344 replies to this topic

[BLOOD WOLF]

Savage Wolf, on 24 April 2017 - 05:24 AM, said:

Which would naturally result in a W/L ratio of 1. Otherwise, what would you think the ideal W/L would look like?

so It doesn't follow from matching 12 people against an equally rated 12 people means if they played a series of 10 matches that the score would be 5:5.

The W/L would be whatever it happens to be.

[Almond Brown]

drunkblackstar, on 21 April 2017 - 06:41 AM, said:

Imagine cooking machine.

This cooking machine makes 12 biscuits. Sometimes 8 buscuits are burnt, sometimes 9, sometimes 10.

After 12 times you used it, it would be safe to say that this cooking machine makes burnt biscuits.

But instead you are confused. You feel that its important to answer several questions before stating that it is a burnt buscuits machine.

"Time of the day? Current voltage in the network? Dough used?"

There you go. You solved your own issue. How can 1 machine "burn some" and not all in any one instance. Obviously you have "never used" a biscuit cooking machine, thus lack the knowledge to implicate the machine as the source of the "burnt" biscuits. Well done...

[Savage Wolf]

BLOOD WOLF, on 24 April 2017 - 05:35 AM, said:

so It doesn't follow from matching 12 people against an equally rated 12 people means if they played a series of 10 matches that the score would be 5:5.

The W/L would be whatever it happens to be.

What is 5 divided by 5?

[MOBAjobg]

drunkblackstar, on 19 April 2017 - 09:06 AM, said:

...*snip

Care to do the calculation for this screenshot of a match that I've rated as the most imbalanced in my own experience.

[Too Much Love]

MOBAjobg, on 24 April 2017 - 07:02 AM, said:

Care to do the calculation for this screenshot of a match that I've rated as the most imbalanced in my own experience.

Yeah, why not.

These are results:

Avg W\L:
Winners - 1,43 Losers - 1,08

Avg K\D:
Winners - 2,03 Losers - 1,46

Avg MS:
Winners - 277 Losers - 243

As in the most samples the winners had all 3 variables advantage, they had higher WL, KD and MS. Besides, the differences were huge 32% in WL, 40% in KD and 14% in MS.

NB. I didn't find aHOI from the winners team in the leaderboard.

So yes, your intuition didn't lie, this match as the most matches in MWO was anbalanced.

[Dimento Graven]

Yeah, I just figured that name was changed due to it being against TOS or whatever, in my calcs which came out differently, I figured him as a 1.0 w/l ratio as if this was his first match ever played.

Yeah, yeah, probably not the 'right' way to do it, but the numbers I had weren't all that different from what drunkblackstar got.

So yeah, those teams did not have matched skills. The match had teams with very unbalanced skills.

Edited by Dimento Graven, 24 April 2017 - 08:01 AM.

[Too Much Love]

Dimento Graven, on 24 April 2017 - 08:01 AM, said:

Yeah, I just figured that name was changed due to it being against TOS or whatever, in my calcs which came out differently, I figured him as a 1.0 w/l ratio as if this was his first match ever played.

Yeah, yeah, probably not the 'right' way to do it, but the numbers I had weren't all that different from what drunkblackstar got.

So yeah, those teams did not have matched skills. The match had teams with very unbalanced skills.

You can compare our findings. I made my calculations in the initial spreadshit (№14)
https://docs.google....xejU/edit#gid=0

[Dimento Graven]

Dimento Graven, on 24 April 2017 - 08:01 AM, said:

Yeah, I just figured that name was changed due to it being against TOS or whatever, in my calcs which came out differently, I figured him as a 1.0 w/l ratio as if this was his first match ever played.

Yeah, yeah, probably not the 'right' way to do it, but the numbers I had weren't all that different from what drunkblackstar got.

So yeah, those teams did not have matched skills. The match had teams with very unbalanced skills.

No need, I figured out I typo'd 2 of the losers w/l ratios, which is where that difference came in.

If you assume a 1 for the w/l ratio for AH0I, it brings the winner's avg down .04, a small enough difference that, "meh"...

You and I, and I would hope most reasonable people, will agree, MM could have done better had PSR been a bit more refined.

I mean c'mon, one side had only one guy with a W/L ratio of under 1, while the other team had 5 guys who lose more often than win...

It was MORE than 50/50 that the team with the least amount of skilled players was going to lose...

Edited by Dimento Graven, 24 April 2017 - 08:11 AM.

[BLOOD WOLF]

Savage Wolf, on 24 April 2017 - 05:41 AM, said:

What is 5 divided by 5?

that's division. your thinking of 5/5. I said 5:5, which is a ratio.

Edited by BLOOD WOLF, 24 April 2017 - 10:43 AM.

[Tarogato]

BLOOD WOLF, on 24 April 2017 - 10:43 AM, said:

that's division. your thinking of 5/5. I said 5:5, which is a ratio.

Except that you can express it in terms of one team's wins and losses.

You might have Team1:Team2 as 5:5

but you also have Team 1 W/L as 5/5 (or Team 2 for that matter, since they went 5-5 in your example)

BLOOD WOLF, on 24 April 2017 - 05:35 AM, said:

so It doesn't follow from matching 12 people against an equally rated 12 people means if they played a series of 10 matches that the score would be 5:5.

The W/L would be whatever it happens to be.

Actually, if matchmaker was perfection, then yes, the WLRs should be 1.0. It's theoretical, but it's also the ideal, the goal, and the expected result. Any difference from 1.0 simply represents the error in the system. The error is inherent and unavoidable because players are human creatures, but it's a divergence from the theoretical/predicted model nonetheless.

[BLOOD WOLF]

Tarogato, on 24 April 2017 - 11:07 AM, said:

Except that you can express it in terms of one team's wins and losses.

You might have Team1:Team2 as 5:5

but you also have Team 1 W/L as 5/5 (or Team 2 for that matter, since they went 5-5 in your example)





Actually, if matchmaker was perfection, then yes, the WLRs should be 1.0. It's theoretical, but it's also the ideal, the goal, and the expected result. Any difference from 1.0 simply represents the error in the system. The error is inherent and unavoidable because players are human creatures, but it's a divergence from the theoretical/predicted model nonetheless.

not really. You can never account for the variables in this game. every action a team makes would lead it to either win or lose.

So thats 2 equal teams with a 24 and each players is equally skilled. So if the end matchscores were perfect and unchanging. Lets also say that the greater matchcsores determines who wins or loses. Its going to be 12(a)=12(a). So your saying by lieu of a perfect MM, that the actions by both teams would always be the same. 12(100)=(12(100). whatever 100 actions or moves that team made, no matter what they did the matches would be 1:1.

Even a perfect MM doesn't guarantee results. Even theoretical mathematics doesn't get you anywhere, because even a=action doesn't tell me if the shot fired even hit. or the movement left to b4 was a good idea. Those moves can not be accounted for and the MM can not predict what people would do. in theory I can create some perfect scenario, but finding it in the real world is highly unlikely.

Edited by BLOOD WOLF, 24 April 2017 - 11:28 AM.

[Ghogiel]

drunkblackstar, on 22 April 2017 - 02:59 AM, said:

Read the original post.

People have tried to explain it to you several times already, I don't know if you are just trolling or a bit stupid, but one last time. The MM uses PSR to match teams. You MUST know the PSR of all the players in the matches so you can see if the MM is making the teams even, or there is no way to tell.

[Dimento Graven]

Ghogiel, on 24 April 2017 - 02:48 PM, said:

People have tried to explain it to you several times already, I don't know if you are just trolling or a bit stupid, but one last time. The MM uses PSR to match teams. You MUST know the PSR of all the players in the matches so you can see if the MM is making the teams even, or there is no way to tell.

Yes, we know what people have said, however, PSR is flawed.

Even PGI has said that even an average player will eventually grind their way to Tier 1.

Why? Because PSR is "win biased". You can win 10 matches, and lose 10 matches, and still have a NET POSITIVE gain in PSR ranking because the penalty for losing matches is so much less than what's awarded when winning one.

If the only values the MM is using when assembling matches is PSR Tier and 'mech weight class, the teams in our matches won't always, and probably only rarely ever be, evenly skilled.

People of Tier 1 having W/L ratios of less than 1 is possible with the current system, and if your W/L ratio is less than 1, it means you're losing more than you're winning, and that brings to question the skill you bring to the team.

Because PSR is flawed, MM can't do its intended job well.

But at the very root of it, because we have such low populations, the thresholds for Tier rankings in assembling teams is wider than originally intended.

Both population and the PSR system need to be fixed for MM to be able to do its job.

[Ghogiel]

Dimento Graven, on 24 April 2017 - 03:01 PM, said:

Yes, we know what people have said, however, PSR is flawed.

Even PGI has said that even an average player will eventually grind their way to Tier 1.

Why? Because PSR is "win biased". You can win 10 matches, and lose 10 matches, and still have a NET POSITIVE gain in PSR ranking because the penalty for losing matches is so much less than what's awarded when winning one.

If the only values the MM is using when assembling matches is PSR Tier and 'mech weight class, the teams in our matches won't always, and probably only rarely ever be, evenly skilled.

People of Tier 1 having W/L ratios of less than 1 is possible with the current system, and if your W/L ratio is less than 1, it means you're losing more than you're winning, and that brings to question the skill you bring to the team.

Because PSR is flawed, MM can't do its intended job well.

But at the very root of it, because we have such low populations, the thresholds for Tier rankings in assembling teams is wider than originally intended.

Both population and the PSR system need to be fixed for MM to be able to do its job.

PSR being garbage doesn't matter to the MM or to the OPs assertions that the MM is making unfair matches.

It doesn't matter if it's Elo, trueskill, PSR or anything else, what matters is knowing if the matchmaker is making an even distribution between the teams of whatever variable it is, in this case it's PSR. But the OP doesn't know any PSRs, is not tracking the PSRs, so can't know if the teams are equal according to the match maker.

The match maker might be doing it's job fine for all we know, just because PSR is garbage doesn't mean the PSR spread isn't good.

[Dimento Graven]

Ghogiel, on 24 April 2017 - 03:53 PM, said:

PSR being garbage doesn't matter to the MM or to the OPs assertions that the MM is making unfair matches.

It doesn't matter if it's Elo, trueskill, PSR or anything else, what matters is knowing if the matchmaker is making an even distribution between the teams of whatever variable it is, in this case it's PSR. But the OP doesn't know any PSRs, is not tracking the PSRs, so can't know if the teams are equal according to the match maker.

The match maker might be doing it's job fine for all we know, just because PSR is garbage doesn't mean the PSR spread isn't good.

You missed the point.

PSR is currently insufficient to ensure that both sides of the battle are evenly matched, thus MM is at least occasionally failing one of its primary objectives, to create matches with balanced teams.

It's like being given an old chewed up yard stick to measure millimeters.

Yeah, you can do it, but you're not ever going to be 100% accurate...

[Tarogato]

Ghogiel, on 24 April 2017 - 03:53 PM, said:

PSR being garbage doesn't matter to the MM or to the OPs assertions that the MM is making unfair matches.

To me, that's like auditioning people for a choir based on how well they play an instrument. Sure, these are two extremely related and intertwined skills, and you may find a correlation, but it's still not fair.

Or matching casual pickup basketball by BMI. Sure, maybe the fit people would beat the overweight people, but what does BMI tell you about how good two people are at basketball if they have the same BMI, even if we know that playing basketball regularly can "improve" your BMI?


Point is, PSR does not equal actual skill level. It equals experience. Better players can fill their experience bars faster, and worse players can fill their experience bars by playing more, but at the end of the day both players have a full experience bar, they achieved the same goal by different means, and their actual skill differs greatly. The MM could put together the perfect match, where all 24 players are max Tier 1... identical PSR. And it can still be outrageously imbalanced. So checking PSR means nothing when it comes to evaluating the effectiveness or "fairness" of the matchmaker.

[Steel Claws]

What in life is fair. Should we hand out participation trophies.

Match maker takes the people that drop, tries to balance the mechs by weight, and find the best selection within a limited amount of time.

In effect you end up getting the luck of the draw balanced out for mech weights in a fairly short amount of time. Perhaps you would like to wait 10 minutes between drops while waiting for something a little more fair? Once upon a time it did that. Then people complained it took too long to get matched up. You suckers are never happy.

Edited by Steel Claws, 24 April 2017 - 05:43 PM.

[BLOOD WOLF]

Steel Claws, on 24 April 2017 - 05:39 PM, said:

What in life is fair. Should we hand out participation trophies.

Match maker takes the people that drop, tries to balance the mechs by weight, and find the best selection within a limited amount of time.

In effect you end up getting the luck of the draw balanced out for mech weights in a fairly short amount of time. Perhaps you would like to wait 10 minutes between drops while waiting for something a little more fair?

it accounts for elo as well, but yea the people complaining are the ones who lost the match and are crying that the other team got gold. So now they got to make up anything to feel better.

Some people just don't understand how a matchmaker works.

Edited by BLOOD WOLF, 24 April 2017 - 05:42 PM.

[AnimeFreak40K]

My understanding of how the Match Maker works...or, rather is supposed to work is that it is relatively fair/balanced across the board.

...at least on paper...

In practice, the Matchmaker ends up falling apart because of the population at any given time.

The reason why population is a factor is because certain hard-coded limiters that were put in place to ensure balance, such as Tier/Skill Level (the optimal setup is that there will be no more than 2 levels between the highest/lowest tiered players; a T4 player will never see a T1 or 2 player, for example) or Classes (2 Assaults on 1 team, means 2 Assaults on the other team, for example) are pulled back after a certain amount of time of searching for a match.

IIRC, the game should put folks into a match no matter what after about 2-3 minutes of searching (I could be wrong here), in which case, the MM is just doing what it can to find somebody...anybody to put a match together.

These things are not always apparent on the NA servers during peak-times, especially in the solo-queue...but are felt and very apparent (almost painfully so) on the Pacific Servers especially on off-peak times.

The reason I say that population is a driving factor for MM fails is two-fold:
1 - The player-base as a whole has stated repeatedly and made it quite clear to PGI that people are *NOT* interested in balanced matches. That is, if given the choice of "You can get in a match quickly *OR* you can get a balanced match, but you cannot have both", the population, as a whole, has unequivocally stated that they want fast matches. They do NOT want to wait. They will take 10 terribly balanced matches over waiting for 1 good one.
2 - With #1 in mind, and my description of how the Matchmaker is supposed to work, a limited population ensures that the 'release valves' to keep things nice and balanced are either opened sooner or just left open.

Another factor that the Matchmaker cannot account for, no matter what, is the snowball effect of what tends to happen in a game more often than not.
That is, two evenly matched teams both in skill-level and tonnage are duking it out. One person out on one of the teams is going to stagger first. It could be a lucky shot, poking a little too long, or any number of factors...the end result is that the team is now down one mech...and things are no longer balanced and the tide will shift as the other team takes advantage of the situation (12v12 turns to 11v12 then 10 or 9 v12 and so on). Yes, things can balance out the other way and the tide can turn...but by the time you get to the mid-game, the likelihood of that happening drops with a quickness.

TL;DR - The matchmaker isn't unfair per se (though it is designed to provide challenging matches), but limited population and the desire to have fast matches over balanced ones throws off the balancing algorithms. Also, the matchmaker can't account for the reality of a snowball effect between teams; whether they're balanced or not.
On a side note, everything I'm talking about is in regards to the solo queue. When talking group queue, all of the problems (especially population) are magnified, especially when dealing with 'large' groups (that is anything bigger than 4).

[vandalhooch]

Dimento Graven, on 23 April 2017 - 08:29 PM, said:

You've repeatedly told everyone who doesn't agree with you that their wrong, their numbers are somehow wrong, even though the numbers are gathered from the system, therefore their analysis are wrong, they don't have enough numbers, blah, blah, blah, blah, blah, blah, blah, BLAH...

I get that you don't actually understand how statistics work, but now you're just being a petulant child.

Quote

At the very least, the numbers from the stomps show that AT A MINIMUM, for those matches, MORE OFTEN THAN NOT, the winning side has higher performing players than the other.

Duh.

Quote

In these instances MM fails to provide truly balanced teams.

Are those types of imbalances more common than they should be? You can't possibly say because you don't even know how often they occur at at all.

Quote

Even with the hundred or so samplings provided by Tarogato and the OP, it was a very obvious thing to observe from the numbers.

I don't see how you can keep denying that, yet you do.

I'm not denying anything. I'm pointing out that you are claiming that the numbers support an assertion that they simply don't.

Quote

Gee, do I have to stick my hand in a thousand pots on stoves to understand that stoves can make things hot?

Or can do I get to stop when my hand is only mildly scalded and get to conclude, "Y'know, there might be something going on here..."

Your analogies are getting dumber and dumber.

Quote

Ok Mr Webster, please define it for us. I've used the most common definitions bandied about on the forums by just about everyone, up to and including, PGI employees when they speak on the subject, so, YOU define it.

Take a look at what Taragato did. He gave imbalance a mathematical definition. That's what you need. Once we have that, we can start to look at some actual data. You're the one claiming that the matches aren't balanced. So give us your definition of a balanced match.

Quote

I'm tired of attempting to get past the fingers in your ears and the loud "NYAH NYAH NYAH NYAH" coming from your mouth...

So you're a 'porn scientist' now? Cool!

But seriously the numbers were there, gathered from the leader boards. In a 12v12 I would say any number above 2 is significant, ie: 3 or more.

I could have sworn that's what we were seeing from the numbers.

What number above 2? 2 what? You can swear all you want, doesn't make it true.

Quote

Greater number of low skill/noobs in LRM boats, on PH against a number of greater skilled/experienced players in whatever 'mechs they happen to choose? You're not seriously asking that question are you?

I'm asking you to prove that there is no difference to the outcomes of teams like that on Polar Highlands versus HPG.

I'm not asking what your opinion is about how it will go, I want you to prove something.

Quote

That's damn near a dream seal clubbing scenario, the noobs/low skills won't stand a chance. You are of course assuming that all the experience players only brought flamer LOLbuilds or something, I take it.

So the imbalance you are complaining about is a team of Tier 1 killers versus a full team of Tier 4 noobies? Nice that you wanted to have a constructive conversation. No hyperbole there at all.

Quote

Maps have little bearing on how players apply their skills. Players can have favorite and least favorite maps, may know the terrain better, or what have you, but as far as affecting their ability to aim, pilot a 'mech, utilize situational awareness, get the most out of their 'mech, so on and so forth, the things that experienced/skilled players do oh so much better than noobs/low skill players, again, it'll be a difference that makes no difference.

Sounds like a great hypothesis. Now, get the data to support it.

Better yet, give us a screen shot of your profile stats showing that you have the exact same WLR on every single map.

Quote

Ignore maps as irrelevant because a good player doesn't suddenly forget how to aim, pilot his 'mech, or press the R key to target what he's shooting at just because he's on a certain map, while a bad player won't be able to consistently do these things on ANY map...

And in your mind all of these imbalanced matches involve teams at the extreme end of the skill spectrum? You don't want to look at any other levels of skill? So much for the discussion you claimed I refused to engage in, I guess.

Quote

Don't need it, the speed at which an average player moves up in rank is not actually relevant. The fact is PGI has admitted that you play enough games, unless you're doing unreasonably badly consistently, you're going to go up in rank.

The fact that you don't think speed of Tier rising matters just shows how very, very bad at data analysis you are. You are not good at this. Quit while you are behind.

Quote

The numbers appeared to show a general direction in which stomps occurred, where MM has created teams of lopsided skills.

No it didn't. No it didn't. No it didn't.

How many matches with lopsided teams did not result in a stomp? You don't know. You can't say that lopsided teams create more stomps because you don't even know what the background rate of stomps is.

You are terrible at this. Just stop.

Quote

That's the way the data was 'leaning', I dunno, what word do you want to use then?

Certainly not "trending." And since the data doesn't show the pattern you think is there, there isn't any word I would use to describe it because it isn't actually there.

Quote

And his data appears to show that when stomps occur MM has created teams of lopsided skill, ergo, it failed to make evenly matched teams.

And what do you make of the fact that 20% of the time the weaker team not only won but actually STOMPED the stronger team? Does that sound like Taragato's system is accurately identifying strong and weak teams?

Quote

Again Mr. Webster, you define it then.

You're the one claiming that matchmaker is failing to create it. I'm just asking you to define what it is. Right now it seems that an unbalanced match is anything that you think is unbalanced because . . . reasons?

Quote

I was using generally accepted terms, terms even PGI uses.

Great. Plug those generally accepted terms into your handy dandy text-based statistical analysis software then. You've got one of those, right?

Quote

Y'know, kind of like how we define stars: "a self-luminous gaseous spheroidal celestial body of great mass which produces energy by means of nuclear fusion reactions"

Great. Now tell me how you can tell the difference between a class-F star from a class-G star from a class-K star. Do you just go hot, medium and cold? What about class-A stars?

Quote

We don't have to count the number of hydrogen atoms before we can declare, "Yes, it's a star."...

"Yes, it's a match. . . "

"Is it a balanced match? How do you know? Does everyone agree that it's a balanced match? Why not?"

Your attempts at analogies are terrible. You don't know enough about any of these subjects to even be trying to use them to play "gotcha."

Quote

No, they don't. They figure the odds based on past performance. They have all sorts of statistics on the teams, the individual players, home vs away, temperature variances, time of day of the game, whether it's a weekday game vs. a weekend game. Sometimes the numbers they have makes NASA look like they're playing bingo...

Bwaaa haaa haaaaa.

Do point spreads stay the same from when the initial line is set? Care to explain why not?

Quote

So yeah, Vegas has what they feel is a 'reasonable certainty' of how a game is going to turn out and offers odds based on their outcome. As Vegas is more often right than wrong, and most gamblers ate betting their 'gut', Vegas makes pretty good money on it.

You are a statistical illiterate.

Quote

Me? When I go to Vegas, I stick to the best odds in the house: Blackjack and craps for me, sir.

Best odds in the house = lowest house advantage. All casino games have an inherent house advantage.

Quote

That's exactly how a professional client denier would be paid to respond.

'Yeah, which CO2 sources are raising the temperature of the earth? Is the coal based ones, or the ones from burning petroleum? Do we care about burning wood? Should we dived it all by the number of cows farting methane?'

So, in other words, you have absolutely no idea how to average a player's WLR and their damage per match to determine their "skill level" despite your claim that it would be better than what PGI does.

You are a pompous know-nothing. Got it.

Edited by vandalhooch, 24 April 2017 - 07:50 PM.