27 replies to this topic

[N0V0CAIN]

Roadbeer, on 29 March 2013 - 02:49 PM, said:

I swear to the MODs if I see one more Elo thread, I'm going to punch a frackin kitten.

http://mwomercs.com/...a-elo-is-great/

[Taemien]

Leiska, on 29 March 2013 - 02:41 PM, said:

My brand new LRM boat Stalker-3H has the following stats:

This is why.

You're using a weight class you're not normally using, which means your ELO for that weight class is low. And losing in it is just dropping it further. You have four ELO scores, one for each weight class.

There are two reasons for the losses. One is being unlucky. Lower ELO players usually don't win by a slim margin, but rather in a landslide. This is due to the other team being lower skilled and inexperienced too. What normally happens is someone gets impatient and draws enough fire to die. Another does this, probably trying to help the first. You're down 6v8 and the other team realizes its advantage and presses forward and wipes the rest of you out. This is how alot of CB games played out. reason I say unlucky is because it was someone on your team that did it each time.

The second reason is you might have a premade on the other team. What happens here, and I'm going to use your screenie as an example and assume the DC peeps are grouped. You and 12 other solo peeps join the queue as the same time as a premade group of three. The queuer slots the premade (I believe groups get priority so they don't get 'failed to find game' and its easier to slot solos after) and then fills in the rest. You got unlucky and didn't get the group on your team. Even in a grand melee, the group can focus fire much easier and thats why you get smoked.

ELO does work, but when there is only a limited number of people queuing at a time, it gets wonky. I mean say a 4 man premade of TOP ELO scores queues up and only 12 solo BOTTOM ELO people queue up at the same time. Well the match maker goes, "yep 16 players, 8 here and 8 there, launch!" Remember if there is a thousand players online, only 50-100 are not in a game, and out of those only a small percentage are hitting launch at the same time.

For ELO to work 100%, we need hundreds to a thousand queuing up at the same time. That won't happen in MWO. Those of you that played other MMOs with group queues know that even with millions of players, those games fail to find groups in minutes.

To get the best results, play at primetimes. Non-primetime you might as well not even assume ELO is on, because effectively it isn't. It can't be, there's not enough players for it.

Edited by Taemien, 31 March 2013 - 11:34 AM.

[Gremlich Johns]

Sable Dove, on 29 March 2013 - 02:50 PM, said:

The LRMs are not weak; at least not large amounts of them. Only ECM prevents LRM boats from being overpowered.

The problem with ELO is that it barely attempts balance; it just rounds up 12-16 players, throws them on teams, and picks which one is likely going to win. If you PUG most of the time, ELO does nothing for you, because you can't rely on your teammates; they might be great, they might suck, but if you're consistently expected to lose, and you lose, then your ELO doesn't go down, and it just keeps putting you in games where it expects you to lose.

Really, unless you're with a team, your win/loss is entirely random, as the game doesn't really try to balance; it just predicts who will win or lose.

Elo matches team members by skill and does not seem to also balance the teams based on their tonnage.

[FunkyFritter]

Leiska, on 29 March 2013 - 02:41 PM, said:

Now, as you can see from the stats, even with the very weak LRMs, I'm worth more than 1.5 average mechs, yet I lose 3/4 matches.

This is an excellent example of how not to use KDR. Killing the bad players on the enemy team while the good ones get into position isn't helping your team very much.

[Cato Phoenix]

I think its biggest effect is probably averaging out the groups against the better-than-average pugs.

[Taemien]

Just understand by the basis in which ELO works, it needs more players queuing up to make it work better.

No amount of coding to the ELO will make it make low skilled players more skilled.

[Roadbeer]

ELO @ WORK

Edited by Roadbeer, 31 March 2013 - 12:18 PM.

[MischiefSC]

19 matches?

Come back when it's 190.

You're in an LRM boat, it's a feast/famine mech. It's also not a good choice for pugging.

You're doing a bunch of damage but without killing people. That's the issue with support mechs. Hop in a 2xAC20 Jag. You'll be getting 5 kills for 400 damage and it'll help your team. LRMs only help your team when you're focus firing or you're using your LRMs to drive high-risk members of the enemy team into cover (splatcats, Boomjags, etc) into cover while your teammates gang up on and kill the more vulnerable and now isolated enemies.

Finally, Elo is for the whole match. Not just you. It's not going to only drop bad people with you but good people on the other team. The other team will be made up of the same caliber of players as your team.

No offense is intended here and all due props for trying to run an LRM Stalker build, it's a tough ride to deliver in but when done well can really help your team. However if you're consistently having issues then you need to look at the common denominator. that's going to be you or your mech. Maybe it's how you play, maybe it's how you drive, but your team is going to be the same as the other team. At worst you should end up at a 1.0 win/loss. If you're losing more than you're winning it means that you are negatively impacting your teams success.

that's the math of it. It takes a good 50-100 games to play out statistically though, we all have bad runs. I've had an almost 20 game losing streak before. It was painful. I think I had like 1 or 2 kills. I had to stop playing for the day, I felt like I'd been beaten with a stick for an hour. Came back the next day and did fine again.

Happens. Don't get caught up in it. Change things up if it's not working for you. I never could get Stalkers to work for me, just doesn't suit my playstyle. LRM or brawler. Find what works and do that.

Finally