Jump to content

Psr Should Be Changed To Better Reflect Player Skill


71 replies to this topic

#41 MrMadguy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,308 posts

Posted 29 January 2016 - 03:04 AM

View PosttortuousGoddess, on 29 January 2016 - 02:50 AM, said:

And please don't break out into conspiracy theories like that, it's rude.

PSR is behaving way to similar to Microsoft TrueSkill. And this MM is patented and has to be licensed. TrueSkill - is the same old ELO. The only difference - is that rank change at the end of match is distributed among players according to their Match Score. So there are no hardcoded Match Score thresholds, that determine, whether your rank will increase, decrease or stay the same, when you Win or Lose. If system predicts result of match correctly and decides, that rank should not be changed - you will see "=", no matter how low your Match Score will be. Because X * 0 = 0 no matter, how big X is.

Edited by MrMadguy, 29 January 2016 - 03:05 AM.


#42 Ratpoison

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 851 posts

Posted 29 January 2016 - 03:24 AM

View PostMrMadguy, on 29 January 2016 - 03:04 AM, said:

PSR is behaving way to similar to Microsoft TrueSkill. And this MM is patented and has to be licensed. TrueSkill - is the same old ELO. The only difference - is that rank change at the end of match is distributed among players according to their Match Score. So there are no hardcoded Match Score thresholds, that determine, whether your rank will increase, decrease or stay the same, when you Win or Lose. If system predicts result of match correctly and decides, that rank should not be changed - you will see "=", no matter how low your Match Score will be. Because X * 0 = 0 no matter, how big X is.

So you think that since matchmaker behaves kind of like another matchmaker, that PGI is covering up their relation to it as well as a nasty legal battle? That doesn't sound at least a bit silly to you? Okay then...

#43 Mead

    Member

  • PipPipPipPipPipPip
  • Little Helper
  • Little Helper
  • 338 posts

Posted 29 January 2016 - 03:44 AM

One consequence of having a low score on a win be a decrease penalty is that no one will be inclined to capture anything when they could be shooting mechs instead. We'll have three Skirmish modes where the only distinguishing feature is different quantities of bars at the top-middle of the screen.

I sure wouldn't bother with it any more. It's already crappy enough to run around capping and get what is basically a slap in the face for helping land a win.

#44 MrMadguy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,308 posts

Posted 29 January 2016 - 04:09 AM

View PosttortuousGoddess, on 29 January 2016 - 03:24 AM, said:

So you think that since matchmaker behaves kind of like another matchmaker, that PGI is covering up their relation to it as well as a nasty legal battle? That doesn't sound at least a bit silly to you? Okay then...

I don't see other reason to hide MM details from us.

I just got sick of MM being completely inconsistent. For the past two weeks all matches have been extremely terrible, including yesterday. But today all of a sudden I started to win and having good matches. My question is - why can't I always have such quality of matches?

#45 Ratpoison

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 851 posts

Posted 29 January 2016 - 04:23 AM

View PostMrMadguy, on 29 January 2016 - 04:09 AM, said:

I don't see other reason to hide MM details from us.

I just got sick of MM being completely inconsistent. For the past two weeks all matches have been extremely terrible, including yesterday. But today all of a sudden I started to win and having good matches. My question is - why can't I always have such quality of matches?

Um, did you not think that maybe it's kept private so that they don't have everyone trying to game the matchmaker to do what they want? I promise PGI isn't out to get you. It will be okay.

#46 MrMadguy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,308 posts

Posted 29 January 2016 - 04:49 AM

View PosttortuousGoddess, on 29 January 2016 - 04:23 AM, said:

Um, did you not think that maybe it's kept private so that they don't have everyone trying to game the matchmaker to do what they want? I promise PGI isn't out to get you. It will be okay.

Game the MM? You mean intentionally performing badly to drop your rating? Do you understand, that if MM would work correctly, then how much you would win via dropping your rating - that much you would lose during trying to drop it. Law of reward conservation.

#47 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 29 January 2016 - 04:53 AM

View PostMrMadguy, on 29 January 2016 - 04:09 AM, said:

I don't see other reason to hide MM details from us.

I just got sick of MM being completely inconsistent. For the past two weeks all matches have been extremely terrible, including yesterday. But today all of a sudden I started to win and having good matches. My question is - why can't I always have such quality of matches?


You sort of answer it yourself there, if good matchmaking == you winning, then half the players will always have bad games... Posted Image

Seriously though, Matchmaking and Scoring are different things. Related but different. Afaic nothing is really hidden with the PSR, but details of the MM is largely not open to us. There in the actual matchmaker I am pretty sure there is room for improvement, the MM has a reputation for being streaky and gameable. No idea if that's statistically true or not, but that's the rep... in any case, it would be comforting if PGI could afford to be completely open with the MM code, like in open source models, that gives credibility to the robustness of the system.

Edited by Duke Nedo, 29 January 2016 - 04:53 AM.


#48 MrMadguy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,308 posts

Posted 29 January 2016 - 05:03 AM

Nope. Started to lose again. It always happens like that: 2-3 good matches, then lose streak again.
Posted Image

#49 MrJeffers

    Member

  • PipPipPipPipPipPipPip
  • 796 posts
  • LocationThe Rock

Posted 29 January 2016 - 10:21 AM

View PostMrMadguy, on 29 January 2016 - 03:04 AM, said:

PSR is behaving way to similar to Microsoft TrueSkill. And this MM is patented and has to be licensed. TrueSkill - is the same old ELO. The only difference - is that rank change at the end of match is distributed among players according to their Match Score. So there are no hardcoded Match Score thresholds, that determine, whether your rank will increase, decrease or stay the same, when you Win or Lose. If system predicts result of match correctly and decides, that rank should not be changed - you will see "=", no matter how low your Match Score will be. Because X * 0 = 0 no matter, how big X is.



PSR is nothing like elo or true skill and performs nothing like them. You are so blindly wrong it's not even funny. There is no prediction or relative performance to other players involved.
Your rank can't drop on a win, even if you are AFK the entire match. The only time that PSR drops is on a loss, but if you perform well enough on a loss you can get a slight increase. This is shown in Paul's graphic that has already been posted in this thread.

There are *fixed* hardcoded match score levels that determine the outcome. On a Loss a match score of less than 250 is a decrease. 251-400 is a neutral performance, and 401 is an increase.
On a win greater than 100 points is an increase but only Paul/PGI know the tiers of increase, but one could assume they are the same scale as the losses.

It in no way is performing any measure against other players like elo systems, it is completely unrelated and why it's referred to as a glorified XP bar because it's a fixed table of match score vs win/loss determines your point gain/loss in tier. That is the only criteria.

Edited by MrJeffers, 29 January 2016 - 10:24 AM.


#50 vandalhooch

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 891 posts

Posted 29 January 2016 - 10:35 AM

View PostKilo 40, on 28 January 2016 - 02:18 AM, said:


no no no....

I mean across both teams. the top six players, from both teams increase, while the lowest 6 across both teams decrease. so using your example, The 6 super guys with score 500-550 get uprated, while the 6 super guys with score 450-500 and The 6 noobs with score 20-30 stay the same and the 6 noobs with score 0-20 get downrated because they suck so much.

also your example sucks and isn't very realistic.


That is the easiest system you could think of to game. Players could intentionally do bad for a few matches in order to remain in a low Tier so that they can farm new players. We already have veteran players who take pride in their ability to game the system we have now. Your idea would make it even easier to troll and grief new players.

No thanks.

#51 MysticLink

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 214 posts

Posted 29 January 2016 - 10:49 AM

I like the system as is. I've seen better quality matches over all. Just keep in mind, you can't win them all, and you will be happy. Just try to win more then 50% of the games, but if you don't, it's not the end of the world. Of course more then half the population is going to be on the losing side of win/loss ratio given there is some people with much higher win loss ratios then normal. So if you on that side, just remember most people are.

#52 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 29 January 2016 - 11:48 AM

View PostLord0fHats, on 27 January 2016 - 11:33 PM, said:

I think the change that's really needed is that people should lose PSR even when their team wins if they do poorly enough.


How then do you propose to reward players performing the role of "forlorn hope"?


View PostXX Sulla XX, on 27 January 2016 - 11:45 PM, said:

You are exactly right. Being able to go down on a win is a must. For example having some one playing in a group and they make like 5 damage each game and the rest of the team is making 400-900 each game and they are winning. 5 damage person should be going down.


How do you adequately reward the player who decided to forgo killing the enemy by instead scouting, distracting, baiting, capping, etc. to get a win?

Edited by Mystere, 29 January 2016 - 12:05 PM.


#53 Triordinant

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,495 posts
  • LocationThe Dark Side of the Moon

Posted 29 January 2016 - 11:53 AM

They should wipe the slate clean and make everyone play 100 matches of Solaris 1 vs 1. PSR will be a lot more accurate at the end of that. Posted Image

#54 Mech Jager

    Member

  • PipPipPipPipPip
  • The Money Maker
  • The Money Maker
  • 122 posts

Posted 29 January 2016 - 12:04 PM

How about a 10 step system that everyone can see and game that is based totally off of individual performance while we keep tier system for matchmaking. If someone has a positive KDR after 100 matches, I want them on my team. If someone is continually getting great damage scores and very few wins, I do not need another assault sniper/lrm in the back row (or god forbid stock loadouts).

PSR is a rating based on 100s of games, and the bad games stick out more especially if the player really was an army of one.

I am also pretty hopeful that PSR uses more specific numbers than just 1-5 (like say 3.158) to be figured into the MM. This would mean everyone could be t1 and still be meaningfully separated by MM.

I think what the OP is saying is they want a badge they can show off to others over winning games.

#55 JernauM

    Member

  • PipPipPipPipPip
  • Survivor
  • Survivor
  • 132 posts

Posted 29 January 2016 - 12:48 PM

View PostMech Jager, on 29 January 2016 - 12:04 PM, said:

I think what the OP is saying is they want a badge they can show off to others over winning games.


You are incorrect. What I want is a matchmaker that reliably produces balanced matches. In this regard, the system we have now is going to degrade over time as the playerbase inexorably trends upward in tier and everyone is improperly grouped into the same matchmaking pool.

#56 wanderer

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 11,152 posts
  • LocationStomping around in a giant robot, of course.

Posted 29 January 2016 - 12:56 PM

You want PSR to tier better?

Make the score required to increase a tier bar higher as you go from Rank 5 -> 1. Likewise, increase the minimum score needed to prevent losing PSR, leaving the space between them the same.

And increase the -amount- of loss as well. If you turn in a T5-worthy "lose" result at T3, it should hurt you more than barely missing the cut.

Keeping your tier should be an ever more difficult struggle.

Edited by wanderer, 29 January 2016 - 12:58 PM.


#57 Pendaelose

    Member

  • PipPipPip
  • Elite Founder
  • Elite Founder
  • 69 posts

Posted 29 January 2016 - 12:57 PM

View PostMech Jager, on 29 January 2016 - 12:04 PM, said:

If someone has a positive KDR after 100 matches, I want them on my team.


I think you're really hitting something here, but KDR ratio is only important on the direct damage dealers. A good scout or a pilot who plans and communicates well can win the match without firing a shot.

More important than KDR, I would look at WLR. If you have WLR greater than 1 you're definitely doing something right. The greater the WLR the better you're doing it. You can be a great shooter, a great scout, a great bullet sponge... if you help your team win consistently you are doing something right, and I want you on my team. He're the real kicker... if you just plain suck, but you always party with your awesome friends that always carry your *** and the rest of the team, your going to have good WLR despite sucking... but you know what, if you always drop with your awesome friends... I still want you on my team! (btw, bring your friends!) Seriously though, if the group keeps "Joe" a good WLR then it must mean he doesn't drop without his friends.

It's important to note that rewarding ranking by WLR isn't the same as rewarding or penalizing for a single win or loss.

My opinion is that the win/loss ratio should be tracked for a fairly narrow window of time (1-2 weeks) with older games dropping off. Only your "current" WLR is important. So long as your ratio is greater than one, for each game you play you should get a small bump up in Ranking. If your ratio is negative, for each game you play you should get a small bump down. The more you play the more tolerance the system has for a single bad game.

The amount of bump received should be reflected by the ratio itself. Having a 1.01 WLR should give a minuscule bump, while having a 2.5 WLR should be a significant bump. To prevent a few games from creating excessive leapfrogging in ranks we can cap the rate of rank change.


I would also suggest that a player have a separate ranking for each weight class. The worlds best scout pilot may not be worth very much in an Atlas.

Edited by Pendaelose, 29 January 2016 - 01:11 PM.


#58 MrJeffers

    Member

  • PipPipPipPipPipPipPip
  • 796 posts
  • LocationThe Rock

Posted 29 January 2016 - 01:15 PM

View PostPendaelose, on 29 January 2016 - 12:57 PM, said:


I think you're really hitting something here, but KDR ratio is only important on the direct damage dealers. A good scout or a pilot who plans and communicates well can win the match without firing a shot.

More important than KDR, I would look at WLR. If you have WLR greater than 1 you're definitely doing something right. The greater the WLR the better you're doing it.

<snip>



Exactly. I'd rather have a team made up of people with a 2:1 W/L ratio than a team of 2:1 KDR. Sure there is correlation between the two but good teammates bolster the performance of the other team members and may not have as good a number of kills. And the closer the skills are within the team the more even the distribution of damage and match score becomes.

#59 Felicitatem Parco

    Professor of Memetics

  • PipPipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 13,522 posts
  • LocationIs Being Obscured By ECM

Posted 29 January 2016 - 01:18 PM

Guys...

The reason PSR is an XP bar is because it was designed under the assumption that the longer you play, the better you become. You learn more about game mechanics, weapons attributes, map layouts, etc.

The reason it's hard to go down in PSR rating is because you can't really "unlearn" what you have learned from playing all those matches under your belt.

The Tier system is designed to put distance between the most experienced players and the newest players. Yes, you will have non-elite players up in the highest Tier simply because they have been playing many matches per day for months. However, those people need some stiffer competition, because they obviously haven't gained enough tough training from fighting newer players.

The massive proportion of PSR points ts that are attributed to Wins (and the pittiful points you get from a loss) are meant to curb the rewards of people that don't work as part of the team. That lone Raven3L "sniper" who inflicted 800 damage but let his team get wiped out? He should not get a massive reward as if he clinched victory for the team. As said earlier by many here, dumping LRMs into targets can inflict great damage, but not neccessarily create victorious conditions.

People who are often on Losing Teams should not be rising in PSR because you're not winning. W/L is a better indicator of skill than K/D.


Does PSR derive too much weight from Damage Inflicted? Yes.
Does the concept of the system suck? No.

Edited by Prosperity Park, 29 January 2016 - 01:21 PM.


#60 MysticLink

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 214 posts

Posted 29 January 2016 - 01:22 PM

View PostMrJeffers, on 29 January 2016 - 01:15 PM, said:



Exactly. I'd rather have a team made up of people with a 2:1 W/L ratio than a team of 2:1 KDR. Sure there is correlation between the two but good teammates bolster the performance of the other team members and may not have as good a number of kills. And the closer the skills are within the team the more even the distribution of damage and match score becomes.


I've had games where I see the atlases are to scared to move in, and I'm in lighter assault then them, but then do the moving in, and tank, and we win the game as result.

Sure I die...but I make sure we won and that we didn't wait to get lasered to death or lrmed to death or sniped to death in bad trading war we were engaged in (we were mostly brawlers).

Should I get rewarded? Yes. I did some damage, not too much, but I lead the team.

Some people want to stand at the back, take no damage, deal most damage, and are not very willing to go frontline till near the end of the game. Sure they might get the most damage on the team, but they deserve to be penalized or not rewarded if they lose.

I think this is going to bread a better way than a team of full of people that just won't everyone else to tank damage but not them, and just do the most damage and kills, then say look how good I did by being an assault who stood in the back while my whole team died and I didn't tank, but look at all the damage. No the system should punish to a degree that play style. And if you in heavy and your assaults have taken a lot damage, and still don't want to front line and lose, you deserve to penalized. You will always do the most damage if you make sure you aren't the one taking damage, but just dishing it out. That's not a measure of skill, and you kdr will always be high if you do that cause u kill a bunch of weak ones at the end. It's not a measure of skill.

If you dish enough damage to make the team win, and you not tanking doesn't make them lose, then sure, the psr will reward you. Otherwise, I like that it punishes your play style.

Edited by MysticLink, 29 January 2016 - 01:25 PM.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users