but was predicted to loose, his elo would barely change say to +145 elo, this could continue for many matches before he falls back into his correct elo. Meaning without clamping (where you restrict the max elo change over x number of games to a certain % or number) players going on streak can completely break their elo for a long time (20+ matches).
Duke Nedo, on 05 September 2014 - 07:19 AM, said:
I hope I explained it better with the points above.
4 user(s) are reading this topic
0 members, 4 guests, 0 anonymous users