48 replies to this topic

[Foxfire]

I remember that there was a point where you are being matched one to one for a similar weight class... now I'm seeing wonkyness like yee olden days of early closed beta where you'd have groups of assaults dropping against mediums and lights.

[Amsro]

The high amount of 4 man lances dropping makes matchmaking near impossible when the whole lance drops in similar mechs.

There aren't many options other then limiting lances to 1 of each class of mech, or 3 of each class per team (12vs12).

That would almost instantly balance matches better. But would also limit mech options.

[BladeXXL]

Match Making is just a Fairytale...

I've pugged yesterday mostly having:
- trial-noobs in my team against 1 or 2 4-man-premades with a lot of poptarts (2xAC20 or PPC/Gaus-Combos)
OR
- 200+ Tons difference between the sides (if I was on the havier side: we won, else we lost)!

[Zeus X]

Match making is becoming more and more of a problem as time goes on.

For the last week my team mates and I been getting almost nothing but trial pugs, while the other team is getting fairly decent eight players and resulting in a roflstomp cause our pugs can't breach 100+ damage.

Until PGI can separate the trial pugs from normal pugs, these matches are just going to be trash.

[5th Fedcom Rat]

Last week I lost more games in a row than I had ever lost since starting the game. This was dropping solo exclusively. I certainly didn't lose all my skill at this game in a month, nor was I using bad mechs.... all I can point to is the matchmaker. Namely its seeming inability to balance skill properly, mech loadout/size properly, 4-man premades properly (and of course, PPC stalkers, but if ELO was designed right it would put equal numbers on both teams). Anyway, I've just stopped playing the game for now.

[Smokeyjedi]

5th Fedcom Rat, on 24 June 2013 - 03:34 AM, said:

Last week I lost more games in a row than I had ever lost since starting the game. This was dropping solo exclusively. I certainly didn't lose all my skill at this game in a month, nor was I using bad mechs.... all I can point to is the matchmaker. Namely its seeming inability to balance skill properly, mech loadout/size properly, 4-man premades properly (and of course, PPC stalkers, but if ELO was designed right it would put equal numbers on both teams). Anyway, I've just stopped playing the game for now.

A whole lot more of this!!! The tournament only Exacerbates these issues. ---Like moths to a flame!

[Fooooo]

Im guessing the no restrictions on groups, but keep the groups in a seperate queue (only bringing in people from the solo queue when 100% needed) is probably looking a lot more promising for them.

It would open up 5-11 mans.

So

11man + 1LW

10man + 2man
10man + 2 LW

8man + 2man + 2LW
8man + 4man

7man + 2man + 3man
7man + 4man + 1 LW


5man + 5man + 2man

etc etc etc etc etc etc etc.

Throw all the groups into a "side" queue , it tries to match the teams group sizes as best it can and only pulls in solo players when its been waiting 2 or 3 mins or something, or when it requires them to fill out a solo slot because of the odd group sizes allowed.

Edited by Fooooo, 24 June 2013 - 04:14 AM.

[knight-of-ni]

Yes, I agree with the OP. Something seems different. I've noticed more, newer (read - *very* new) players on my team, and I've also noticed unbalanced drops.

Yesterday was getting predictable... Tourmaline, conquest, no light mechs and only 1 assault mech on our team.... Yep - we just got dropped against a 4-man group with 3 light mechs and 1 ppc stalker. Called it!

As a pug, I learn to just roll with the punches, but yesterday seemed pretty extreme. Wins were very hard to come by.
Will try again. Thank you sir may I have another.

[Balsover]

I'm thinking with the lower playerbase, the matchmaker just doesn't have a big enough pool to draw from in order to fill balanced teams. It eventually has to start accepting any players to fill a game or you would be waiting for a very long time.

[LauLiao]

Weight matching isn't necessarily the be-all and end-all. Had a match recently with One heavy and one assault on our side, the rest were mediums and lights. The other team was comprised entirely of heavy/assaults except for 2 mediums. We won and not by out-capping. The final count was something like 8-0 or 8-1. Not always true, but good pilots can generally overcome weight differentials.

[Matroid]

They had class-matchmaking for a while, and most people were happy. the worst that could happen is your team drops with a Commando, Dragon, and Awesome, and the enemy team gets a Jenner, Hunchback, and Atlas


Then somewhere in the last few months (March time frame??) they implemented the ELO-Matchmaker. this takes into account a bunch of stuff, and almost doesn't include tonnage. so when people figured this out, they drop with 4 Assaults... and the enemy team (~within a 2 minute window) gets 4 random mechs. if they aren't Assaults too, then it's almost always a stomp

[Goose]

DCM Zeus, on 24 June 2013 - 02:10 AM, said:

For the last week my team mates and I been getting almost nothing but trial pugs, while the other team is getting fairly decent eight players and resulting in a roflstomp cause our pugs can't breach 100+ damage.

Smurfing.

You were facing smurfs …

[Mawai]

Well ... unless PGI releases some statistics or heuristics on how their match maker is supposed to work we will probably never know for sure.

However, the objective of the matchmaker is to produce balanced matches which could be won by either side. So far, in my experience it has generally obtained this objective at least in the certain parts of the ELO spectrum in each mech class that I get exposed to ...

The number of rofl stomps is generally less than with earlier versions of the match maker (though they still happen) ... and close results, where there are only 1 or 2 mechs left standing on one team, appear more frequent.

The problem with the match making is trying to weigh ELO and weight class. How much is a high ELO player in a light mech worth compared to a lower ELO player in an Atlas? In addition to this, any reasonable communication or organization of a group is generally more important that either ELO or tonnage.

So there are three main factors that need to be included in a match balance ... and in my opinion probably in the following order:
1) groups
2) pilot ELO (or some estimate of their combat effectiveness on a team .. maybe ELO based on W/L isn't the best choice?)
3) mech

So of all the balance factors ... mech tonnage comes in last ... so it isn't surprising to see matches with imbalances in tonnage IF the PGI matchmaker is weighting pilot ability higher than the mech they are running. Unfortunately, tonnage is the most visible factor in any match so it is easy to point to it and say the match maker is broken because there is an imbalance in the tonnage on the two sides. However, grouping and ELO are likely much more significant factors in determining the outcome of the match than pure tonnage and that information is not generally available before or after a match.

I've had very close matches where the tonnage difference was on the order of 200 ... sometimes won by either side. The big issue is that the extra tonnage is a buffer. The side with the lower tonnage can't afford mistakes or bad luck ... which are bound to happen. Light mechs can be extremely effective but if someone gets a lucky shot and legs the light as it comes around a corner ... then the light mech is toast. A Jenner can take out an Atlas by itself unless the Atlas pilot is both good and lucky ... or if the Atlas has support of team mates. So mech tonnage is clearly not the only factor.

Anyway, I think the match maker can probably use additional tweaking ... particularly in generating balanced teams in terms of groups ... but overall looking at the tonnage on the end game screen is probably not the best way to judge whether the match maker is working or not.

P.S. If it isn't in place already ... I think PGI needs separate ELO or combat values for each player when running solo or when grouped ... since the W/L record (upon which ELO is apparently based) will likely be quite different for the two cases.

[BladeXXL]

I think this is where the tournaments are for.
PGI tries to find a formula to measure "skill".
Hopefully they replace the current ELO as fast as they can and we can enjoy balanced games. ^^

[Amsro]

BladeXXL, on 25 June 2013 - 05:36 AM, said:

I think this is where the tournaments are for.
PGI tries to find a formula to measure "skill".
Hopefully they replace the current ELO as fast as they can and we can enjoy balanced games. ^^

The main problem with Elo is that it requires bulk data, lets say around 1000 matches to accurately get your "Elo Bracket or Score" to where it should be.

Resulting in all ... ALL players required to play 1000 matches in all classes they wish to play in order to get to thier Elo level, to receive balanced games. Little extreme no?

[zhajin]

Amsro, on 25 June 2013 - 07:02 AM, said:

The main problem with Elo is that it requires bulk data, lets say around 1000 matches to accurately get your "Elo Bracket or Score" to where it should be.

Resulting in all ... ALL players required to play 1000 matches in all classes they wish to play in order to get to thier Elo level, to receive balanced games. Little extreme no?

this is true, though with so many random variables involved in every match, that number is likely much higher than 1000. the fact is this is a heavily team based game, and if you play solo you will have to accept the fact that there will be a lot of things out of your control. really the player base is far to small for elo to work well, given the vast number of things that can effect a match win or loss.

Edited by zhajin, 25 June 2013 - 08:58 AM.

[Krivvan]

Over hundreds of games over months I haven't seen a trial pug even once.

I absolutely hate class matching. I want asymmetrical weight games. A heavier team doesn't inherently have an advantage over a lighter team. Especially outside of 8v8 games, a few Lights can absolutely destroy a team double their tonnage.

Wouldn't it be boring to know that no matter what the enemy team will have the same exact weight classes as you?

[Foxfire]

Eh.. there is one thing about having asymmetrical games.. and another about having two sides with a significant weight difference.

I pilot light mechs and I've had quit a few matches where there have been 4 lights on my side where we just dominate... but then I've seen many matches where I've been the only light on the side(with everything else heavy and assault). I don't care about having perfect weight balance but I do think that the game is funner when there is a bit of variety and not one extreme vs the other.

[BladeXXL]

Amsro, on 25 June 2013 - 07:02 AM, said:

The main problem with Elo is that it requires bulk data, lets say around 1000 matches to accurately get your "Elo Bracket or Score" to where it should be.

Resulting in all ... ALL players required to play 1000 matches in all classes they wish to play in order to get to thier Elo level, to receive balanced games. Little extreme no?

That is not true!
You start with a ELO score of 1300 and can loose or win max. 50 points per match.
So if you s.u.c.k in every single game and loose 50 points, it takes 26 matches to land on ground zero!

all can be read here: http://mwomercs.com/...ost__p__1626065

[Mawai]

Just finished a match that pretty much illustrated some of the points I made above regarding matchmaking.

Dropped in an Assault match on Alpine. There were 5 assault mechs on the opposing team and one on ours. The tonnage difference was 145 tons (620 to 475 I think). Only one light mech/side (both Ravens) and all the ECM was on the opposing team (3 Atlas DDC ... neither Raven had ECM).

We lost ... but it was surprisingly close given the difference in tonnage and ECM. The remaining mechs on the winning side were pretty beat up (red CT) ... except possibly their LRM DDC. A few more hits and the match could have gone either way. They had 2 Atlas and a Highlander survive the match.