76 replies to this topic

[Mawai]

Wow ... lots of vitriol and misinformation in this thread.

1) We do know (more or less) how the matchmaker worked as of December or so 2014 (based on comments made in these forums by the developer who wrote the code ... I don't think it likely we will ever get a better information source than that).

Given that PGI has spent almost all of their development effort in the last 6 months on CW ... I think it is reasonable to assume that the MM has NOT changed in that time frame (despite apparent conspiracy theories to the contrary).

Can I guarantee that? No. But it seems likely.

In addition, from my personal experience, I have noticed NO change in matchmaking for the matches I am in, for at least the past 8 to 10 months or longer.

2) KDR is irrelevant to the calculation of Elo or matchmaking. So is damage done ... or any other in game performance metric. Keep in mind that these metrics may not indicate whether the player actually contributes to winning the match or not. I've seen a player in a Timberwolf score 5 kills, 3 assists, over 1000 damage ... most of it after the rest of the team was dead (at least 4 of the kills happened then ... and a lot of the damage as he sprayed lasers all across the opponents).

Do any of these stats matter in a game? No. The only thing that matters is whether you win or lose.

As a result the MM tries to populate the teams with about equal Elo value players across weight classes. It then calculates which side it expects to win based on Elo. The rewards for the match depend on the actual win/loss outcome of the match when compared to the prediction. If you win, your Elo will go up .. if you lose it goes down .. but the amount of change depends on the probability of the predicted outcome.

What this means is that you could hypothetically win one match and lose three and your Elo could still be higher after the four matches than before you started. Similarly, you might win three and lose one but your Elo might drop. However, over the long term ... if the MM is placing you appropriately, then your W/L should approach 1. If you are getting better at playing the game then your wins will likely go up .. and so will your Elo. If you are leveling a new mech then you might not be as effective, you might lose more often and your Elo drops.

3) Personally, I am either in a good place with Elo on heavy mechs or I have a higher tolerance for bad matches (or perhaps the larger number of heavy mech players just makes it easier to find more folks in the same Elo range as I am) ... whatever the reason, I don't get very many matches with people doing their own thing. On the other hand, I will almost always state in team chat ... "Stick together, focus fire, have fun! " ... "let's regroup with the assaults in XX" ... etc. I try to minimize random derpiness just by saying a few words. Does it work? I have no idea but on the other hand, I can't find as much to complain about in matches where I play a heavy mech.

Mediums have also bee pretty good recently.

Playng my lights might be a different story but I also tend to take some chances I shouldn't in light mechs just because they are so much fun

Edited by Mawai, 14 July 2015 - 07:38 PM.

[Idealsuspect]

Mors Draco, on 14 July 2015 - 06:32 PM, said:

OOOooo We are showing off screenshots? Look at this malarkey...1000 damage and NO KILLS. Damn team full of kill stealing hippies I tells ya.

Bro you want some kills ? Stop use lrms ...

With pulse laser pinpoint you just need 150 damage for get a kill ( dire atlas ) by only yourself and it will take you 30 seconds if you are a god.

With lrms you will need inflict near than 500 damage point and it will take you more than a minute ( if the target is in open space ) god or last of noob ....

Morality, lots of damage don't mean automaticly you are good ....i don't said you arent i dont know you at all.
Sorry for engrish but too late for using google translator. xD

Edited by Idealsuspect, 14 July 2015 - 07:43 PM.

[Jman5]

Idealsuspect, on 14 July 2015 - 07:41 PM, said:

Bro you want some kills ? Stop use lrms ...

With pulse laser pinpoint you just need 150 damage for get a kill ( dire atlas ) by only yourself and it will take you 30 seconds if you are a god.

With lrms you will need inflict near than 500 damage point and it will take you more than a minute ( if the target is in open space ) god or last of noob ....

Morality, lots of damage don't mean automaticly you are good ....i don't said you arent i dont know you at all.
Sorry for engrish but too late for using google translator. xD

I did a test one time where I ran a laserboat and a LRM boat in the testing ground. Each one was quirked for their respective weapons and I timed how long it took to kill a mech while I was running full speed from left to right.

You know what the result was? There was no real difference in Time to Kill between the two. Everyone likes to pretend LRMs are some horribly inefficient weapon, but when you put people into games where you and your opponent are running and gunning in and out of cover we're all a lot less accurate than we like to admit.

I have done plenty of 1v1s and baring some fluke 50 meter ambush, the LRM mech almost always kills more quickly than the meta mechs.

The high damage of lrm mechs has less to do with spread. It has more to do with

1. High DPS and superior heat efficiency
2. Long range
3. indirect fire

I just don't buy the argument that these meta mechs I see doing 200 damage are never missing and only hitting CT. There is no evidence to suggest that people running lasers or ballistics are more accurate than someone running LRMs. Especially once you get beyond 300 meters.

[Idealsuspect]

Jman5, on 14 July 2015 - 10:11 PM, said:

I did a test one time where I ran a laserboat and a LRM boat in the testing ground. Each one was quirked for their respective weapons and I timed how long it took to kill a mech while I was running full speed from left to right.

You know what the result was? There was no real difference in Time to Kill between the two. Everyone likes to pretend LRMs are some horribly inefficient weapon, but when you put people into games where you and your opponent are running and gunning in and out of cover we're all a lot less accurate than we like to admit.

I have done plenty of 1v1s and baring some fluke 50 meter ambush, the LRM mech almost always kills more quickly than the meta mechs.

The high damage of lrm mechs has less to do with spread. It has more to do with

1. High DPS and superior heat efficiency
2. Long range
3. indirect fire

I just don't buy the argument that these meta mechs I see doing 200 damage are never missing and only hitting CT. There is no evidence to suggest that people running lasers or ballistics are more accurate than someone running LRMs. Especially once you get beyond 300 meters.

In fact in testing ground... all loadouts seem legit and powerfull, in game isn't same.

More than that with difference of ping, pilots skill, target skill, difference of loadout ( tag artemis, radar depr, target decay, targeting computer ) and the location of duel, one lrm boat will work nice or bad depend of pilots of course BUT a lrms user who have 1000 damage point with 0 kill it can happend a more than a laser boat with 1000 damage and 0 kill...


You talk about long range and indirect fire but don't worry good pilots know how advance to a lrms boat without walking in open space testing ground static targets can't even move for dogde .

I would enjoy have a duel with you, a laserboat ( or all others pinpoint boat like balistic, srms in fact it's same ) versus your lrms boat in every weight duel you want for see if your theory is really legit or work only agaisnt average players while they are fighting ( yours teamate for example ) or testing ground mechs.

[EightBitKnight]

Just comes down to every mech has its off days. Lrms can do just as good as the other weapon types but depend heavily on your team to be fully effective.

[Throat Punch]

EightBitKnight, on 15 July 2015 - 04:43 AM, said:

Just comes down to every mech has its off days. Lrms can do just as good as the other weapon types but depend heavily on your team to be fully effective.

Yep. My lrm boat is meant to add armor melting fire support to my buddies in their Timber Wolf and Ebon Jaguar. They call out targets, or narc them, and I let missiles fly. The mech i'm using in that screenshot isn't supposed to get kills, its supposed to be an assist machine that provides support for the other mechs to move in and finish the job. I was just being silly when i was complaining about not getting kills. Killing stuff in a Mad Dog A is just a bonus with the build I use. I know I lack the punch to kill most mechs out right with just missiles. Besides assists, component destruction, etc pay more than kills do anyway.

[Jon Gotham]

Mcchuggernaut, on 14 July 2015 - 02:05 PM, said:

So I broke 3.2 K/D in my Dragon (First time I have done that in any mech. There's just something about it that works for me.), and started noticing something. When piloting it and spectating, I am now consistently seeing players who can't aim, fail to press "R" and target enemies, wander aimlessly, make dumb decisions, and seem to have no idea in general what they are doing. I can do everything right, deal over 1000 damage with six kills, and my team will often fall apart and still lose, despite me practically doing the work of half a team by myself. That's an extreme example, but you get the picture. I'm seeing a huge increase in, um, "questionable skill" when spectating after crossing the 3.0 mark.

So my question is this: Does the matchmaker balance the teams based on your skill with individual mechs? I was under the impression that it just took into account your over-all skill level, but after seeing what I have been seeing for the past few days, I'm starting to wonder...

I usually hover around a 2.20 KD overall, and have not managed better than a 2.60 in any of my other mechs, individually. I have learned to expect having to carry a little with them, but not like I have been having to with my Dragon.

So, thoughts?

Most players in games like this simply aren't very good. Take WOT for example. It had a mode called XVM that measured player stats putting them into dark red, red, dark orange, orange,yellow, dark green, green, cyan and purple. Dark red being the lowest and purple being the best.
Dark red and red is basically bot level play skill.
I used to be a cyan level player, on the vast majority of my pug teams I'd be the solo cyan with at LEAST 10 reds. Some of these players had played over 10 thousand battles. 10 thousand attempts at something and 0 improvement.
I'd assume this is no different?

[Lugh]

Johnny Z, on 14 July 2015 - 03:16 PM, said:

Stealth "Omni mechs are not OP" post.

They aren't but you won't do the math to see it for yourself.

[Remarius]

War Beast, on 14 July 2015 - 04:06 PM, said:

Again. You DO NOT know for a fact. You DO NOT have the formulas in front of you. Your reply about town hall gives no specific hard data does it ? Just a general statement from tow hall.

You honestly 100% do not know the details on current MM.

Again you are guessing based on previous statements. Guessing as there has been no detailed mm data in a long time. Things could easily have changed and or been modified.

AGAIN, you DO NOT know.

It is that simple. You are the one who needs to rethink what you are saying.

And unless you are providing a CURRENT dev provided mm matrix, you dont know.

Guesses, assumptions ...

The paranoia and madness in this one is strong....

[Paigan]

Mcchuggernaut, on 14 July 2015 - 02:05 PM, said:

So I broke 3.2 K/D in my Dragon (First time I have done that in any mech. There's just something about it that works for me.), [...]

Sader325, on 14 July 2015 - 02:08 PM, said:

Stealth "I'm better than all of you" post.

Mcchuggernaut, on 14 July 2015 - 02:16 PM, said:

Not at all. I have only middling results with some of my other mechs.
[...]

So. Do I read that right as "Dragon is OP as hell and must be nerfed hard"?
I mean with a quirk of like 200% DPS it's not so surprising...

Or which one shall it be?

Edited by Paigan, 15 July 2015 - 05:50 AM.

[EightBitKnight]

All that lrm talk makes me want to pilot my A1 and MDD-A.

** should've been A1, haven't played my C1 in while either was on my mind.

Edited by EightBitKnight, 15 July 2015 - 06:01 AM.

[Jman5]

Idealsuspect, on 15 July 2015 - 04:28 AM, said:

In fact in testing ground... all loadouts seem legit and powerfull, in game isn't same.

More than that with difference of ping, pilots skill, target skill, difference of loadout ( tag artemis, radar depr, target decay, targeting computer ) and the location of duel, one lrm boat will work nice or bad depend of pilots of course BUT a lrms user who have 1000 damage point with 0 kill it can happend a more than a laser boat with 1000 damage and 0 kill...


You talk about long range and indirect fire but don't worry good pilots know how advance to a lrms boat without walking in open space testing ground static targets can't even move for dogde .

I would enjoy have a duel with you, a laserboat ( or all others pinpoint boat like balistic, srms in fact it's same ) versus your lrms boat in every weight duel you want for see if your theory is really legit or work only agaisnt average players while they are fighting ( yours teamate for example ) or testing ground mechs.

I'm usually on during the evening (North America) if you're on just let me know.

kamiko kross, on 15 July 2015 - 05:33 AM, said:

Most players in games like this simply aren't very good. Take WOT for example. It had a mode called XVM that measured player stats putting them into dark red, red, dark orange, orange,yellow, dark green, green, cyan and purple. Dark red being the lowest and purple being the best.
Dark red and red is basically bot level play skill.
I used to be a cyan level player, on the vast majority of my pug teams I'd be the solo cyan with at LEAST 10 reds. Some of these players had played over 10 thousand battles. 10 thousand attempts at something and 0 improvement.
I'd assume this is no different?

The people that spend 10,000 matches blaming pugs, or balance never stop for a moment to consider their own shortcomings. It's such a poisonous mindset. The sad thing is improving dramatically does not take very long. They just have to take that first step and admit they're not as good at something as they could be.

I dunno, personally I love finding some hole in my game that I can now patch up.

Edited by Jman5, 15 July 2015 - 09:59 AM.

[Mcchuggernaut]

Paigan, on 15 July 2015 - 05:49 AM, said:

So. Do I read that right as "Dragon is OP as hell and must be nerfed hard"?
I mean with a quirk of like 200% DPS it's not so surprising...

Or which one shall it be?

I have been asked this a lot, and I gotta tell you: I mastered my 1N, and to do so I had to elite the two other c-bill buyable Dragons. It took about 50 matches with both of them, and it was NOT pleasant. Even shuffling around engines and modules from other mechs to try and make up for the shortcomings of a newly purchased mech, and even after upgrading them both with endo and double heatsinks before I actually took them into a game to get XP, they did not perform well. I struggled to get kills and put up decent damage numbers. With the 5N I ended up with a 1.48 K/D and an even number of wins and losses. In the 1C, I did even worse: two more losses than wins, and a paltry 1.21 K/D.

I have been playing MWO since the days of open beta, and consider myself a competent veteran pilot. I am of the opinion that if I can't put up a 2.0 K/D and a 1.3 W/L ratio in a mech when I'm in tryhard mode, then that mech or build isn't good enough. So in answer to your question: You wouldn't see a veteran player piloting a Dragon of any sort by choice except the 1N, and only because of it's quirks. That being said, it's a glass-cannon skill mech. Deadly in the right situations, but useless if you lose the fragile AC/5 arm, and with extremely vulnerable XL side torsos (Which you MUST have, to make weight for 2 AC/5s and sufficient ammo.). Also, pinpoint high-alpha is king over DPS, always has been, and the 1N is a DPS machine. So it IS OP. Against inexperienced players, or people with bad aim. Once you start playing other vets, though...not so much. Honestly, I think it's about where it needs to be, and it's other variants actually need a slight boost. Before the quirkening I would laugh to myself every time I saw a Dragon on the battlefield, and think "Time for an easy kill!". Now, I check to see if it's a 1N, and if it isn't I still laugh. If it is, I aim for the right arm or torso with an alpha, and if I take it off I know I have already won or crippled my enemy.

Edited by Mcchuggernaut, 15 July 2015 - 10:09 AM.

[DONTOR]

Mcchuggernaut, on 14 July 2015 - 03:21 PM, said:

Definitely not. They had to quirk the arse off of the Dragon to make it playable. Anyone who piloted a Dragon before the quirkening was either experienced or suicidal.

Fixed it for you.

[Mcchuggernaut]

DONTOR, on 15 July 2015 - 09:52 AM, said:

Fixed it for you.

Wait, are you telling me you did? I question your sanity and am jealous of your clearly massive steel balls.

[Alistair Winter]

I keep seeing Mcchuggernaut in my games, so I can confirm that he's trapped in the steering wheel underhive... with me.

[DONTOR]

Mcchuggernaut, on 15 July 2015 - 10:10 AM, said:

Wait, are you telling me you did? I question your sanity and am jealous of your clearly massive steel balls.

Dual PPC + UAC5 Dragon Fang was pretty effective in the PPC AC5 Meta days, fastest PPFLD heavy you can make

[Roadkill]

War Beast, on 14 July 2015 - 03:44 PM, said:

No. Truth is we dont actually know what pgi uses at this time. All we have is guesses based on previous statements. They could have changed (added and / or removed) criteria at any time and not have, nor had to have, informed us.

Especially possible when you consider they have announced upcoming elo and mm changes.

Incorrect, but with the correct intent.

We know what Elo is and how it works. Elo is a specific, well-defined system. PGI uses Elo as one input into its Matchmaker, among other things. It's those other things that we don't know about or really understand.

People confuse Elo with the Matchmaker all the time, but they're completely separate systems. You could have a Matchmaker without having Elo ratings. You can also have Elo ratings without having a Matchmaker.

[Roadkill]

Kristov Kerensky, on 14 July 2015 - 04:20 PM, said:

Elo is a flawed ranking system for team games that rank individuals based on team performance, fact, so the MM currently, and going forward after the changes are made to how Elo ranking is determined, is a flawed system because it's using bad data. Until they stop using Elo, that won't change.

Not at all. Elo is a mathematically proven ranking system. It can work equally well in 1-vs-1 games and multiplayer games, but that depends on how it is used.

Elo is not a flawed system. The way that PGI is using Elo is flawed. Sorry to be pedantic about it , but that's an important distinction.

[Roadkill]

Kristov Kerensky, on 14 July 2015 - 05:15 PM, said:

Novakaine, I had a drop last night, I did over 700, 3 kills, rest of my team COMBINED didn't break 500. 11 people, under 500 total damage combined.

Screenshot or it didn't happen. ;-)

I had a game like that once. 700+ damage, 3 kills (no other kills on team), the rest of my lance and all of Bravo had less than 100 damage. Charlie had 2 in the 125-150 range and 2 just over 200. Roughly 1200 damage total from the rest of the team.

I thought I'd out-damaged the rest of the team, but when I actually added it up it wasn't even close. I'd really need a screenshot to believe that an entire team did < 500 damage... that's < 50 damage each and as much as we like to rant about it, < 50 damage really doesn't happen all that often not counting disconnects/afks.