Jump to content

Stats Study: Matchmaker Is Unfair

Balance

344 replies to this topic

#61 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 19 April 2017 - 07:54 PM

View PostGhogiel, on 19 April 2017 - 03:56 PM, said:

I expect that there would be quite a different out come depending on the source. Someone with a 5 W/L is going to put about a 20-25% difference on that stat compared to the T1 avg on every game.


You're only 8% of your teams value, approximately. So if you're a 5 W/L in pug queue (winning about 83% of your matches, or 5 out of every 6) but your team of 11 is all at a 1.0 (winning 1 out of 2 matches, or 50%) then your team average W/L is 52.7%.

Which is to say your ability to carry is heavily mitigated if you're playing with Team Proto-Fries.

The real problem I can see is if the MM can't re-shuffle teams and players right up to the point it launchers.

So the MM is trying to feed people into a virtual lobby to make a match. It gets the first 4 people in, 2 and 2, pretty easily. Very well matched. Then however it's got 3 players readying up, 2 good, 1 bad. So now you've got 3 people on 1 team, 4 on the other. The team of 4 has a lower average PSR because it's got 3 averages and a spud and the team of 3 has 3 averages. So then the next person to ready up is a rockstar. It puts him on the team with 4 because the average PSR is lower, making population 5 v 3 but because their average value is split 5 ways it dilutes the value of the new rockstar relative to the 3man. If you add 2 more averages to the 3man so the population # is equal the PSR is off but it doesn't see that right now because it's got 1000, 1000, 1,000, 600, 2000 (average of 1125) vs 1050, 1050, 1150 (avg 1083). So supposing you get some more players, all pretty close to average, you end up with a 11 v 9 virtual lobby that's going to grab whoever is in queue longest and approximately fits tonnage and drop - without looking at PSR at all, meaning the odds are good that both sides will have at least 1-3 potatoes, 6-9 averages but only 1 side will have 1 or 2 rockstars.

etc. etc. Add to that its attempts to balance tonnage and the strong probability that the last 2-4 players are filled it by whoever is waiting longest and approximately fits tonnage and....

yeah.

If the virtual lobby can not shuffle sides then the trickle nature of new players becoming available combined with the need to match weight classes combined with the widening net as waits continue means there's absolutely room for brutally mismatched teams.

Also, because good players are correspondingly stronger in groups, 3 good players and a mix of spud and average on one side and 1 good player with 11 spud and average players may look pretty similar on PSR but strongly favor the 3 good players in performance. Once you max out T1 you're not 'heavier' in the MM; so Proton + Heimdelight + Edmeister is considered 'equal' to PhoenixFire + El Bandito + drunkblackstar, assuming everyone has pretty much maxed out their T1 XP bar.

Compare their W/L and KDR of them though? Pfft.

A thicker population would largely fix this however as the MM would have a well balanced enough pool for skill and weight classes to fill with. If the population isn't there and it has to trickle-fill, the sides will be unbalanced if it can't re-shuffle before launch.

You could create a TrueSkill style system that ranks you based on your success with individual chassis, even weapons and builds. Even down to that detail vs the population on the other team; you could create a very granular system with a prediction model at launch comparable to what most casinos use for sports events.

However if it's trickle-filling population and can't shuffle it then it's still going to end up unbalanced often enough that it wouldn't be worth the effort.

#62 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 19 April 2017 - 09:14 PM

I've done a similar study here: https://mwomercs.com...is-of-the-12-0/


If you want to go into more detail, I have a spreadsheet built that does all the analysis, all you need to do is input data (player names on each side, what mechs, which side won and I suppose what the score was in kills), and it does everything else automatically. Lemme know if you're interested in collecting data for it.

#63 Trollfeed

    Member

  • PipPipPipPipPipPip
  • 328 posts

Posted 19 April 2017 - 09:24 PM

Cooperation is a very big force multiplier in multiplayer games (doh!). Better players will capitalize on plays you make, like flanking for example while potatoes will bury themselves underground and let you die.

All those actions will show themselves on your stats, but game companies appear to be very reluctant to use them. Usually we see it in games that have real competitive scenes. A decent enoug matchmaker isn't that hard to implement, it just seems that they don't see any need for one.

Edited by Trollfeed, 19 April 2017 - 09:29 PM.


#64 Savage Wolf

    Member

  • PipPipPipPipPipPipPipPip
  • The Wolf
  • The Wolf
  • 1,323 posts
  • LocationÅrhus, Denmark

Posted 19 April 2017 - 09:39 PM

View PostSilentFenris, on 19 April 2017 - 11:34 AM, said:

I didn't say anywhere in my post that a Win/Loss of 1.00 is "good" or "bad". I did say poor QUALITY of matches results by having the Matchmaker attempt to maintain a 1.00 ratio on players.

And poor quality matches aren't bad? No matter what how you wish to exactly describe it, it would be nice with an explination as to why that should be the result? And what you would prefer.

View PostSilentFenris, on 19 April 2017 - 11:34 AM, said:

I don't subscribe to your "deathspiral" theory for 12-0 matches. 12-3 and 12-4 losses seem more plausible to account for one team overpowering as the weaker team begins loosing mechs. Even a moderate amount of cooperation and focus-fire should in at least 1 or 2 casualties on the Winning team. To achieve a 12-0 the Loosing team has to make a major blunder, typically by spreading out so their teamates can be picked off a few at a time and doing minimal damage before their mech is reduced to a smoking wreck.

The occational mistake by one team that coincides with a good move by the other can easily lead to 12-0 matches. They are still rare, but happens. Even seen them with the losing side having decent damage numbers. Often these matches also go super fast, the losing team crumbling before they can say WTF happened.
It's just the nature of the mechanics.

#65 Appogee

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 10,966 posts
  • LocationOn planet Tukayyid, celebrating victory

Posted 20 April 2017 - 12:58 AM

View PostSavage Wolf, on 19 April 2017 - 09:43 AM, said:

Uhm... what you describe there is close to perfect. With a perfect matchmaker everyone should average about 1 in W/L ratio.

No, it's the opposite of perfect.

A perfect matchmaker would have me hovering around 1:1 Win:Loss ratio all the time ... not ramping up to 1.1 with a run of easy wins, only to then have a run of impossible carry missions forcing me back down to 1.0.

#66 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:03 AM

View PostTarogato, on 19 April 2017 - 09:14 PM, said:

I've done a similar study here: https://mwomercs.com...is-of-the-12-0/


If you want to go into more detail, I have a spreadsheet built that does all the analysis, all you need to do is input data (player names on each side, what mechs, which side won and I suppose what the score was in kills), and it does everything else automatically. Lemme know if you're interested in collecting data for it.

Great minds think alikePosted Image

As far as I understand, you collected data, had roughly the same results, but a bit different conclusions (you think that the problem is in blindly following PSR).

What is intriguing for me is that you've showed that there was almost always an advantage on the winning side in terms of WL, KD and MS, but that difference is not so huge as I would expected. I quickly checked your numbers and in some cases the spread was big, in others - very narrow.

By the way, how did you get the info from the leaderboard? Is there any automatical way to do that?

#67 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 20 April 2017 - 01:06 AM

View Postdrunkblackstar, on 20 April 2017 - 01:03 AM, said:

By the way, how did you get the info from the leaderboard? Is there any automatical way to do that?


Yes, I have a spreadsheet that averages all of the leaderboard seasons and just looks up players when you paste them in.

#68 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:12 AM

View PostTarogato, on 20 April 2017 - 01:06 AM, said:

Yes, I have a spreadsheet that averages all of the leaderboard seasons and just looks up players when you paste them in.

wow, it looks like space program to me:)

I've added a link to your study in my OP.

So you don't think that the inbalance in teams is intentional. Your guess is that it's a mistake of the PSR system?

Edited by drunkblackstar, 20 April 2017 - 01:13 AM.


#69 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 20 April 2017 - 01:19 AM

View Postdrunkblackstar, on 20 April 2017 - 01:12 AM, said:

wow, it looks like space program to me:)

I've added a link to your study in my OP.

So you don't think that the inbalance in teams is intentional. Your guess is that it's a mistake of the PSR system?


Yes. There's absolutely no reason to make matches intentionally imbalanced. What would be the advantages of such a system?

It is also not possible to have the matchmaker try to force you toward a 1.0 WLR. For instance, like many people claim... putting a high performing player on a team that is destined to lose. But as a result of that, matchmaker is destining the 11 other players to lose, who already are underperforming, so it's pushing them even further beneath a 1.0 WLR. It just doesn't work, despite how many people seeming to think that the system out there to get them and work against them. You don't see low skill players raving about how well the matchmaker treats them, handing them easy wins all the time.

#70 PhoenixFire55

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 5,725 posts
  • LocationSt.Petersburg / Outreach

Posted 20 April 2017 - 01:24 AM

View PostTarogato, on 20 April 2017 - 01:19 AM, said:

Yes. There's absolutely no reason to make matches intentionally imbalanced. What would be the advantages of such a system?


Guaranteeing ~1.0 W/L for the majority of the playerbase, thus creating an illusion of a properly working MM.

#71 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:27 AM

View PostTarogato, on 20 April 2017 - 01:19 AM, said:

Yes. There's absolutely no reason to make matches intentionally imbalanced. What would be the advantages of such a system?
I agree that at first it seems absurd and the sort of a conspiracy theory.

But on a second approach there could be some logic used in online gaming marketing. This system gives the opportunity to win to everybody. Maybe the guys in marketing think that "bad players" would lose constantly and, as a result, they will leave the game. This system gives them the chance to be paired with the "winning team" and to win. As a drawback it spoils the game for everybody else time after time.

Edited by drunkblackstar, 20 April 2017 - 01:27 AM.


#72 PhoenixFire55

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 5,725 posts
  • LocationSt.Petersburg / Outreach

Posted 20 April 2017 - 01:28 AM

View PostTarogato, on 20 April 2017 - 01:19 AM, said:

It is also not possible to have the matchmaker try to force you toward a 1.0 WLR. For instance, like many people claim... putting a high performing player on a team that is destined to lose. But as a result of that, matchmaker is destining the 11 other players to lose, who already are underperforming, so it's pushing them even further beneath a 1.0 WLR. It just doesn't work, despite how many people seeming to think that the system out there to get them and work against them. You don't see low skill players raving about how well the matchmaker treats them, handing them easy wins all the time.


It does work. Since the amount of players who win is same as the amount of players who lose for every single match, you can simply put everyone who won their previous match into a "destined-to-lose" team for the next one and visa versa. The amount of players who "carry", i.e. have enough impact on the match to overcome the "destined-to-lose" part is negligible compared to the total amount of players. You can always find enough bads who randomly won a T5 match to counter said "carry".

#73 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:29 AM

View PostTarogato, on 20 April 2017 - 01:19 AM, said:

For instance, like many people claim... putting a high performing player on a team that is destined to lose. But as a result of that, matchmaker is destining the 11 other players to lose, who already are underperforming, so it's pushing them even further beneath a 1.0 WLR.

This system guarantees the winning and the losing for everybody. It doesn't aimed against "good players". The "good ones" also guaranteed to have victories:)

#74 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 20 April 2017 - 01:31 AM

View PostPhoenixFire55, on 20 April 2017 - 01:24 AM, said:

Guaranteeing ~1.0 W/L for the majority of the playerbase, thus creating an illusion of a properly working MM.

View Postdrunkblackstar, on 20 April 2017 - 01:27 AM, said:

I agree that at first it seems absurd and the sort of a conspiracy theory.

But on a second approach there could be some logic used in online gaming marketing. This system gives the opportunity to win to everybody. Maybe the guys in marketing think that "bad players" would lose constantly and, as a result, they will leave the game. This system gives them the chance to be paired with the "winning team" and to win. As a drawback it spoils the game for everybody else time after time.


How would you design a matchmaker that does this though? Every time you put a high performing player on a team, he sways the odds of that team further toward winning. You have to stack players on his team that have a high chance of losing in order for that high-performer to lose the match. And if you stacking loss-destined players to incite a loss, all you're doing is making their already bad WLRs worse. I don't see how such a system could even work. In a 1v1 scenario, yes, it works. But this is not 1v1.

#75 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:37 AM

View PostTarogato, on 20 April 2017 - 01:31 AM, said:

How would you design a matchmaker that does this though? Every time you put a high performing player on a team, he sways the odds of that team further toward winning. You have to stack players on his team that have a high chance of losing in order for that high-performer to lose the match. And if you stacking loss-destined players to incite a loss, all you're doing is making their already bad WLRs worse. I don't see how such a system could even work. In a 1v1 scenario, yes, it works. But this is not 1v1.

It seems to be rather simple to build.

You need to watch 2 variables to do that:

1) Players actual performance (easily tracked by stats in leaderboard)

2) Players recent losing or winning data (PGI apparantly have it).

There are good and a bad ones who have won "too much", and good and bad ones who lost "too much". The thing is how to mix them properly after that.

#76 PhoenixFire55

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 5,725 posts
  • LocationSt.Petersburg / Outreach

Posted 20 April 2017 - 01:38 AM

View PostTarogato, on 20 April 2017 - 01:31 AM, said:

How would you design a matchmaker that does this though? Every time you put a high performing player on a team, he sways the odds of that team further toward winning. You have to stack players on his team that have a high chance of losing in order for that high-performer to lose the match. And if you stacking loss-destined players to incite a loss, all you're doing is making their already bad WLRs worse. I don't see how such a system could even work. In a 1v1 scenario, yes, it works. But this is not 1v1.


The amount of "high-performing" players is negligible compared to total population. For each such player there are hundreds of mediocre and bad players. Same way there are far more matches between mediocre/bad players than there are matches with good players, i.e. there are always plenty of bads who won their previous match and plenty of bads who lost their previous match (you can use whatever criteria here, one previous match, several matches, total W/L, etc.), enough to team up your high performer with bads who were winning enough just before against mediocre ones who were on a losing streak prior.

#77 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 20 April 2017 - 01:41 AM

Sounds a bit sophisticated, but how would you explain the losing or winning streaks?

You can win 10 matches in a row easily, then you get 10 stomps defeats. How is that? Is it a coincidence?

Edited by drunkblackstar, 20 April 2017 - 01:42 AM.


#78 Widowmaker1981

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Widow Maker
  • The Widow Maker
  • 5,031 posts
  • LocationAt the other end of the pretty lights.

Posted 20 April 2017 - 01:45 AM

View PostAppogee, on 19 April 2017 - 09:27 AM, said:

I know it's weird, but it feels to me like win:loss ratio is still being used in the background to allocate people to teams.

I say that because my win:loss ratio cycles up to 1.1 then down to 1.01. Up to the top of that range, then down to the bottom again.

And it's been doing that for two years. It's as if when I get to 1.1 some kind of switch kicks in and I get a run of awful teams that not even Proton could carry.

I have no proof for this, other than that my win:loss stat seems like it hasn't behaved in a random fashion over the course of thousands of matches.


I dont think thats right - i have a 2.29 W/L this season, and i have pretty much not played group queue at all (maybe 10 games or so). I cant look at overall win loss because i do play in groups a lot, but it definitely does not feel like MM is 'trying' to keep me at 1.00.

Im pretty sure the issue is entirely caused by the idiotic upward biased tier system, meaning, as far as MM is concerned, some random baddie who plays a decent amount is considered Protons equal.. the range of abilities in T1 is incredible - id put myself in the top 5-10%, and im not even in the same league as the best players.

Matchmaker only looks at tiers, the tier system is junk. Put junk in, get junk out.

#79 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 20 April 2017 - 01:52 AM

View PostPhoenixFire55, on 20 April 2017 - 01:38 AM, said:

The amount of "high-performing" players is negligible compared to total population. For each such player there are hundreds of mediocre and bad players. Same way there are far more matches between mediocre/bad players than there are matches with good players, i.e. there are always plenty of bads who won their previous match and plenty of bads who lost their previous match (you can use whatever criteria here, one previous match, several matches, total W/L, etc.), enough to team up your high performer with bads who were winning enough just before against mediocre ones who were on a losing streak prior.


So then why do you often get teams stacked with all the best players on one side, and nobodies on the other? Because if you really think that is how the matchmaker works (tries to force people toward 1.0), than it is failing even worse than the one we actually have (which is trying to match players of equal skill.)

#80 Flitzomat

    Member

  • PipPipPipPipPipPipPipPip
  • WC 2018 Top 12 Qualifier
  • WC 2018 Top 12 Qualifier
  • 1,108 posts
  • Location@ the bowling alley

Posted 20 April 2017 - 01:58 AM

The better the player the more often he starts the death-spiral in his own favour.

That´s why no system will ever achieve 1.0 W/L for all

P.s. Good work on the stats. I run my own little thing and I was wondering if there is something like this for the Leaderboards. Would be real nice to apply filters but I have no capabilities whatsoever to program smth. like this.





8 user(s) are reading this topic

0 members, 8 guests, 0 anonymous users