92 replies to this topic

[Exoth3rmic]

Genewen, on 08 June 2013 - 07:28 AM, said:

I did not talk about a 50:50 chance to win a specific match. This should be quite obvious, especially with the quote of an PGI official. The goal of elo is to give you an overall number that is roughly 50% wins and 50% losses. Please try to read properly what people are writing. Thanks.

I read what you wrote, and as I said, it isn't true. Your hypothesis is that you should have a 50:50 win ratio over time, this is demonstrably and factually incorrect.

It might be how you would like it to work but it is not in point of fact the case. As in the very link I put in my post the goal of the MM is not to push you to 50:50 wins it is to reward you proportionately to the weighted probability of you winning which is something that will depend on the players joining and qued at a the time you wish to play.

The MM isn't working within tight enough limits to push the top players towards 50:50 win loss. I am unaware of any dev statement that says it should be doing so.

Top Elo players would be qued for a very long time if this was the case.

[Appogee]

theNemesis, on 08 June 2013 - 05:29 AM, said:

Is a KDR of 5.1 common? I don't mind admitting mine is below 1. I got killed a lot in the first few weeks w/few kills.

I only started about three months ago, and after a month when I first checked my stats, it was 0.5, which was personally mortifying.

I've managed to get my overall k/d up to 1.75 in the two months since, mostly in Mediums and Heavies. I only ever PUG.

Interestingly, my k/d in Assaults (3 varieties of Atlas) over the past month is between 2.5 and 3. Again, only ever pugging.

Conclusions:
1. Pugging in trial mechs is brutal on your k/d.
2. The heavier your ride, the higher your k/d is likely to be.

I can't imagine ever getting to 5.0. It would take a stats wipe and me playing Assaults in an organised clan to even have a change at that.

Edited by Appogee, 08 June 2013 - 09:02 AM.

[Waking One]

My guess is they implemented the most ******** possible kind of matchmaker tied to elo which is the better you are the worse players get put on your team to "balance it out".

[Genewen]

Exoth3rmic, on 08 June 2013 - 08:34 AM, said:

I read what you wrote, and as I said, it isn't true. Your hypothesis is that you should have a 50:50 win ratio over time, this is demonstrably and factually incorrect.

It might be how you would like it to work but it is not in point of fact the case. As in the very link I put in my post the goal of the MM is not to push you to 50:50 wins it is to reward you proportionately to the weighted probability of you winning which is something that will depend on the players joining and qued at a the time you wish to play.

The MM isn't working within tight enough limits to push the top players towards 50:50 win loss. I am unaware of any dev statement that says it should be doing so.

Top Elo players would be qued for a very long time if this was the case.

Pray tell me, why do you ignore the statement of a PGI official that I linked? You say that you are not aware of a statement that underlines my statement? No wonder, you are GODDANG IGNORING IT WHEN IT IS ALREADY JUMPING IN YOUR FACE! So once again, just for you:

srccoder, on 07 June 2013 - 09:31 AM, said:

all that elo does, is trend a player towards a 50/50 win/loss ratio. It doesn't equal skill, it's just a correlation at best. so the way you can check matchmaking, is to track you win/loss ratio on your end over time.

Edited by Genewen, 08 June 2013 - 09:06 AM.

[Exoth3rmic]

Look, that "staff" is wrong. Its pointed out within the very thread in which the false assertion is made. We will continue that in a private discussion with the very staff member who made the assertion concerned, just so we can clarify this for you.

For the purposes of this thread, its quite clear to me, to Chemie and probably even to you that the population of trial mechs has gone up within the games we are playing. Your assertion that this is matchmaker pairing him with "low" Elo players to push him to 50:50 losses has not been true up to this weekend.

It is very rare in the 2000 matches i've played since Elo was introduced that I see a stock mech, let alone a trial mech and I have seen plenty this weekend already.

As I suggest its either the server population spiking and hoovering up all the limited number of skilled players or something was changed.

And now, kind sir, to our private conversation.

Edited by Exoth3rmic, 08 June 2013 - 10:34 AM.

[Bobdolemite]

a 5.1 K/D is the stat of someone who plays almost exclusively in groups and does.... ok. 8.0+ kd is not unheard of in preforms, Theres nothing imbalanced about that is there? (having 8 times more kills than the average player all due to grouping is definitely not evidence of a broken game mechanic)

My kd is 1.38 which I tend to be proud of being a PUG only player.

Matchmaker seems to be working, but then again it never worked that well. I tend to agree with OP that less people are playing hence the longer queue times.

Edited by Bobdolemite, 08 June 2013 - 10:34 AM.

[Thuzel]

Nicholas Carlyle, on 08 June 2013 - 07:02 AM, said:

[...]
Because PGI purposely hides all the numbers.
[...]
So all of you guys are blowing hot air pretending to have an idea of what is going on. But none of us do because PGI refuses to release anything because they are so damn scared of what might happen.

It wasn't because they were scared, or at least, not entirely because of it.

The real reason is that MWO is being treated as only a mid-term project. Also, that PGI's epic QA failures would be highlighted by the player count (as it was in CB). Fluctuating playercount numbers would significantly reduce current revenue and would let people know when the end is actually near.

The thing is, this game is on auto-pilot right now. Either it was a conscious business decision, or it's just the way things have worked out, but IGP/PGI want to maximize revenue without investing too heavily into the future of this game. The only way to do that is to smooth out any obvious deficiencies caused by a lack of investment. One of the easiest ways is to deny us any real information about the game's performance. They want us buying into the game until the bitter end, or at least until the point where they would be held legally liable for refunds.

They've also probably expected a significant dip in the player count while we've been waiting for CW, and they want to gloss that over. We're due for the first phase of it very soon, and as soon as people realize that the first phase will be more shallow than a kiddie pool and that we won't get any kind of decent meta-game for another 4 or 5 months, people are going to be upset.

Just my opinion though. Based on how they've handled every significant backlash so far: deny, hide, ignore, and backpedal if all that fails.

[jeffsw6]

Appogee, on 08 June 2013 - 09:01 AM, said:

I only started about three months ago, and after a month when I first checked my stats, it was 0.5, which was personally mortifying.

PUGging will do that to you. Also, K/D is not everything. When you disarm an opponent, do you mindlessly chase him down to score a kill, or do you move on to another enemy mech which is still a threat to your team? Many players do the first, have a few more kills, but win less matches.

Thuzel, on 08 June 2013 - 11:04 AM, said:

The thing is, this game is on auto-pilot right now. Either it was a conscious business decision, or it's just the way things have worked out, but IGP/PGI want to maximize revenue without investing too heavily into the future of this game.

I don't think that is true. I think they are just so bad at game-play balance that their impressive efforts in graphics, audio, etc. are failing because they don't understand that game balance is what dictates whether or not the game is actually fun.

[bloodnor]

i miss the currently online counter we used to have in our clients.

ELO does seem broken at the moment

[Thuzel]

jeffsw6, on 08 June 2013 - 11:12 AM, said:

I don't think that is true. I think they are just so bad at game-play balance that their impressive efforts in graphics, audio, etc. are failing because they don't understand that game balance is what dictates whether or not the game is actually fun.

I think that too. It seems almost definite that their idea of "balance" isn't shared by a lot of us for one reason or another.

What makes me think they've been hedging their bets for the last 6 months has been that things haven't sped up nearly as much as they should. Based on when they started taking in serious money and hiring additional people, we should have seen content delivery and core gameplay improvements seriously accelerate around March. But we haven't, and not only that, but we've seen only a few indications that they will accelerate in the future.

Just look at the test server situation. This game has been in play for a year now (a little over actually), and we're still waiting on a decent test setup. Even with the monumental QA issues we've seen, which should have been an extreme motivator, we're still waiting.

[Tarrasque]

Personally, I see it this way -

The matchmaker matches based on total team Elo. So you could have a score of over 9000, but it doesn't matter. You're not being matched against other people with a similar score, you're put with a random set of other players, then the matchmaker finds another 8 players who's score is within a threshold.

You could, and definitely should be getting paired with any and everyone who plays this game. Seeing the same players in matches is largely indicative of playing for a span of time that someone else is, and your launches are close enough that you get put into the same games together.

I'm definitely no scrub myself, and I see trials, new players, elite mechwarriors, and everything in between. I have noticed longer MM queues, but that can also be explained by the times that I tend to play.


TL/DR; It doesn't matter what your Elo score is - you're going to be matched against new players regardless.

[HarmAssassin]

The Devs have said that the goal of ELO is not to pair you with people of equal skill, but to balance everyone win/loss and Kill/death ratios to 50/50. That means one of two things will happen.

The higher your ELO the better the players you'll be placed against, and the worse the players will be that you are teamed up with to bring your kill death ratio and win loss ratios back to 1:1.

[Hauser]

HarmAssassin, on 08 June 2013 - 12:19 PM, said:

The Devs have said that the goal of ELO is not to pair you with people of equal skill, but to balance everyone win/loss and Kill/death ratios to 50/50. That means one of two things will happen.

The higher your ELO the better the players you'll be placed against, and the worse the players will be that you are teamed up with to bring your kill death ratio and win loss ratios back to 1:1.

Horse droppings.

If anything you've got cause and effect completely turned around. The purpose of a match maker that uses Elo is to keep the new and the experienced players apart. Everybody should face people at own level, resulting in a win/loss ratio that goes closer to 1:1.

If it ain't working, that's something to talk about. But don't go spreading that junk.

Edited by Hauser, 08 June 2013 - 12:34 PM.

[Echo6]

theNemesis, on 08 June 2013 - 05:29 AM, said:

Is a KDR of 5.1 common? I don't mind admitting mine is below 1. I got killed a lot in the first few weeks w/few kills.

Its been coming up fast but i doubt I'll ever get near 5.

This guy probably has more than one account. His learning was probably done on another account. Cause gotta keep up images, ya know?

[ICEFANG13]

Nicholas Carlyle, on 08 June 2013 - 07:02 AM, said:

The fun thing here with this thread is...

NO ONE KNOWS ANYTHING.

Because PGI purposely hides all the numbers.

We have no idea whether the MM is working, we cannot see any meaningful statistics, we cannot see our ELO's.

We can't see the population.

We know nothing.

So all of you guys are blowing hot air pretending to have an idea of what is going on. But none of us do because PGI refuses to release anything because they are so damn scared of what might happen.

I do know that generally I have PUGs who don't seem to know where the mouse button is. Drives me crazy to see multiple mechs not break 100 damage. I understand it happens, but this its a little crazy to have 4 or more not break it in assault mechs.

Echo6, on 08 June 2013 - 01:38 PM, said:

This guy probably has more than one account. His learning was probably done on another account. Cause gotta keep up images, ya know?

I had a KDR of 15 when I piloted the ezmode Jenner-D in closed beta, and this is my only account. Just because someone has a high KDR doesn't mean its not legitimate. For instance, my KDR was at 10, but then with seismic and worthless SRMs, and HSR (which I do like, but its increased their power as mine, as in the firepower, has lowered), and people packing huge alphas and only useful close range missiles (SSRMs), my KDR is only 7 now. Back when SRMs were dangerous, I raised my KDR during the hero tournament from 9-10, and that was all solo play. I also prefer to play 8 mans, but 8 man is dead and so is my merc groups. Don't assume anything.

[Renthrak]

I don't know what's going on with the matchmaker either. As people have said, none of us really know.

Still, I speculate that a part of the problem is that the MM is unnecessarily complex. Matching players based on their exact skill level (ELO) is fine in theory, but in practice, it just doesn't seem to work that way. It takes so much work for the server to match based on ELO and team averages that we routinely see massive weight imbalances, as the MM is forced to eventually discard weight matching just to find enough players.

I gave the whole problem some thought the other day, and I came up with something that I think can help:
http://mwomercs.com/...eight-matching/

[Taemien]

Not showing a player count has nothing to do with being scared to show the numbers. No one shows the numbers. Blizzard has never shown the exact numbers. The only time they announced their numbers is when it was a marketing desicion, 9 million + is a good time to show it off.

When you have less than a million, its not as good, and its definitely not good to show exact numbers. There is legal and proprietary reasons behind it. The reason they showed it in CB was because if you went spouting off about it during, they could pursue litigation against you for it.

Its best not even to bring up why, or should, or anything about exact numbers. No one gives that info out. No one has given that info out, and no one will. If you want to be the first by all means, make your little online game and post the numbers. But don't expect PGI or anyone else to, there's no reason and no precedence.

Just because you think there's a good number to show why the match maker isn't working right, doesn't mean PGI doesn't see it. You're not some cool awesome genius that can run the numbers. They are quite capable of seeing whats going on and know what they want to do next with those numbers. Because at the end of the day, they are going to make the final decision and do what they want, not what we want or think. We're players, not businesspersons, and if we were into business, we'd be competition, not allies/help.

[ICEFANG13]

I ran over this the other day, and wondered what it said about Mech, and Mech Warrior, and MechWarrior Online. Its kinda cool and it is real data, although you may disagree its relevant, I remember looking stuff up on google before downloading this game.

http://www.google.co...0Warrior&cmpt=q

[Deathlike]

I don't think MM is strictly the issue. It's the players in the queues that are affecting this, and it's not their fault.

In an optimal world, there are plenty of people playing, using plenty of different mechs and chassis and everything is hunky dory.

The reality.. the meta is heavy/assault based (meaning more tonnage mismatches), people fed up/tired of MWO (for whatever reason, they are not playing the game), and the MM is put into a situation that it must "even out" the ELO of the players in the match. More people are playing Forumwarrior than playing the game, thus resulting in the situation that you are stuck with more players in trial mechs.. because the queues have more newbies than regular players. This trickle down effect is affecting the 8-man just as much.

If there was a new gameplay mode and/or some portion of CW to test, this dynamic would change. Right now, we're arguing over the MM for stuff that isn't going to be resolved... the lack of interest by the community. That is what we're all suffering from.

[Fate 6]

Chemie, on 08 June 2013 - 07:08 AM, said:

Needless to say with that many games, I have played my share of assault, heavies and lights. Not mediums because PGI killed role warefar. I could list off my stats by class since ELO started but that is not the point of my post. As stated, I gave this info to avoid flames about being low ELO. I am not low ELO but I have seen way too many newbs.

Take your pick

1. MM turned off
2. Good players have quit and queues are empty

I don't buy there enough experienced players running trials for Lolz. I can see that happening a bit not becoming a standing order.

I think it's number 2. I spend more time on the forums these days, and that's never a good thing when players are talking about the game more than playing it. It's not like LoL where I have to have a free hour to make sure I finish a game, MWO matches take 5-10 minutes so I should never have an issue of not enough time.

I don't even remember the last time I saw half the guys from DWAR on the TS3 server or in-game. Granted I didn't have TS3 installed on this computer for a couple weeks, but it was trending downwards before that and now the server only has a couple people on and neither are online in MWO.