23 replies to this topic

[Mawai]

Hi All,

This is more of an observation than anything else.

I like to play a wide variety of mechs .. heavy, light, medium ... rarely assaults... probably in that order. By design I have a different Elo in each weight class ... and it shows.

This is all PUG matches ...

In heavies, my matches are almost uniformly good. Folks seem to generally do the right thing, they typically play reasonably well ... group up ... listen when someone reports intel or suggests a push etc. I would guess that my Elo in heavies puts me in a good spot at the moment

Mediums seem to be mostly about the same ... I have been playing my HBK-4G with good success recently ... the matches are also generally good.

However, last night I pulled out my lights .. in particular my Jenner ... which has been one of my favourites since closed beta. There are two problems here ...

1) it always takes a couple of matches to adjust my playstyle to the fast twitchy response to do well in a skirmisher light.
2) I have so much fun with the speed in them that I am tempted into doing less than intelligent things ... often you will get away with these actions (and they can help your side win the game by distracting opponents at a crucial time) but they are also high risk and sometimes you don't survive it

Anyway, as a result, my Elo in lights appears to be less than heavies and mediums ... since there seemed to be a distinct difference in the matches I played last night in my Jenner compared to my usual matches. (Besides the fact that the Jenner seems to lose its CT far faster than I recall ).

- in one match my side literally went in 12 DIFFERENT directions. I could hardly believe it, as the entire group seemed to decide to go their own way ... some encouragement in chat to get folks to regroup worked more or less but it took almost two minutes! I think we still lost but at least we put up a decent fight instead of the 12:0 Roflstomp we were headed for ..

- there were builds in use that are uncommon in the usual games I play (e.g. Atlas DDC with ECM, LRM20 and 2ERLL that plinks at targets it can't really see from the backfield ... remarkably ineffective in general).

- a lot more peeking and far less spontaneous coordination ... it wasn't a matter of watching and saying to yourself ... "He is pushing so I will support him" ... it was more "Let him peek first and I will peek after so he gets blasted".

- Lone Dire Wolves ... (and not ones that just got left behind).

- etc

The short story is that the side that coordinated better usually won. One side or the other depending on which picked up the momentum would usually domino the other team. There was no "carry harder" ... I didn't see any signs of great players being tossed to the dogs with bad players ... individual skill wise the players seemed fairly comparable but winning was a random event based on luck or whichever side achieved minimal coordination.

How do you escape from here? ... Try hard to coordinate your side ... your individual play will have an impact on match outcome over the long term but stressing out about "carry harder" is a waste of time ... in the long run play better and play smarter so that your contribution will average out to more wins than losses. Avoid peeking, avoid getting stuck off on your own, avoid dying early, support your lance or choose a wingman, dont run off to do your own thing ... at least if you want to increase your chances of winning.

Focus fire destroys opponents ... this is not a 1:1 game and as much fun as it can be to play it that way ... you are increasing your chances of losing by playing that way ... which will extend your time in Elo hell

[MeiSooHaityu]

One big ELO problem is it treats all mechs of one class equally. No matter what Light, Medium, Heavy, or Assault you play, it all carries the same ELO.

So, you play your Storm Crow all the time, you establish this high Medium mech ELO, that's OK ad long as you stay in that Storm Crow.

BUT, now you feel like taking out a medium that you perform less in (say a jump sniper medium like a Vindicator). The Vindi has the same ELO rating as your Storm Crow (since they are both Mediums).

Now the matchmaker is throwing you in with a weaker team than usual thinking you're going to carry a bit more than your current Medium normally does. This is more likely to result in a stomp.

Since (for quirk purposes) mechs were organized into tiers, I think using a similar system should be integrated into matchmaker. Not only do you have an ELO rating for weight, it is broken down further by tier within that weight class.

Not a perfect solution, but it might help a bit.

[Nightmare1]

I hate it when I pick an un-Basic'd Light Mech so that I can level it, and the MM decides that I need to carry the team.

Lousy matches those...

[Duke Nedo]

I'd say step 1 is to separate Group Elo from Solo Elo, that would help a lot against these weird event matching...

Step 2 afaic could be yo merge all class Elo's into one Elo. While not perfect either, it would let you converge on your "destination Elo" quicker than when your Elo is fragmented between chassi and queues, making you play 4 or 8 times as many games before hitting your Elo... or some kind of spill-over mechanics where if you gain Elo in Heavies, then your Elo in Lights/Mediums/Assaults would get half that increase as well, and vice versa on a loss.

Otherwise I can just confirm the OP, I have quite different Elo's in different classes it feels like, Mediums being the lowest... PUG-bossing ftw if trying to affect Elo.

[Theodore42]

Mawai, on 01 June 2015 - 05:04 AM, said:

...

The short story is that the side that coordinated better usually won. One side or the other depending on which picked up the momentum would usually domino the other team. There was no "carry harder" ... I didn't see any signs of great players being tossed to the dogs with bad players ... individual skill wise the players seemed fairly comparable but winning was a random event based on luck or whichever side achieved minimal coordination.

How do you escape from here? ... Try hard to coordinate your side ... your individual play will have an impact on match outcome over the long term but stressing out about "carry harder" is a waste of time ... in the long run play better and play smarter so that your contribution will average out to more wins than losses. Avoid peeking, avoid getting stuck off on your own, avoid dying early, support your lance or choose a wingman, dont run off to do your own thing ... at least if you want to increase your chances of winning.

...

Yep.

Some drops I don't use viop at all, but sometimes I feel like I'm babysitting my team a little bit. If they are responding to what I'm saying, I keep saying it, and usually they end up in a better position and win.

I don't mind getting dropped in low Elo matches at all. Its usually and easy win because the other team isn't coordinated. Some people think that they have an Elo of 2500 and everyone else on thier team has an Elo of 500. Then the other team has a better variety of Elo scores so therefore they have more experienced players. I'm pretty sure this never happens to the extent that it wins matches.

They assume matchmaker matches 2500 Elo + 11(500 Elo) players vs 12(666 Elo) players. Really, one 2500 Elo player can't carry against 12 players with an Elo of 666 just because his team is all at an Elo of 500?

Obviously there would be variation in the other 11 players, but the difference between 11 players with an average Elo of 500 isn't that different from 11 players with an average Elo of 666. You should still be able to win.

Use your experience to win it for the newbies. You probably don't even have to fire a shot.

Or take the loss and enjoy a lower Elo and easier games. If you have a problem with carrying, this is probably what you want.

Probably the biggest source of imbalance in the Elo scores comes from bad data in the group queue. And I heard PGI is going to split the Group and Solo Elo scores, can anyone verify this?

[Weeny Machine]

I wonder what would happen if you turned off the matchmaker for one weekend. I bet you wouldn't even know the difference.

I have had so many games where my team either absolutely destroys the enemy or gets destroyed. And there is nothing you can do about it.

Heck, I just come from a match were we lost 4 mechs in 20 seconds - and the match had barely started. Even with the highest ELO possible no one could prevent this mess.

[sycocys]

I find many times to be somewhat the opposite. When I play standard I generally play the mechs I enjoy and have set up to my liking rather than using it for leveling mechs - its just easier to do that in CW with the passive xp.

If you are sitting somewhere dead in the middle of the elo system it will probably seem to be much more average across the board, with less low skill and less high skill players popping in. But if you are ranked higher, especially depending on the times you play (lower server population) you'll get some really, really, really unbalanced teams.

Carry harder is a good challenge though, so it doesn't really bother me much. More kills and damage for me if half the team wants to pretend they are snipers or actually think they are supporting the front line by firing LRMs from 1500m.

[MeiSooHaityu]

Bush Hopper, on 01 June 2015 - 05:45 AM, said:

I wonder what would happen if you turned off the matchmaker for one weekend. I bet you wouldn't even know the difference.

I have had so many games where my team either absolutely destroys the enemy or gets destroyed. And there is nothing you can do about it.

Heck, I just come from a match were we lost 4 mechs in 20 seconds - and the match had barely started. Even with the highest ELO possible no one could prevent this mess.

I honestly think it is "better" with the ELO in place, but not much better.

In theory it works, but it isn't flexible enough to compensate for different chassis performance (only weight class performance). Then there is the whole population issue that ends up throwing less than idea candidates into a match to fill the empty slots.

I don't think turning it off will be better. In closed beta, almost every match was a stomp. Matches like 8 to 0 or 8 to 1 was VERY common.

We still get a lot of stomps, but I feel it is marginally better with ELO. It just needs to be fine tuned better.

[Theodore42]

I win 12-0 or 12-3 way more often than I lose 0-12 or 3-12. What does that mean? I think it means when I'm on a "losing" team, I get them to 6 or 8 kills.

And when I'm on a winning team, I cause extra stomp. When my team is up 4 or so kills, I've been known to say "Nobody die, let's 12-0 these kids," and stuff like that. I do it to crush the confidence of the high Elo players on the other team so they don't try as hard to carry. If I can get everyone at a high Elo to think they're in Elo Hell, I'll always win.

Edited by Theodore42, 01 June 2015 - 06:08 AM.

[Navid A1]

The major problem of ELO is that it only takes weight class into account.
witch means that a gargoyle, a KGC and an awesome are all the same as far as elo goes. (BV anyone?)
the worse matter is that the mech skill level is not considered either ( you wanna level a new mech? ... yup...good luck carrying)

but we have to understand, MWO is very different from the general shooter concept.
on one (positive) hand it is more challenging. meaning that it gets harder to push back the more you give the enemy any opportunity. However, on the negative hand, it does not let you choose before you drop. so you end up with no ecm against a team full of lrm and 3 narc users. or you end up on alpine in a short range build.
ELO is not going to help much when a large group of non-JJ laservomits drop on terra therma.

Imagine a 1000 ton ball on top of a very steep hill. with 2 groups of people pushing it on opposite sides... no matter how elo balances the game based on skill. the nature of the situation dictates that the outcome will be decided in the first minute. even a slight movement of the ball to one side means its over for one of the teams.

Edited by Navid A1, 01 June 2015 - 06:09 AM.

[C E Dwyer]

Duke Nedo, on 01 June 2015 - 05:34 AM, said:

I'd say step 1 is to separate Group Elo from Solo Elo, that would help a lot against these weird event matching...

Step 2 afaic could be yo merge all class Elo's into one Elo. While not perfect either, it would let you converge on your "destination Elo" quicker than when your Elo is fragmented between chassi and queues, making you play 4 or 8 times as many games before hitting your Elo... or some kind of spill-over mechanics where if you gain Elo in Heavies, then your Elo in Lights/Mediums/Assaults would get half that increase as well, and vice versa on a loss.

Otherwise I can just confirm the OP, I have quite different Elo's in different classes it feels like, Mediums being the lowest... PUG-bossing ftw if trying to affect Elo.

Seperating group and solo ELO's will have virtually no effect.
Only effect it will have is if a bad player groups with ones that are a lot better, he will get into a higher elo bracket, without really learning the game and its problems, so that if and when he goes solo, he will get flattened and be a weak link, until his loss rate takes him into a lower elo.

[Mawai]

I agree with Mei ... I think the matchmaker would be much better if it factored in ..

- Elo
- mech tonnage
- mech loadout
- XP skill in the chassis

.. to come up with a "battle value" for the specific mech you are dropping with in a particular match and then used the battle values as an additional constraint.

As a first step, I would be interested to see some data mining to see the effect of mech XP, actual mech tonnage and mech loadouts on match outcomes. At the present time they just use Elo and weight class ... but it would be pretty straight forward to analyse the other factors to see whether or not they do a better or worse job of predicting match outcomes and thus likely being a better basis for forming even matches.

P.S. I am against a single value of Elo for all mechs though since I know from experience that I play each class of mechs quite differently ... I may bring the same situational awareness and aim to the game but not necessarily the same mindset or tactics ... so I think different Elo/chassis weight class is justifiable ... but it would be interesting to see the matchmaker include some of the other factors.

Edited by Mawai, 01 June 2015 - 06:15 AM.

[C E Dwyer]

Bush Hopper, on 01 June 2015 - 05:45 AM, said:

I wonder what would happen if you turned off the matchmaker for one weekend. I bet you wouldn't even know the difference.

I have had so many games where my team either absolutely destroys the enemy or gets destroyed. And there is nothing you can do about it.

Heck, I just come from a match were we lost 4 mechs in 20 seconds - and the match had barely started. Even with the highest ELO possible no one could prevent this mess.

MM is turned off in CW
Far more QQing about how unfair it is there

[CDLord HHGD]

I have to wonder if it places me with others of the same elo (if so, then I suck worse than I think I do) or if it tries to average the player base with both highs and lows (which is what I feel it does to me).

[Almond Brown]

Duke Nedo, on 01 June 2015 - 05:34 AM, said:

I'd say step 1 is to separate Group Elo from Solo Elo, that would help a lot against these weird event matching...

stuff

Can you show any "documentation" as to the existence of those "Group vs Solo" elo ratings please? I have not heard nor read of such a thing, thus separation of such, would seem to be impossible.

[Theodore42]

Almond Brown, on 01 June 2015 - 06:28 AM, said:

Can you show any "documentation" as to the existence of those "Group vs Solo" elo ratings please? I have not heard nor read of such a thing, thus separation of such, would seem to be impossible.

I'm so sure I've heard or read that PGI was looking at, or working on, separating group and solo queue Elos. Maybe it was mentioned in a Town Hall? I looked a few weeks ago and couldn't find any documentation, and just now I looked again real fast but didn't find anything. So if anyone finds something, link it here please!

[Duke Nedo]

Cathy, on 01 June 2015 - 06:08 AM, said:

Seperating group and solo ELO's will have virtually no effect.
Only effect it will have is if a bad player groups with ones that are a lot better, he will get into a higher elo bracket, without really learning the game and its problems, so that if and when he goes solo, he will get flattened and be a weak link, until his loss rate takes him into a lower elo.

It would change a lot imo, average players dropping regularly as 12-er in the group queue will build Elo that thay can't live up to in the solo queue. Then comes solo queue tournament....

[jaxjace]

Lights I havent been seriously into since they nerfed the jenners, but I have them all, and when I take them out, good lord they are fun. (seriously how does a myst lynx and a commando carry a team with 6 kills)

Mediums are a step up but tend to be a bit more hard pressed,

Heavies thankfully I have not played a ton of but I have a decent collection of them, all mastered out. Hit and miss, usually a hard fight or a roll one way or another.

Assaults for me are when I crack my knuckles, crack my neck, crack a mountain dew, turn my cap around, and get my game face on because the opposition will either be extremely stiff, or I will be leading an 11 man of trials.

[Bilbo]

Theodore42, on 01 June 2015 - 06:50 AM, said:

I'm so sure I've heard or read that PGI was looking at, or working on, separating group and solo queue Elos. Maybe it was mentioned in a Town Hall? I looked a few weeks ago and couldn't find any documentation, and just now I looked again real fast but didn't find anything. So if anyone finds something, link it here please!

Groups are given an modifier based on the size of the group. Don't know if it was implemented all or if it is still in. You'll have to dig through the Karl Berg thread. It is mentioned in there.

[Necromonger Commander]

Spent the last few days leveling some new mechs. I had forgotten how bad the underhive was....