Jump to content

Psr Issues And Why It Is Going To Break The Game In The Long Run

General Gameplay

21 replies to this topic

#1 MrKvola

    Member

  • PipPipPipPipPipPip
  • The Scythe
  • The Scythe
  • 329 posts

Posted 01 October 2015 - 08:37 AM

I'll try to explain where this article is wrong and why the system does/will not converge according to actual player skill level: http://mwomercs.com/...-tiers-and-psr/

Let's break down some of the clarifications then.

#1

This is a constant, not much to discuss here.

#2 and #3

Now this is horribly off the mark.

Since you can gain PSR rating when you meet one of 4 conditions (win/v. high, win/high, win/medium, lose/high), but lose only on 2 conditions (lose/medium, lose/low). Considering you score roughly 50/50 wins/losses the system naturally converges to Tier 1.

Perhaps you can influence the win/loss ratio by how much you carry or fail, but still the system will have a tendency to move you up. And still a lot depends on how much your teammates carry or fail. All you may influence is the speed of progress. You really have to do terribad AND have a lot of bad matches to actually drop in PSR rating.

What also influences the result is how the criteria for low/medium/high/very high score is determined. From what I have experienced it seems to me that the division is static. Perhaps it is not and I just did not analyze the situation properly. I believe it needs to adapt dynamically to scores achieved by all players on the team and divide players into groups according to what was achieved (and most probably achievable) in the match that is being considered. Because if you find yourself in a roflstomp (on the losing side) you may have very little chance to do much better than anybody else - regardless of your skill. On the other hand not much of a contribution seems to be required to gain points if your team wins.

Furthermore, as already discussed elsewhere, damage done strongly influences your match score (half of damage dealt is added to total point gain). You could just be running around shooting random targets into random spots (or just hurling various missiles at them) thus achieving the biggest damage tally and possibly top score while the total contribution could be questionable. If the team wins, OK, gain for most. If you lose, gain for you, loss for other, perhaps more skilled players.

Is the match score metric flawed? Perhaps. And most probably.

Is the PSR rating change flawed? Most definitely. A balanced system that converges to a point when the player reaches his skill limits would need to look like this:

Posted Image

The player's PSR would still improve - when the player's skills (aiming, loadouts, awareness, etc.) improve, because he would be more effective. Yet still his rating would oscillate around a value that is determined by his abilities.

In the current state players are rewarded for activity, not for real player skill or anything. The more matches you play the higher your PSR. What we end up with in the end is a lot of frustration, because no matter how good the matchmaker is, if the PSR system is broken:

1) Players level up to a point where their opposition is overwhelming. While they play their teammates may still carry them and they stay on the same tier level - but the game will not be enjoyable for them. Because their PSR will not drop/will not drop quick enough to make a difference. This may end up being a significant player deterrent.

2) Players with lower skill enter higher tier - again, a game spoiler - because they are not enough of a contribution to the team (even if the team wins) and may spoil the game for others. This mostly incites forum unhappiness. :)

And yes, I recognize they are basically two sides of a coin. But that is the result of an imbalanced system.

TLDR: What needs to be recognized is that we need the system to converge around a player's skill, not the top tier. The current system does not do that.

A reward system can be implemented to reward players for playing a lot, etc. Just please do not make the skill rating system that.

#2 TLBFestus

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,519 posts

Posted 01 October 2015 - 08:44 AM

Paul said:

That's because with experience comes better player skill.


Paul doesn't seem to acknowledge that there is a ceiling to skill level and that varies from player to player. Not everyone would "bubble up" to the top. Some of us will get passed and then held down because of the more skilled players above holding us back.

Not that that is a bad thing.

#3 Rebas Kradd

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,969 posts

Posted 01 October 2015 - 09:48 AM

This misunderstanding is a mix of a) Paul making a faulty assumption that playing always makes people better and B) everyone having bad reading comprehension.

If playing makes you better, you'll move up.

If playing doesn't make you better, you'll stay where you are.

Simple as that. :)

#4 Poisoner

    Member

  • PipPipPipPipPipPip
  • Veteran Founder
  • Veteran Founder
  • 440 posts

Posted 01 October 2015 - 09:55 AM

I'm good with my PSR.

#5 Chuck Jager

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 2,031 posts

Posted 01 October 2015 - 09:59 AM

View PostTLBFestus, on 01 October 2015 - 08:44 AM, said:


Paul doesn't seem to acknowledge that there is a ceiling to skill level and that varies from player to player. Not everyone would "bubble up" to the top. Some of us will get passed and then held down because of the more skilled players above holding us back.

Not that that is a bad thing.

Playing a super high number of games is called tenacity and practice. Most coaches and bosses see this as important as being a star player.

#6 Speedy Plysitkos

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,460 posts
  • LocationMech Junkyard

Posted 01 October 2015 - 10:10 AM

View PostPoisoner, on 01 October 2015 - 09:55 AM, said:

I'm good with my PSR.


im also good with my PSR. Only those 5 kills, 540 damage, 320 matchscore thing..........
:angry: :lol:

#7 MoonUnitBeta

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,560 posts
  • LocationCanada ᕙ(⇀‸↼‶)ᕗ

Posted 01 October 2015 - 10:57 AM

We could have a more accurate and efficient rating system if the match score we receive was given context to the match that it came from...

Edited by MoonUnitBeta, 01 October 2015 - 10:58 AM.


#8 Love in an Annihilator

    Member

  • PipPipPipPipPip
  • Knight Errant
  • 106 posts

Posted 01 October 2015 - 11:16 AM

View PostChuck YeaGurr, on 01 October 2015 - 09:59 AM, said:

Playing a super high number of games is called tenacity and practice. Most coaches and bosses see this as important as being a star player.


I could play a Million Football matches and still wouldn't become Leo Messi.

Of course experience is Important, but there is a (different) ceiling for everyone you're not going to break, even if you play MWO 24/7.

Edited by Rick Sanchez 1895, 01 October 2015 - 11:16 AM.


#9 Moomtazz

    Member

  • PipPipPipPipPipPipPip
  • The Vicious
  • The Vicious
  • 577 posts

Posted 01 October 2015 - 11:17 AM

I think they should simply rename it to Player Win Rating. It really is not an accurate reflection of individual skill, but people think it is because of the "Skill" in the name. It takes a very solid game to increase PSR on a loss anyway, so the whole rating is extremely biased to Wins.

#10 TLBFestus

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,519 posts

Posted 01 October 2015 - 11:22 AM

View PostChuck YeaGurr, on 01 October 2015 - 09:59 AM, said:

Playing a super high number of games is called tenacity and practice. Most coaches and bosses see this as important as being a star player.



Well IF I had a coach that told me I could be a Star Player, I'd end up telling him to ********** and go join my buddies for a beer and some wings, because he wouldn't know what he was talking about.

#11 Ex Atlas Overlord

    Member

  • PipPipPipPipPipPipPipPip
  • Survivor
  • 1,018 posts

Posted 01 October 2015 - 01:04 PM

View PostMrKvola, on 01 October 2015 - 08:37 AM, said:

Considering you score roughly 50/50 wins/losses the system naturally converges to Tier 1.


You can't assume this....

Therefore everything based on this assumption falls apart.

You might as well have your first assumption be "assume you're a Tier 1 player, then the system will place you in Tier 1 eventually".

/thread

Edited by The Atlas Overlord, 01 October 2015 - 01:05 PM.


#12 Revis Volek

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 7,247 posts
  • Google+: Link
  • Facebook: Link
  • LocationBack in the Pilots chair

Posted 01 October 2015 - 01:16 PM

View PostThe Atlas Overlord, on 01 October 2015 - 01:04 PM, said:


You can't assume this....

Therefore everything based on this assumption falls apart.

You might as well have your first assumption be "assume you're a Tier 1 player, then the system will place you in Tier 1 eventually".

/thread



I agree..

That's a bad assumption...

3,257 / 2,690

Thats my W/L ratio, Its not 50/50.

Edited by DarthRevis, 01 October 2015 - 01:17 PM.


#13 MrKvola

    Member

  • PipPipPipPipPipPip
  • The Scythe
  • The Scythe
  • 329 posts

Posted 01 October 2015 - 01:48 PM

By saying "Considering you score 50/50" I meant you are placed in the correct tier and games should be competitive (balanced) for you. And your rating should not climb because of that. Of course if you are placed below your skill level you will win more - that is based on your actual skill being superior to your rating. The exception being perhaps if you reach the upper ceiling and still manage to achieve more. Kudos at that point.

Ultimately when you reach skill ceiling you should see wins/losses come to a draw - where it is not certain (ideally 50/50) if you win or lose a match - because the opponent(s) should be equally matched to you (your team).

I was not assuming anything, I was considering a starting condition. Thereby saying if I had a balanced game scores my rating would still rise using the current rating system. And therefore the system is flawed.

Edited by MrKvola, 01 October 2015 - 01:48 PM.


#14 Hydrocarbon

    Member

  • PipPipPipPipPipPipPip
  • WC 2017 Qualifier
  • WC 2017 Qualifier
  • 659 posts

Posted 01 October 2015 - 03:26 PM

View PostMrKvola, on 01 October 2015 - 08:37 AM, said:

Posted Image


While I agree with that change in PSR/tier movement, I have a feeling it's actually working as intended better than people suspect. I know some people that usually end up in the ++ category & play as much or more than I do, yet they're in tier 2.

The other thing is most people's perception of their own skills is generally very different from reality. It works both ways - some people think they're god's gift to MWO, others think they suck


From what PGI has stated so far, my guess is they could potentially re-adjust/reset current tiers based on any changes to the formula. I doubt they discard the individual data. If they see that some tier 1 players are getting crazy scores while the majority keep shifting between tier 1 & 2, they may add in additional filters to the formula.

#15 El Bandito

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 26,736 posts
  • LocationStill doing ungodly amount of damage, but with more accuracy.

Posted 01 October 2015 - 04:24 PM

View PostMrKvola, on 01 October 2015 - 01:48 PM, said:

By saying "Considering you score 50/50" I meant you are placed in the correct tier and games should be competitive (balanced) for you. And your rating should not climb because of that. Of course if you are placed below your skill level you will win more - that is based on your actual skill being superior to your rating. The exception being perhaps if you reach the upper ceiling and still manage to achieve more. Kudos at that point.

Ultimately when you reach skill ceiling you should see wins/losses come to a draw - where it is not certain (ideally 50/50) if you win or lose a match - because the opponent(s) should be equally matched to you (your team).

I was not assuming anything, I was considering a starting condition. Thereby saying if I had a balanced game scores my rating would still rise using the current rating system. And therefore the system is flawed.


If one reaches T1 and still has 50/50 WLR, then he deserves to be there. If he can't, he will fall back to T2. I see no issue here.

Edited by El Bandito, 01 October 2015 - 04:26 PM.


#16 MrKvola

    Member

  • PipPipPipPipPipPip
  • The Scythe
  • The Scythe
  • 329 posts

Posted 01 October 2015 - 04:26 PM

The question stands: will he?

#17 El Bandito

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 26,736 posts
  • LocationStill doing ungodly amount of damage, but with more accuracy.

Posted 01 October 2015 - 04:27 PM

View PostMrKvola, on 01 October 2015 - 04:26 PM, said:

The question stands: will he?



Depends on his skill, hence PSR. I have not seen a single T1/2 player who has negative WLR.

Edited by El Bandito, 01 October 2015 - 04:42 PM.


#18 Sjorpha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,478 posts
  • LocationSweden

Posted 01 October 2015 - 04:57 PM

I agree completely with OP.

The bias towards PSR increasing over time needs to be removed, the scenarios in which it decreases needs to mirror those in which it decreases so that you actually stop at your current skill level.

You should not be able to increase PSR simply by playing a lot of matches with average performance, which is currently the case.

PSR needs to change much much faster both up and down. There are many players who don't play a lot of public queue, they will never converge to their actual skill level under the current system. One example would be members of CW oriented units who may be very good but almost never drop public, or some competitive players who mostly play private training sessions and tournaments but not casually. A player may also get better very fast over a period of intense training, or a very very good player may lose his edge quite quickly if he decreases his time investment. It's not good to have those players stuck in the wrong place for long.

I also think there should be some kind of modifier to PSR for unleveled mechs, it's not fun for anyone to have players dropping in suboptimal mechs if the matchmaker assumes they will perform as well as ususal.

#19 El Bandito

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 26,736 posts
  • LocationStill doing ungodly amount of damage, but with more accuracy.

Posted 01 October 2015 - 05:07 PM

View PostSjorpha, on 01 October 2015 - 04:57 PM, said:

I agree completely with OP.

The bias towards PSR increasing over time needs to be removed, the scenarios in which it decreases needs to mirror those in which it decreases so that you actually stop at your current skill level.

You should not be able to increase PSR simply by playing a lot of matches with average performance, which is currently the case.

PSR needs to change much much faster both up and down. There are many players who don't play a lot of public queue, they will never converge to their actual skill level under the current system. One example would be members of CW oriented units who may be very good but almost never drop public, or some competitive players who mostly play private training sessions and tournaments but not casually. A player may also get better very fast over a period of intense training, or a very very good player may lose his edge quite quickly if he decreases his time investment. It's not good to have those players stuck in the wrong place for long.

I also think there should be some kind of modifier to PSR for unleveled mechs, it's not fun for anyone to have players dropping in suboptimal mechs if the matchmaker assumes they will perform as well as ususal.


Yeah leveling up new mechs is certainly pain at the higher level. Also, since PGI did not take account for CW matches or private training matches in making the PSR, those players--if they are good--can rise up tiers very fast, since their overall number of matches are low.

Edited by El Bandito, 01 October 2015 - 05:08 PM.


#20 Sjorpha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,478 posts
  • LocationSweden

Posted 01 October 2015 - 05:27 PM

View PostEl Bandito, on 01 October 2015 - 05:07 PM, said:

Also, since PGI did not take account for CW matches or private training matches in making the PSR, those players--if they are good--can rise up tiers very fast, since their overall number of matches are low.


You can't assume their overall number of matches is low, they may have played a lot in public earlier while being bad, then focused on CW or private match training after that. In that case their old matches are irrelevant to their current skill level but will still prevent them from getting the correct PSR.

There is no sense in using a players accumulated total matches for PSR btw, that means you move slower in PSR the more you play, which is really stupid. PSR should only reflect your last 100 matches or something like that.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users