76 replies to this topic

[TOGSolid]

Today was supposed to be the reintroduction of weight matching but it obviously hasn't been implemented yet and there has ben zero communication.

The Elo matching is still horribly broken. While matches may be fairing better tonnage wise, there is still a huge discrepancy between skill levels on the teams. Good players should not be getting matched idiots running LRM stalkers who can't figure out how to lock on, nor should we be grouped up with people who shoot the ground more than their opponents. This averaging system is complete bunk. Hawken, of all things, has a much better player skill system that keeps things fun and competitive. The only reason I can think of to average things out like this is because there aren't enough players to run a proper Elo matchmaker.

[Adamar Ash]

I have noticed the matches being much closer in weight.

[ohtochooseaname]

All it takes for a steamroll is one player, making a stupid decision, and one other player to follow him. Then it's 8 vs 6, and you've got a huge uphill battle. Also, getting popped by a 5 ppc stalker can ruin your day.

[Phoebe]

After 3 days and countless of lost games is still the same.. most of the time in the losing team, 10 loses in a row... only time I could get some consistent wins is when I stopped pugging and played in a group (during the new elo i pugged like 95% of the games, so it's not that I have high elo thanks to teammates)... This game is becoming really frustrating, I had 2.6 win rate before the new elo system, now it's probably around 0.10 and I still lose all the time....

[Kageru Ikazuchi]

Phoebe, on 19 April 2013 - 01:09 PM, said:

After 3 days and countless of lost games is still the same.. most of the time in the losing team, 10 loses in a row... only time I could get some consistent wins is when I stopped pugging and played in a group (during the new elo i pugged like 95% of the games, so it's not that I have high elo thanks to teammates)... This game is becoming really frustrating, I had 2.6 win rate before the new elo system, now it's probably around 0.10 and I still lose all the time....

This is exactly why the Elo scores needed to be recomputed ... my experience has been exactly the opposite.

I've played about 1000 matches since Elo seeding started, and my W/L ratio was about 1:1 ... since Elo was introduced, that number had been going steadily down ... now it appears to be trending back up.

It appears that the recalculation of Elo brought your scores way up, and my scores way down ...

[Hekalite]

Since it's difficult to be objective when you are in the middle of a losing streak I decided to record the results of my matches to see just how well the system is currently working (or not) for me. After each match I would jot down the kills per side, if it was a win or loss, and if I subjectively felt the match was unbalanced. I did not track weight mismatches, but I did keep track of number imbalance due to DC etc.

Here is what I found over the last 30 matches:

Overall W/L: 15/15

Kills per side:
8/0 or 0/1: 1
8/1 or 1/8: 2
8/2 or 2/8: 6
8/3 or 3/8: 6
8/4 or 4/8: 3
8/5 or 5/8: 3
8/6 or 6/8: 2
8/7 or 7/8: 0

Other: (i.e. kills per side when some mechs were left alive or teams were not full)
7/7: 1
7/5: 1
6/6: 1 (this cap fight was actually a really good match)
4/3: 1 (boo, cap swap)
1/0: 1 (boo, cap swap)
0/0: 1 (boo, cap swap)
?/?: 1 crashed out

- hard to judge skill vs skill in the 4 cap matches
- Subjectively I felt like I was on the receiving end of a stomp 6 times and on the winning side of a stomp 3 times.
- Worst losing streak was 6 in a row.
- Best winning streak was 5 in a row.
- 6 matches did not have equal numbers either due to a DC or starting down 1.

So to sum up, based on my numbers the system seems to be working very well overall for me, but it can still a little rocky at times. Definitely a step in the right direction with this patch though.

Edited by Hekalite, 21 April 2013 - 01:28 PM.

[Devil Fox]

I still see largely unbalanced teams, it depends on the game mode you play... go conquest and one team will be 4-6 light's against the other team which would be 1-2 light's with the rest heavy and assault. Or go into assault, it'll be all heavy/assault against a team that may range from 3-4 mediums up to a few heavy and assault.

Their reason for not introducing the weight matchmaking component was for the outlier players (which shouldn't the system be widening it search enough to find other players) and 8man's. I don't understand why they would care about 8man's, before elo when all we had was weight class matchmaking the 8man queue was exempt from it. I can't see why they still can't decouple the weight matchmaking for the 8man queue, hell some people still haven't noticed that the 8man queue is affected by elo also.

All disappointing really since they really haven't given any real notice that we're not going ahead with it. The throw in this gimmick tourney which no doubt is being abused, and really all you're doing is inflating all the assault mech elo's which really turns this into assault online (which it has been for a week now).

[Kageru Ikazuchi]

Apostal, on 21 April 2013 - 04:15 PM, said:

... really all you're doing is inflating all the assault mech elo's ...

Now that they've fixed the Elo computation, it should be refining a participant's Assault Elo, not inflating it, right?

[Elder Thorn]

i played a couple of matches now.
I realise the system isn't completly implemented yet, but iÄll jsut type my thoughts as they are.

The first thing that came to my attention was, that i had to play Conquest on Alpine with 4 Lights on the enemy team and 2 on my team. It resulted in a loss by cap, as you can imagine.
The next few matches, nothing really stood out to me, but after a while i realised, that i see way more close matches and for a long time i even got an epic 1on1 in the end. Might be coincidence, btu for now id say this is definatly on the right way

[ArtemisEntreriCRO]

just had a match where my team had 2 assaults (me in atlas and a 8q, probably trial) against 2 ppc stalkers, atlas and an awesome
matchmaking does not work

Edited by ArtemisEntreriCRO, 23 April 2013 - 05:27 AM.

[Venraa]

For all those saying they have lots of games where it is 8-0, 8-1 or 8-2, you can't just use this number as an indication the game wasn't balanced.

As has already been said it only takes a couple of mechs on one side to wander off on their own and get taken out quickly for the game to dramatically turn (I am sure you have all seen those games with the Atlas doing 4 damage before being taken out). Just the simple act of one team sticking together and the other team splitting up is enough to decide the game with equal piloting and shooting skills.

Another thing to look at is how damaged are the mechs that survive? Failing to secure a kill can end up being the downfall of your team. An example from one of my games yesterday. I was in a stalker on Tourmaline. We lost 2 mechs pretty quickly. I ended up at the front taking a lot of enemy fire. Lost my left torso and went to yellow internals on the CT. The enemy stalker I was engaging got killed at that point and his team took cover and did not follow up and take me out. I was able to pull back and stay behind our atlas. We won 8-3 and I had 4 kills and 2 assists. If the enemy team had only finished me off the game would have been very different. This is not uncommon in my games. Even though the scoreline may not be close the actual game was and the survivors are very badly damaged and the game could easily have gone either way. Maybe if people targeted the enemy more often they could see how badly damaged some of them are and finish them off.

While I don't know my ELO I would put myself in the average category maybe just a little above halfway.

[LackofCertainty]

Have had one epicly close game since the patch. One of our highlanders got stuck in river city, and the 2 enemy survivors (one was a stalker missing a leg, other was a cataphract with no torso armor left and red/yellow/orange internals) decided to go cap win instead of trying to fight the stationary highlander.

Have had 1 8-0 stomp fest and 1 0-8 soul crushing defeat. (also a few 6-2 stomps)


Not enough games played yet for this info to be anything other than anecdotal, but I'm keeping my eye on it.

[Zylo]

I'm only running in groups of 2, 3 or 4 and I'm still seeing most matches ending up being 8-0 or 8-1 with a few being a bit closer and the very rare 8-7 matches. It's either an easy win or a fast loss, the current Elo system doesn't seem to produce many matches that feel balanced and seems less fun that the exact weight class matching that often resulted in closer games. Game type = Assault. Wins generally seem to go to the team with more assaults or heavies unless those heavies are Dragons.

In matches where the Assaults and Heavies were close (regardless of type), the win generally went to the team with more Gauss/PPC/ERPPC boats but that is a separate balance issue not related to matchmaker. These matches tended to be closer, 8-3 or 8-5 in most cases.


I ran a few conquest games and the team with more lights regardless of type seemed to win. The game sample was very small for conquest compared to assault mode.

[Jonathan Paine]

The current incarnation of the matchmaker seems to fail at both matching tonnage and skill most of the time. Good games are now the exception, bad games the norm.
Here is a really bad example (from a rapidly filling folder of examples):
630 tons vs 425 if I calculated correctly...

[Zylo]

Jonathan Paine, on 24 April 2013 - 06:18 PM, said:

The current incarnation of the matchmaker seems to fail at both matching tonnage and skill most of the time. Good games are now the exception, bad games the norm.
Here is a really bad example (from a rapidly filling folder of examples):
630 tons vs 425 if I calculated correctly...

This is an unexpected result as I noticed the game type was conquest and the lighter team should have an advantage as long as you guys can split up the enemy team to get some kills and go for the resource gathering win.

Of course with that many assaults if they fire enough PPCs at you eventually they will get lucky and hit but that's a different balance issue.

[Jonathan Paine]

Zylo, on 24 April 2013 - 06:39 PM, said:

This is an unexpected result as I noticed the game type was conquest and the lighter team should have an advantage as long as you guys can split up the enemy team to get some kills and go for the resource gathering win.

Of course with that many assaults if they fire enough PPCs at you eventually they will get lucky and hit but that's a different balance issue.

Caustic Valley. Seize volcano, dominate map with long range weapons.

Oh well. Got to celebrate the 20th client crash of the day.

[Liquid Leopard]

I've had several games in a row where one team was all heavy or assault mechs, while the other team has a bunch of speedy lightweights. This usually results in a base cap, or "death by dizziness" for the heavy team. I've had several games where one side has ECM mechs and the other doesn't. I just had a game where the other team had 6 founders, and my side was a bunch of nobodies, so that was the fastest loss I've ever seen. Elo system, my @$$.

Edited by Liquid Leopard, 25 April 2013 - 03:13 PM.

[Lostdragon]

Matchmaker is still badly broken. Just played a Conquest game. My team was 6 assaults and 2 heavies. The enemy team had 4 lights. It was no fun at all.

[Duncan Aravain]

The weight mismatch is getting out of hand. I had a match with 6 assaults on the other team and 1 on ours. This is supposed to be an improvement? Add the complete hashing of ELOs; I've had both tourney winners and complete noobs on my team the same match. Last match, we had 3 DC'd mechs on our side during the match. Guess what side won? Other than that, the game is working as intended...............

[01Shanx]

1 in 5 matches getting kicked to desktop is 1 in 5 too many. Should not be more than MAYBE 1 in 1000 matches. I've run the tool too no effect. With an i7- 3770 3.4Ghz machine running win 8 64 bit it IS NOT my machine, further I have no trouble running ANYTHING else, including MMO's with 200 or more players in the same area. It's almost like your server is failing to maintain connection.