67 replies to this topic

[Flak Kannon]

HI Yellonet,

I have played with you quite a bit.

You know, its really a crap shoot. Even when a team is LOADED with players I know to be good players, things can spiral out of control quickly. Say.. that Kodiak that went just 100 meters too far forward, and gets crushed, your team is down 100 tons.. and that could be the difference.



This game can snowball soo very fast. I would say that 75% of the teams that loose the first 2 mechs of a match loose that match, regardless of how many Tier 1 players are on the team.


Losses will happen. Period. I had a streak yesterday of 8 loses in 10 matches. I even switched to my best mech and couldn't stem the tide.


And lastly,

...I often buy new mechs. I just bought the entire Ebon, Orion IIC, and Warhawk lines. All of them are 'Locked', 'basic'ed', 'slow', sluggish mechs in which I have NO experience.

I am Tier 1, but in that mech, there isn't a comparison between me in my new, 'locked','basic'ed' Orion IIC, and a pilot in his Kodiak 3, or Ebon Jag fully 'Mastered', that has played 200-500 matches in it. I don't have the same skills yet, or the mechs abilities unlocked...My skill is constant, but the percentage dif. in my ability in a new mech and one that's unlocked fully, mastered, and in which I have alot of experience is truly night and day at higher Tier levels.

r

Edited by Flak Kannon, 06 October 2016 - 03:54 PM.

[Y E O N N E]

Wintersdark, on 06 October 2016 - 11:31 AM, said:

TLDR: The MM doesn't "try" to do anything; it doesn't add players to counterbalance other players, it's really simple. But low pops and a ill thought XP bar rating system gets us what it gets us.

Source?

[Wintersdark]

Yeonne Greene, on 06 October 2016 - 03:45 PM, said:

Source?

Karl Berg, who wrote it. He clearly outlined step by step how the MM algorithm works in his megathread. I'm on my phone, so I'm not going to search it, but he discussed the MM in detail there. Russ also confirmed some parts in a town hall (third? Don't remember which) with regards to how it uses your specific psr rating, taking the nearest available players first, then stretching the release valves over time.

To find, search for algorithm in his thread (paging Karl Berg...)

In off topic

[Sjorpha]

Mole, on 06 October 2016 - 07:08 AM, said:

I hate this response. It's statistically impossible that one player could be dragging down all 11 other members of his team so hard that he causes them to repeatedly lose. If I can witness teams that have one player that does 1000+ damage and gets 7 kills but his team still loses and I can also witness teams win while they have one or more disconnects or AFKs then this argument holds no water.

In a system of repeated experiments where every variable except 1 is random, the constant variable will dominate over time even if the random variables dominates every single repetition of the experiment.

In other words, if you've played for long enough time then you are the dominant factor behind your win ratio because of the above. Looking at single matches and observing that the other players (random variables) dominate specific results decreases in relevance towards zero the more matches you play.

This is the basic principle that most match making systems is based on, PSR is a hybrid but it relies on this logic as well.

Edited by Sjorpha, 06 October 2016 - 04:26 PM.

[Wintersdark]

I believe the "the matchmaker tries to..." Urban legends comes from the Elo days when people felt the Matchmaker would try to make you lose if you'd been winning a lot, because they'd heard Elo was intended to move people to where they'd have a 1:1 win loss ratio.

That of course was somewhat of a logic error born of narcissism. The matchmaker doesn't care about individual players nor try to accomplish anything with them. It just tries to match teams as evenly as possible. People like to believe the MM is doing things to them though, as if they are special in some way and not just another player to cram into a team. There's 24 players, if the MM had some overarching goal as to setting people up to succeed or fail, it would be monstrously overcomplicated for no good end.

Of course that is to low player counts, teams CANT be evenly matched. Elo took that into account in how ratings where adjusted, but PSR does not.

Edited by Wintersdark, 06 October 2016 - 04:31 PM.

[Sjorpha]

Wintersdark, on 06 October 2016 - 04:31 PM, said:

That of course was somewhat of a logic error born of narcissism. The matchmaker doesn't care about individual players nor try to accomplish anything with them. It just tries to match teams as evenly as possible. People like to believe the MM is doing things to them though, as if they are special in some way and not just another player to cram into a team. There's 24 players, if the MM had some overarching goal as to setting people up to succeed or fail, it would be monstrously overcomplicated for no good end.

Yeah, and the truth is the matchmaker doesn't know anything about you except your current PSR. It doesn't know your W/L ratio, it doesn't know how many matches in a row you've lost or won, it doesn't know any other stat on your stats page. It doesn't know, and it doesn't care.

All it knows is your current PSR, how to match that up with others within it's given constraints, and then how to calculate an increase or decrease to your PSR after the match. That's it, there is nothing else, everything else you think might be there is in your head.

Edited by Sjorpha, 06 October 2016 - 04:44 PM.

[razenWing]

Are we still talking about the conspiracy of the MM when I literally just explained the math behind all the fortune/misfortune?

The game is FIFO, the player base PSR is bell curve. The end.

You can literally derive ALL conclusions from those 2 confirm facts alone. I don't understand why people like to ignore these 2 obvious facts and have 40 other nonsense posts about conspiracies. Geez man...

[Y E O N N E]

Wintersdark, on 06 October 2016 - 04:20 PM, said:

Karl Berg, who wrote it. He clearly outlined step by step how the MM algorithm works in his megathread. I'm on my phone, so I'm not going to search it, but he discussed the MM in detail there. Russ also confirmed some parts in a town hall (third? Don't remember which) with regards to how it uses your specific psr rating, taking the nearest available players first, then stretching the release valves over time.

To find, search for algorithm in his thread (paging Karl Berg...)

In off topic

That's right in line with how I thought it worked.

If it's releasing valves over the duration of the search in order to fill the match, it will still create a game where you have players on a team who are known to carry harder, and if they go down before the other team's carry players go down, you have exactly the scenario I described in my previous post.

It's not necessarily an intentional thing, but that's what the mechanics of the game result in.

[Mole]

I don't know about all these players complaining about terrible team after terrible team. I'm in tier 2 and all I ever play is quick play since FW has been dead to me since phase 2 and I get teams that are fairly decent and work together well over VOIP comms regularly. I speak and communicate often, though there are times where I don't because I just feel like being quiet. At any rate, it seems to me that I win almost as often as I lose with it leaning towards victory a little more often. My leaderboard page reflects this feeling with my win/loss ratio at or above 1.00 in all seasons except season 2, during which it was 0.88. My overall win/loss ratio for all time on my private stats page comes out to 1.08. I refuse to believe that one pilot out of 12 is consistently responsible for their team's win or loss, so I'm not going to try and sit here and tell you to git gud either. Perhaps it's a matter pf perception, and your win/loss ratios really aren't that bad?

Edited by Mole, 06 October 2016 - 07:46 PM.

[Wintersdark]

Yeonne Greene, on 06 October 2016 - 07:39 PM, said:

That's right in line with how I thought it worked.

If it's releasing valves over the duration of the search in order to fill the match, it will still create a game where you have players on a team who are known to carry harder, and if they go down before the other team's carry players go down, you have exactly the scenario I described in my previous post.

It's not necessarily an intentional thing, but that's what the mechanics of the game result in.

Yes and no. While that's possible, it's not controlled.

The MM doesn't "try to maintain an average". Say the seed is (arbitrary numbers, 0-1000 total range) 700. For the first minute, it's grabbing everyone within a tier range of that. Tier=1000/5=200, so 700±100. Its whoever is available and in the right weight classes (dunno if it starts at 3/3/3/3 or 4/4/4/4, and this changes all the time around events etc)... While it's possible this could take a nice even spread of higher and lower players, if there's nobody from 701-800, but 9 people at 605, you could well end up with 1x700(seed player), 9x605... At 1minute, it extends ±100, so now 500-900, it takes you at 801 and Mr.Potatoehead at 500.

Kind of a rough example, but my point is that there's no guarantee that a high ranked player will be "balanced" by a low ranked player, not that the MM will go out of its way to try and acheive that at all.

So, while it's taking people within an equal range (±x rating), it's not actually trying to maintain an average.

Then add to the mix the realities of our rating system in the first place.

So while you can interpret this as "carry harder" the MM doesn't "expect" anything of anyone or "try to" acheive anything beyond adding players within a given rating range.


Also, consider, if the seed player is 950 rating, it's very likely that the majority of players added to the match will be well below his ranking. It can only add +50, but can go down to 850(initial), 750 then 650.

Given its unlikely that players online will be evenly distributed in rating, that is a hell of a lot of random

[InspectorG]

Yellonet, on 06 October 2016 - 06:55 AM, said:

Now that I'm T1 it feels as though I've been in more bad teams than before, and thinking back it seems to me that the higher tier I've had, the larger percentage garbage games due to no comms and generally no team play.
And as I usually get into a match quickly (<30 seconds) I should be with high tierish players, right?

What's up with that?

Even though tiers are mostly XP, surely they should count for something?

CARRY HARDER

That is your privilege.

Back when PSR dropped, I was high T3 then T2. Matches were good. Teammates communicated. Teammates could aim. Teammates adjusted to enemy tactics. And the Lord said, it was good.

Then the bads leveled up...

And then the fall...

Matches became Rabbit Runs. Campsites were rampant. Teammates could not aim and used LBX at 600m...
And Puglania fell into a deep darkness...

The only way out is to redeem yourself and carry harder...

[Green Mamba]

Player base not large enough for tiers to work properly

[Jehofi]

FupDup, on 06 October 2016 - 11:12 AM, said:

I've never been able to understand the "you're the only constant" argument.

Because what you quoted is misleading.
It is as you late state yourself: "The only constant in your games are you!"

FupDup, on 06 October 2016 - 11:12 AM, said:

For one thing, if we accept the premise that Bob is the only constant in all of his losses, this means that, by logic, Bob is also the only constant in all of his victories. So, every loss that Bob experiences is only caused by Bob himself, implying that he's a baddie. But, this also means that every victory is caused solely by Bob and no one else, making him a God among men who can carry hard (really hard).

False, it is not soley caused but in the noise of other players you are the only thing constant. Your teammates average out and you skill remains over alot of games.

FupDup, on 06 October 2016 - 11:12 AM, said:

These two revelations together leave us with absolutely nothing more than what we started the argument with.

It does if you actually understand statistics.

Bob having a 55% win chance is affecting his team positivly, thus winning more and should thus rise in rank to get better opponents and teammates. As he is definitely better than his current average teammate.

Bob having a 45% win chance is affecting his team negativly, thus loosing more and should thus lose in rank to get worse opponents and teammates. As he is definitely worse than his current average teammate.

Bob having a 50% win chance is right where he should be. Until he learns a new trick or 2 ofc.

FupDup, on 06 October 2016 - 11:12 AM, said:

For another thing, if you argue that Bob is the only thing affecting the match, that means you're admitting that only a single person can influence the outcome of a match...there are 23 other persons in the match. If one person has influence on the match outcome, you can sure as heck bet that 23 people have a lot of influence as well.

The is the only one constantly affecting the outcome, looking at an long series of games the influence of his teammates averages out.

Maybe now you will understand the basic principle for rating skill and why "The only constant in your games are you!" is a completely correct statement.

[BluefireMW]

The funny thing, it's not just that your team mates start to get worse as if they can't even move, or hitting a not moving target.
No the best is, that the enemy team seems to shoot precisly and knows what they are doing. Shooting on weak spots know what a UAV is...

There are lots of players with little to know experience in this game and the matchmaker is able to put them all together in the own team.

Edited by BluefireMW, 07 October 2016 - 04:19 AM.

[MOBAjobg]

The Conspiracy Theory at its best, detected.

[Morggo]

I just take matches for what they are and work with that, I guess.
VERY random is the main takeaway I'll agree with in all this, but you're online with hundreds or several thousand folks, many that will be in your MM range so 12 randoms will be random is how I see it.

My worst run was a 26 straight loss streak, probably half were 0-12 to 2-12 stomps. No idea what happened that day but it was tough to keep going even for me through all that but ya do.

The one thing I will say has baffled me the past month or two (and several mention it in this thread)... the wife is wrapping up tier 5 and I'd say 50% - 70% of her games have pretty good comms, about third of the time a really good dropcaller. My tier 3 matches... tend to be very quiet. Like 10% - 30% any comm chatter at all, and once in a while a solid dropcaller.

I would have expected the opposite... T1 alts farming perhaps? Dunno but just... odd.

[Wil McCullough]

cant get worse than t5.

i swear some players actually play with one hand and one brain hemisphere only.

if i had a dollar for every player i've seen that barely moved 10 steps before trying to shoot lurms at targets 2.2km away, i'd be a rich man.

[Wintersdark]

Sjorpha, on 06 October 2016 - 04:41 PM, said:

Yeah, and the truth is the matchmaker doesn't know anything about you except your current PSR. It doesn't know your W/L ratio, it doesn't know how many matches in a row you've lost or won, it doesn't know any other stat on your stats page. It doesn't know, and it doesn't care.

All it knows is your current PSR, how to match that up with others within it's given constraints, and then how to calculate an increase or decrease to your PSR after the match. That's it, there is nothing else, everything else you think might be there is in your head.

Exactly.

This was also directly confirmed by the author of said matchmaker - that it only receives your rating and mech. Nothing else. It doesn't know or care.

It's a really simple system, despite what people want to believe. It tries to get matches with players at the same rating and appropriate classes as possible; with the caveat that populations make it's goal impossible most of the time. That's all it does, no more, no less.

Yet people insist on seeing patterns that aren't there, and (my personal favourite) believing the the matchmaker is building a match of 24 players specifically to accomplish something with them, personally, like somehow they're more important than the other 23 players. I'd argue that, in particular, is somewhat indicative of some serious personal issues.

Wil McCullough, on 07 October 2016 - 08:06 AM, said:

cant get worse than t5.

i swear some players actually play with one hand and one brain hemisphere only.

if i had a dollar for every player i've seen that barely moved 10 steps before trying to shoot lurms at targets 2.2km away, i'd be a rich man.

To be fair, T5 includes people who've installed the game, started it up, and have absolutely zero idea of how to play at all.

Got to expect that =)

[EgoSlayer]

Sjorpha, on 06 October 2016 - 04:41 PM, said:

Yeah, and the truth is the matchmaker doesn't know anything about you except your current PSR.

If only people would understand and acknowledge this truth it would end a lot of the discussions about the matchmaker.

Wintersdark, on 07 October 2016 - 04:12 PM, said:

Yet people insist on seeing patterns that aren't there, and (my personal favourite) believing the the matchmaker is building a match of 24 players specifically to accomplish something with them, personally, like somehow they're more important than the other 23 players. I'd argue that, in particular, is somewhat indicative of some serious personal issues.

Yep - if people followed this thought process through they would understand how unrealistic it is. Because for it to be true and making each match "personal" it would have to be doing that for not only the one player, but for the other 23 players in the match as well.

So the "victim" player who has been on too long of a win streak, needs to carry a group, etc, needs 23 other people that match specific the victim's criteria but that doesn't conflict with their individual personalized match criteria. That coupled with the "skill" rating/PSR, and mech class limitations would require something like 10,000+ concurrent players to create one single match. It just isn't happening folks.

[DrxAbstract]

El Bandito, on 06 October 2016 - 08:16 AM, said:

And so we are back to the part where people say "there is only one constant variable in every game one loses, (or wins)". And since I am that variable, I am pretty happy about my performance.

Still, when one spectates a teammate who seems alien to the concept of torso twisting in T1 matches, it really feels tragically comedic.

Reported for name & shame violation.