10 replies to this topic

[Aresye]

Step 1: Create a group of 7 players.
Step 2: Ensure some of those players are new players/recruits with lower Elo values.
Step 3: ......
Step 4: Profit! (literally)

In all seriousness though, most players wanting to get involved with units want to be involved in larger, more organized drops. I mean, isn't that kind of the point of joining one?

To illustrate why this is now an issue, we need to take a closer look at how the matchmaker worked beforehand, and note, I will be operating under the assumption that my Elo is relatively high based on who I usually run into during solo drops and personal statistics. This is not by any means an epeen boosting post. If it was, I wouldn't mention how Cyleth kicked my ass in a duel yesterday (oops).

Example 1:
Overview: Myself and 2 other longtime members of the Clan create a 4man team with the 4th player being relatively new.
Purpose: Introduce new recruit to small unit teamwork in a more difficult setting.
Expected Result: Mid-High Elo drop with at least 1 other Mid-High Elo premade on the other team.
Actual Results: 50% of matches meet expected result. 25% of matches result in facing very high Elo premades. 25% of matches result in facing very low Elo premades.

Example 2:
Overview: It's late at night. Most of the full members have gone to bed and only a few recruits are still on. I create a 4man group with 3 recruits.
Purpose: Just have fun really.
Expected Result: Mid Elo drop with at least 1 other Mid Elo premade on the other team.
Actual Results: 50% of matches meet expected result. 25% of matches result in facing very high Elo premades. 25% of matches result in facing very low Elo premades.

Notice this is exactly the same result regardless if I have 3 newer players, or 2 average players + 1 newer player. Based on what I know of how the MM works, the results SHOULD be similar because if I'm at a high Elo then I have a much higher influence on the group average (25% overall influence).

With 7+ sized groups however, now my influence is much smaller on the overall average, meaning, if I'm the only high Elo player, then having a mix of average and below average players in my 7+ sized group will have a lower Elo average than a 4man drop.

The result of this, is both comical and sad. I used to think that the pugs on my team were often bad and, "must be low Elo players," but yesterday I truly learned what being in a low Elo game is like, and when you're used to facing players from SJR, Lords, HHoD, etc, all of the sudden you have an ABSOLUTELY INSANE advantage when thrown into a game like this.

How much of an advantage? Around the 8 kills, 1887 damage kind. A result of nearly every single player on the opposing team not familiar with the map, standing completely still in the open, not understanding weapon ranges, missing nearly every shot, etc.



I was hardly using my jump jets (think I jumped a grand total of 3 times in combat) because I figured it would be too much of an advantage, but that did little to influence the game.

So why would I have a problem with this? After all I enjoy winning, and I also enjoy playing in bigger groups, so why would I speak up about this issue when I could very well keep my mouth shut and continuously stomp low Elo players until I reached a ∞:1 KDR?

Simple answer: I believe this is detrimental to units that want to play competitively, and detrimental to players that are already good. There's a reason why people who are good add level 9 bots on Super Smash Bros, or snowboarders who tackle the most difficult black diamond slopes. You don't get better at things unless you're challenged, and as it stands with the current group matchmaker, the larger the group, the lesser the challenge, and thus, the lesser you have to learn, and the slower it takes to improve as both a player and a unit.

So how do we fix this?

[Sahoj]

I suggest a moar broad sample size. Your opponents are still adjusting.

Evaluate again after a month.

I have not been on the group queue since patch but I've also noticed that my stats are doubling up while PUGing in the new MM. I may even hazard to say that:
If the Matchmaker is presenting more balanced matches than before in terms of tonnage a skilled player may see more consistently high battle scores.

Knowing exactly how many lights/mediums/heavies/assaults are on the map is very valuable information in both your mech building and general approach to a battle.

Edited by Sahoj, 04 July 2014 - 07:10 PM.

[wolf74]

If I may ask a simple question, how do you know what ELO they are in?

PGI does not tell us our 4 ELO scores (Yes I said 4) We only know that we have 4 ELO due to a Video Dev Log. (the 4 ELO are for each weight Class Group)

Known info:
Win a Match Raises ELO amount Unknown
Loss a Match Lowers ELO amount Unknown
4 ELO Scores Per Player
One for each Weight Class (Light, Medium, Heavy, Assault)

Edited by wolf74, 04 July 2014 - 09:41 PM.

[HBizzle]

Why do people even have faith in the ELO? I think it is probably broken. No way to tell your own ELO or those of others to see if it is working.

[Aresye]

wolf74, on 04 July 2014 - 09:41 PM, said:

If I may ask a simple question, how do you know what ELO they are in?

PGI does not tell us our 4 ELO scores (Yes I said 4) We only know that we have 4 ELO due to a Video Dev Log. (the 4 ELO are for each weight Class Group)

Known info:
Win a Match Raises ELO amount Unknown
Loss a Match Lowers ELO amount Unknown
4 ELO Scores Per Player
One for each Weight Class (Light, Medium, Heavy, Assault)

I'm not saying you ever know somebody's Elo for sure, but you can get a general idea based on the way they play.

If you watched the video, did you notice that their team was en route to the radio tower around the 9-10min mark? Before the video started recording I was actually on top of the hill earlier around the 13-14min mark. Their team was running around their base, and then they followed 3-4 mechs as they were going up the far cliff (originally I thought they were all heading out of bounds).

One medium mech was completely off on their own attacking our DC'd teammate (that's the one I killed at the start of the video), nearly every one of their mechs were standing out in the open without moving at all, and no artillery/airstrike were dropped.

Average players and competitive players don't play like that, at all.

[spectralthundr]

Has anyone else noticed it's always clan wolf players creating these "new mm is broken threads?" The hell is up with that.

[Deathlike]

I'm going to hazard a guess.

When you're in a big group, the MM does a "hard job" of trying to find players on the opfor. So, most likely Elo is far less a factor. Without a screenshot of the end match scores (and looking for faction groupings), it is hard to determine.

So, even if the opfor DOES have a 7-man equivalent to yours, chances are your Elo does not match up with theirs.

The MM seems to prioritize the follow in a particular order:
Weight Classes on each side (most important)
Team Size (not very important)
Elo (generally ignored as part of the release valves)

When the MM is starving for a particular group it needs, it starts dropping stuff immediately, and Elo is #1 on that list. Team size is "not as important" though because I've had 8-mans with 2 2-man premades.. instead of a straight 4-man. While that can sound like a hindrance (it can be), you have to put in some extra effort to communicate to the entire team.

So, ultimately, Elo is far less of a factor (in BOTH solo and group queues). It doesn't really try to match them up when it is in dire need of a particular chassis.

[ShinVector]

Hmmmm... I recognise some of the players OP was up against in the video.
However.. It seems some of them were piloting crappy clan lights can't really carry...

Soo... Big clan wolf group versus... Some decent ELO players who are leveling their crappy clans light being teamed up with noobies assault mech pilots.

Alpine ER PPC... Enemy team is screwed... Just more of an example the ELO MM screw up of Decent ELO light pilot versus Decent ELO Heavy/Assault mech pilot.

Try piloting that Twolf without ERPPCs... And see how you fare.. But nice damage and kills though.

[ztac]

welcome to the world of broken MM , but obviously it's not broken for those lucky few that have almost constantly 'good' encounters! I'm quite sure that the MM is almost incapable of matching player ratings anyway from my matches.

With above going on it is obviously possible that whilst 50% of players are having ludicrously good encounters , the reverse can be said for the other 50% of players have really dire ones! Hence there will be the 'well done PGI your new MM is great' , opposed to the ' MM is broken threads'. But seriously the MM is broken (my category).

We people in the PUG queue have had to put up with this since the games inception and still have to put up with groups in the solo queue (nothing has really changed for us with the new MM other than the matches for some seem to be more off) , however I am shocked to see that someone said that our ELO is based on wins/losses as this would in no way reflect individual players abilities.

The facts are that this is a team based game and in team based games a players performance is affected heavily by the teams performance, i.e. a good player could be dragged down by being dropped constantly in poorly performing teams (although these matches are when a truly good player can shine , or just die horribly if the enemy team is co-ordinated). In fact pretty much most of your stats are the result of the teams overall performance. It would possibly make more sense that the ELO rating would b based on your match score or XP as this would tend to reflect a players abilities better than a mere win or loss.

Final note to all those 'My K/D ratio is going down or up' people, your K/D is based on team performance as I would imagine that you personally did not do 100% of the damage on the mech but merely got the last shot in. And you would actually be an undesirable player in a team if you were just waiting for an enemy to get to a certain % health before you shot it (to get the kill), incidentally you are aware that you get more C-Bills for an assist than you do for a kill so from a C-Bill point of view it would make no sense!

edit:- just to say sorry for the almost wall of text!

Edited by ztac, 05 July 2014 - 12:44 AM.

[Aresye]

ShinVector, on 05 July 2014 - 12:29 AM, said:

Try piloting that Twolf without ERPPCs... And see how you fare.. But nice damage and kills though.

I've had plenty of games >1200 damage using a Timberwolf with 2 Large Pulse, 3 ER Mediums, and 2 Machine Guns. The ERPPCs replaced the 2 Large Pulse originally, but with 3 ER Mediums and 2 ERPPCs it just ran too hot.

I call it my compromise build. It gives me the much needed PP FLD when I end up against the competitive groups yet doesn't end up being too unfair (well, so it seemed) against non-competitive groups.

Deathlike, on 05 July 2014 - 12:16 AM, said:

I'm going to hazard a guess.

When you're in a big group, the MM does a "hard job" of trying to find players on the opfor. So, most likely Elo is far less a factor. Without a screenshot of the end match scores (and looking for faction groupings), it is hard to determine.

Hopefully these will help illustrate the issue a bit more:
Note: These were all from a total of 4 matches in a row. The match I don't have a screenshot of is the one I did a measly 950 damage (still won).

10+2


7+3+2


6(1DC)+3+2


It isn't group size that's the main issue. It's group size + bad Elo matching that's allowing games like this to happen. I would not be getting the kinds of scores like you see above if my group was being matched against more competitive groups like they were before the MM change.

spectralthundr, on 05 July 2014 - 12:06 AM, said:

Has anyone else noticed it's always clan wolf players creating these "new mm is broken threads?" The hell is up with that.

I have no idea. Perhaps we feel more noble?

I'm just interested in seeing if there's a solution that still allows us to play with more than 4 people yet have a better, more evenly matched enemy team.

ztac, on 05 July 2014 - 12:43 AM, said:

edit:- just to say sorry for the almost wall of text!

Don't worry, you're talking to a guy who's average post is worse.

Edited by Aresye, 05 July 2014 - 03:39 AM.

[MischiefSC]

As a staunch defender of the MM and Elo, I would say that statistically, getting a good Elo match in the group queue is going to be HARD. You've got a lot of criteria that you're trying to meet with a tetris-style environment, only it's Tetris with pieces falling in with between 2 and 12 squares each and in completely random shapes. You've got to place them as quickly as possible with the other 2-12 shaped pieces to make a 24 square block.

Matching group sizes? Dunno if it even tries, it might. It gets pretty close for tonnage - surprisingly so. To match Elo..... hard stuff in the Group queue. Disproportionate number of high Elo folks there, playing with other high Elo folks.

This is part of why 2-4 man casual groups are having an issue. You need those smaller groups who arrive in larger numbers to fill in tonnage-wise. Elo.... if you're close in Elo 50% of the time you're doing well. Especially if you're in a 7man group - you've buggered yourself there. 10mans are likely easier to fill closer to Elo goal. 6mans too. 7mans need either a 3/2 or a single 5man, meaning you're very likely to end up with 2 small, casual groups. The 5man will likely be the high Elo companions.

Think of it like throwing craps dice. You can get a '6' with a lot of different combos, thus giving you more options. You get an '11' with a 5 and a 6, that's it. 10s get a 6 and a 4 or a 5 and a 5.