Login

russian armor

Let's fix ELO automatch

12 Nov 2013, 17:07 PM
#21
avatar of Lichtbringer

Posts: 476

I think canceling the search is a great Idea.
Why should I play 3 Games with 5 min searchtime, when 40from 60 min are wasted because its a stomp?
I am saving myself and other people from unbalanced games and therefore time.
If there is someone in my skillrange in the firstplace this method wont increase waiting times.

On the other hand I enjoy the challenge and my first attempt to solve the problem is to become the best player, so that I will never be matched against someone who is better :D
12 Nov 2013, 17:10 PM
#22
avatar of herr anfsim

Posts: 247

Thanks for the tip. Il do that untill they get arround including basic public lobbies.
12 Nov 2013, 17:26 PM
#23
avatar of Alties

Posts: 49

Aborting is basically the same thing as the checkbox idea, except you have to do it manually instead of it being automatic. You won't get matched with someone on your skill level if there isn't one around. So you'll have to wait who knows how long, until someone who is, logs on.

Perhaps if you get matched up with someone on your level, the game was really good, add them to your friends list and organize 1v1s with them? Rinse and repeat with how many other people you want.

That way you take take the random into your hands and you know what you're getting yourself into.
12 Nov 2013, 17:47 PM
#24
avatar of WilliG

Posts: 157

In Soviet Russia ELO games you!
12 Nov 2013, 22:05 PM
#25
avatar of SgtBulldog

Posts: 688

There definately IS a problem. I just won a 2v2 and it boosted my position 90.000 up (opposition was absolutely newbie).

It shouldn't change my rating that much on one victory after 75 games!?!?
13 Nov 2013, 00:09 AM
#26
avatar of wooof

Posts: 950 | Subs: 1

thats how elo works. i dont know your rank, but if youre somewhere near the middle of the pack, where there are a lot of players, most of them have a very small difference in rating between them. any small change in rating will make you go up in ranks very quickly. just wait til you get closer to the top and it takes 20+ wins in a row to gain a rank.
13 Nov 2013, 07:06 AM
#27
avatar of SgtBulldog

Posts: 688

jump backJump back to quoted post13 Nov 2013, 00:09 AMwooof
thats how elo works. i dont know your rank, but if youre somewhere near the middle of the pack, where there are a lot of players, most of them have a very small difference in rating between them. any small change in rating will make you go up in ranks very quickly. just wait til you get closer to the top and it takes 20+ wins in a row to gain a rank.


But that's the whole point: I'll never get anywhere near the top this way.

The system should match me against players 'somewhat' better when I'm on the way up. But it does not. In 1v1 it matches me against vastly superior players OR total newbies. There's very rarely something in between.

In 2v2 I usually either get a strong teammate against 2 newbies or vice versa. That makes my rank bounce up and down many thousand positions.

Does it not strike you as odd that there would be something like 90.000 players exactly in the middle group? Surely the distribution should be more even?

Could it simply be, that a lot of unused profiles simply take up a hughe chunk at .5 win ratio and the default ELO? If that's how it works, that is faulty. Unused profiles should start at the bottom ranks with no rating.
13 Nov 2013, 07:13 AM
#28
avatar of DanielD

Posts: 783 | Subs: 3



I think with something like an RTS there is actually an even steeper drop-off as skill increases. So if you're one standard deviation away from the mean, but just barely, you have very few options in terms of opponents, and most of those options will crush you. If there were 1,000 people searching, then you'd have good results, of course.
13 Nov 2013, 09:51 AM
#29
avatar of SgtBulldog

Posts: 688

jump backJump back to quoted post13 Nov 2013, 07:13 AMDanielD

I think with something like an RTS there is actually an even steeper drop-off as skill increases. So if you're one standard deviation away from the mean, but just barely, you have very few options in terms of opponents, and most of those options will crush you. If there were 1,000 people searching, then you'd have good results, of course.


That's a probabel distribution, allthough it may be much narrower in COH2 of today.

Still the same, I suspect that little or completely unused profiles make up a large part of the middle. But these players are never available for matching and in that way messes up matching with smaller increments. It sort of teleports you directly from one side of the distribution to the other.

It's speculation, ofc., but if I'm right, those idle profiles should be left out from the calculation.
1 user is browsing this thread: 1 guest

Ladders Top 10

  • #
    Steam Alias
    W
    L
    %
    Streak
Data provided by Relic Relic Entertainment

Replay highlight

VS
  • U.S. Forces flag cblanco ★
  • The British Forces flag 보드카 중대
  • Oberkommando West flag VonManteuffel
  • Ostheer flag Heartless Jäger
uploaded by XXxxHeartlessxxXX

Board Info

398 users are online: 398 guests
0 post in the last 24h
0 post in the last week
28 posts in the last month
Registered members: 49432
Welcome our newest member, weekprophecy
Most online: 2043 users on 29 Oct 2023, 01:04 AM