24 replies to this topic

[Goose]

I like to think I can get mileage out of a light 'Mech, and back during, and after, Top Gun, I tended to see names off of them charts here and there. Ignoring the Usual Number of Cravens in-game, things weren't too bad.

Then 19MAR happened, and I sat out 'till the hot fix; Then Dragons went on sale, and I basically went to level them up during the Heavy Challenge. You can guess how well that went …

In fact, I found out the hard way I'm no good in heavies: Seems I'm the Team Banzai ideal of ether light or assaults.

Durring this time, I sat in on a lot of bad pugs; I'd heard Elo was by weight class, so i figured my results for effort equaled my teams results for whatever it was they thought they were doing.

So now I'm back to in my lights, tring to level up my Spiders, finally.

I can think of a handful of games that went well, or were fun. The wins tended to be me capping out with no teammates left. Most games are, at best, a 8-2 blow-out.

About like all that time I just spent in heavies …

So it seems like I've trashed my Elo, and there's only one Elo per person, and it's so far down, I ether have to carry my pugs (and I just ain't that good), or play way more with a team to "fix" it. I've got "in's" with teams, fine, but still: Seems like two or seven things are broken with matchmaking …

[Davers]

You have 4 different Elo scores, one for each weight class. Your Dragon performance has nothing to do with your light mech Elo. The problem is you are playing Spiders.

[Goose]

And yet the quality of the teams has drastically changed from when I played lights earlier this month, to more closely resembling my recent time in heavies …

[Flyby215]

Haha, I was wondering the same thing.

I suspect it has to do with the long weekend and lots of players being online; under-elo and premades alike are online en-masse. I'm on a wicked 5 win 27 loss record today so far; even my very best mechs are being roflstomped. A few matches it's certainly been me (picked my weaponless Jager by accident!) but most matches I'm getting sub-100 teammates vs 300+ enemy premades.

Don't be upset though. It's like the pre-elo days again, which must of us PUGgers are used to anyway!

[Zylo]

Goose, on 31 March 2013 - 02:55 PM, said:

And yet the quality of the teams has drastically changed from when I played lights earlier this month, to more closely resembling my recent time in heavies …

Elo ratings seem to allow mixing of different ratings as long as the team average matches the enemy team. A member of my merc corp that is pretty good and wins often ended up with a player on the team that claimed to be new, was in a trial mech and performed like a new player. I think when Elo scores get high enough it's expected that a good player can balance a new or bad player but this doesn't seem to be the case which seems to increase the odds of 8-0 wins or losses. It's also not fair to new players to put them into matches where they aren't facing only new players with similar Elo rankings.

[Zylo]

Flyby215, on 31 March 2013 - 02:55 PM, said:

Haha, I was wondering the same thing.

I suspect it has to do with the long weekend and lots of players being online; under-elo and premades alike are online en-masse. I'm on a wicked 5 win 27 loss record today so far; even my very best mechs are being roflstomped. A few matches it's certainly been me (picked my weaponless Jager by accident!) but most matches I'm getting sub-100 teammates vs 300+ enemy premades.

Don't be upset though. It's like the pre-elo days again, which must of us PUGgers are used to anyway!

A member of my group believes he found an exploit some groups may be using to greatly increase the odds of 4+4 sync drops working and has reported this (and I think the players involved). It's possible you may be seeing the effects of sync-drops again. Phase 3 matchmaker didn't seem stop sync-drops. A quick fix would be limiting each team to only a single group.

Edited by Zylo, 31 March 2013 - 03:07 PM.

[Inconspicuous]

I believe it fills one team first (can be pretty random) and then calculates an average and tries to fill the other team using that data.

This would lead to everyone on your team NOT being at the same Elo rating.

I can not prove this of course.

[Davers]

Zylo, on 31 March 2013 - 03:05 PM, said:

A member of my group believes he found an exploit some groups may be using to greatly increase the odds of 4+4 sync drops working and has reported this (and I think the players involved). It's possible you may be seeing the effects of sync-drops again. Phase 3 matchmaker didn't seem stop sync-drops. A quick fix would be limiting each team to only a single group.

According to the latest Ask The Devs, most players of MWO are in groups, and there is pretty much ALWAYS a group in each drop. Some groups communicate more and are more obvious, others don't and get confused with Pugs.

You should always anticipate the other team being 2 4 man groups.

[Training Instructor]

Ran into a lot of really coordinated teams this weekend, while I didn't seem to benefit from this at all on my side mostly.

[Goose]

Zylo, on 31 March 2013 - 03:02 PM, said:

[insert both posts here]

See: The mechanics of how a game is formed, and the odds of it being "meaningful" to Elo (i.e. has a chance of moving the score(s) around) haven't ever been put out, AFAICT; And it clearly goes off the rails a lot, trying to start the game sooner rather then later.

The guys I wind up play with as a team often get a number of players other than 4 or 8, so they tend to try sync dropping. "Try" being the operative phrase …

[Zylo]

Davers, on 31 March 2013 - 03:10 PM, said:

According to the latest Ask The Devs, most players of MWO are in groups, and there is pretty much ALWAYS a group in each drop. Some groups communicate more and are more obvious, others don't and get confused with Pugs.

You should always anticipate the other team being 2 4 man groups.

Right, that's part of the problem currently as the exploit that was reported increased the odds of the groups ending up in the same game. Players are using this to sync-drop and it was observed that the same sync-drop groups were only showing up on a single map. Matchmaker probably places at least a single group against a sync-drop group at minimum, as long as there is at least 1 group on each side.

[Dreamslave]

Through my own playtime, I'm a firm believer that ELo never existed in the first place.

[Zylo]

Goose, on 31 March 2013 - 03:15 PM, said:

See: The mechanics of how a game is formed, and the odds of it being "meaningful" to Elo (i.e. has a chance of moving the score(s) around) haven't ever been put out, AFAICT; And it clearly goes off the rails a lot, trying to start the game sooner rather then later.

The guys I wind up play with as a team often get a number of players other than 4 or 8, so they tend to try sync dropping. "Try" being the operative phrase …

This is why PGI needs to return to any group size and just match similar size groups together first, then match weights second and try to get the Elo ratings as close as possible as 3rd priority.

[Goose]

I'd bet matching group size would be counter productive, time-wise, and would like to see how a match of four pairs faces off with a seven-plus-a-pug.

Maybe the rule would be "within one is close enough." A seven-man could expect to face a eight- or a six-man …

Weight matching: Even with all of the nova-hot light drivers out there, whom could kill off a Stalker on their own, there are other victory conditions out there.

I do think there should be more Elos per person: Something like two per 'Mech ("in-group" and "out") to sort things out. Probably screw up the sample-size per Elo, but we should be playing more, anyways.

[Flyby215]

Zylo, on 31 March 2013 - 03:17 PM, said:

Right, that's part of the problem currently as the exploit that was reported increased the odds of the groups ending up in the same game. Players are using this to sync-drop and it was observed that the same sync-drop groups were only showing up on a single map. Matchmaker probably places at least a single group against a sync-drop group at minimum, as long as there is at least 1 group on each side.

I find this kind of interesting. I would wonder if this was deliberately implemented to satiate the die-hard groups while not saying anything so that non-serious kiddie-groups can't terrorize the masses with 4x4 DDC drops? (sorry if that sounded corny, but it's a legitimate quandary).

Although... I certainly can't criticize people who sync-drop since I too am guilty (although personally, I like killing my friends every bit as teaming with them ["killing" them in game!])

Goose, on 31 March 2013 - 03:44 PM, said:

I'd bet matching group size would be counter productive, time-wise, and would like to see how a match of four pairs faces off with a seven-plus-a-pug.

Maybe the rule would be "within one is close enough." A seven-man could expect to face a eight- or a six-man …

Weight matching: Even with all of the nova-hot light drivers out there, whom could kill off a Stalker on their own, there are other victory conditions out there.

I do think there should be more Elos per person: Something like two per 'Mech ("in-group" and "out") to sort things out. Probably screw up the sample-size per Elo, but we should be playing more, anyways.

Wish I could "Like" this more than once. Stats too as "group" and "non-group" would be worthwhile.

Play more? Eh-heh... heh...heh.... 4,000 matches good?

[Caleb Lee]

Dreamslave, on 31 March 2013 - 03:17 PM, said:

Through my own playtime, I'm a firm believer that ELo never existed in the first place.

ELO exists... my win loss and kill death have drastically been dropping. The level of team I get paired up with is typically horrible, run off and do their own thing, have no situational awareness and bring noob builds.

Frankly, it's so broken ECM is no longer at the top of my hate list.

Of course you templars are sync dropping so you probably have no idea just how bad it is.

[JimSuperBleeder]

ELO in a nutshell: The better you play, the worse your teammates will be.

[Sifright]

Zylo, on 31 March 2013 - 03:02 PM, said:

Elo ratings seem to allow mixing of different ratings as long as the team average matches the enemy team. A member of my merc corp that is pretty good and wins often ended up with a player on the team that claimed to be new, was in a trial mech and performed like a new player. I think when Elo scores get high enough it's expected that a good player can balance a new or bad player but this doesn't seem to be the case which seems to increase the odds of 8-0 wins or losses. It's also not fair to new players to put them into matches where they aren't facing only new players with similar Elo rankings.

I suspect this is an accurate summation.

I don't understand why pgi would do it this way though.

Elo with out bracketing is utterly pointless.

New players should never be forced to play outside of their brackets because they will just get stomped repeatedly absolutely terrible idea.

[Primetimex]

Yes I experienced what ELO is like for a while this weekend on Saturday! Was playing with a 3-man group, inexplicably, we seem to end up in low-ELO games - where some ppl don't understand what glhf means and strolling out in the open has no consequence.

Suffice to say, we won majority of all our matches before we were back against the usual good enemies ...

[Hauser]

JimSuperBleeder, on 31 March 2013 - 05:43 PM, said:

ELO in a nutshell: The better you play, the worse your teammates will be.

Not how the match maker works. It picks a specific elo level and then starts matching people to that level. See the command chair posts about it.

Rather what I think everybody is seeing are new players. New players start with a default (average) elo rating. These are people who have not played that much and thus haven't reached the proper rank yet. Most players themselves are also average, so they get matched up with the recruits.

I've got a 3.00 k/d and 1.48 w/l ratio on my 4P (I would assume that gives me a good rating). When I play that I usually get some interesting matches with fairly good builds.

Edited by Hauser, 31 March 2013 - 06:02 PM.