31 replies to this topic

[MoonUnitBeta]

Big question for this thread - Should the rest of your team's performance affect the weighting of how your match score affects your PSR gain/loss? And should each team be compared for match score differentials?

Key points for discussion that should be considered:
- Roflstomps (the severity of a win/loss) should matter.
- Match scores of team mates (the difference between you and your team) should matter.

If you have time, please read this so we're all on the same page:

Currently, your PSR is determined according to a fixed value. With a certain degree of consistency, scoring above or below a certain limit will define how well of a player you are.

For example, if you and your entire team did absolutely dreadful, and no one got above 100 match score, all 12 of you will face the same fate from the PSR nerf hammer. Everyone receives the largest reduction.

Some will say there's nothing wrong with that, but that would be true if this was a 1v1 scenario, just like ELO (the exact thing that PGI is trying to get away from).
The secret lies within the individual scores within your team's performance (to decide if making drastic changes to your rating makes sense), and how that is compared to the overall performance of the other team (to decide if the match outcome was close enough to warrant drastic changes made to your entire team).

If you did the very best you could have with what you had, you are still only 8.3% of your team.
Why does the other 91.7% have 0 merit when you get rated on your 8.3% contribution?

So far, PSR is damn near exactly like ELO. It's ignoring the complete fact that:

• You're part of a team of 11 other players.
  
• YOU are only 8.3% of that TEAM!
  
• The performance of the rest of your team.
  
• The performance of the overall match.

PSR doesn't care about the rest of the team, just like ELO. ELO only cared about two things, Win or Lose. PSR only cares about 4 things, Win, Lose, If you did good, or not.

Neither system takes your team, your opponents, or even the overall match, into consideration; factors that give your PSR meaning. PSR is making an ass out of u and me by assuming that the rest of your team is exactly where they need to be. PSR is currently based off of perfect-world math.

So what if it was a 11-12 game. Do you know what that means to PSR? It means that there's 24 players who likely have very high match scores. That's all that it means. PSR doesn't care about how close the game was. It only sees "Hey, you had an opportunity to get a blazingly cool match score, here's 25 magical tier points for whatever the hell reason." And then it goes onto your next team mate, and says the exact same thing to him.
Why! This was a close match! Why are people getting separated and pushed collectively forward for a match that doesn't require them to? Keep us here! That's the whole point - to sustain quality matches?
If there were a few players who scored wicked awesome, then push them forward a bit, and, similarly, if there were a few players who didn't do so hot, then push them back a bit. But we can't do that without understanding the match and the teams.



Does this not seem odd to anyone else?
In conclusion, when your performance is compared to your team and both teams are compared to each other, then the importance of a win/loss is incredibly lessened and a fixed scale turns into a reference instead of being the judge.

I feel that with this method we will be able to achieve the correct PSR faster and sustain it with a greater degree of accuracy.

Thank you for reading. Discussion can now take place.

Edited by MoonUnitBeta, 29 September 2015 - 10:42 PM.

[El Bandito]

It is a delicate thing to do without pissing off one group or another.

I personally think if we got stomped so badly that none of us got 200 points, then we all deserve the large rating loss. If some of us got 200 points while the others got 50, then the PSR already takes that into consideration and gives you less fall damage.

Edited by El Bandito, 29 September 2015 - 08:11 PM.

[SuomiWarder]

In general I agree with you. But i would image that PGI thinks they are encouraging "team play" and that a system where you do best by being the highest scorer over your team mates is not "team play". I wouldn't mind seeing PSR reduced if your team wins by a huge kill margin. So that dying in a winning 8-12 effort is worth as much as having the same damage/kills/etc in a 12-0 stomp win by your side.

[MoonUnitBeta]

El Bandito, on 29 September 2015 - 08:10 PM, said:

It is a delicate thing to do without pissing off one group or another.

I personally think if we got stomped so badly that none of us got 200 points, then we all deserve the large rating loss.

You wouldn't be ticking anyone off...

Exactly, you're thinking in terms as a team vs another team.
Now say you were the one who got 500 in a team that lost like that. You did freaking awesome, but you only go up a medium amount? You should be on the OTHER team. The other team who is getting the LARGEST tier boost, and most of them get that just for winning.

Remember, you're only 8.3% of your team. So why are you, someone who got 500 match score on the losing team, only getting a medium raise when 91.7% of your team were the reason for you not advancing?

Edited by MoonUnitBeta, 29 September 2015 - 08:17 PM.

[El Bandito]

MoonUnitBeta, on 29 September 2015 - 08:14 PM, said:

You wouldn't be ticking anyone off...

Exactly, you're thinking in terms as a team vs another team.
Now say you were the one who got 500 in a team that lost like that. You did freaking awesome, but you only go up a medium amount? You should be on the OTHER team. The other team who is getting the LARGEST tier boost, and most of them get that just for winning.

Remember, you're only 8.3% of your team.

Right, and who is to say that next time you won't be on the stronger team's side and rolling over the opposition? It all evens out over the long run IMO, and unless one improves personally, he will have hard time climbing later on.

Edited by El Bandito, 29 September 2015 - 08:18 PM.

[MoonUnitBeta]

El Bandito, on 29 September 2015 - 08:17 PM, said:

Right, and who is to say that next time you won't be on the stronger team's side? It all evens out over the long run and unless one improves personally, he will have hard time climbing later on.

And that's exactly the issue. The long run is so long for you to find out how much of a difference your continual 8.3% contribution makes. And it's also very spikey. The way PGI counter acts the spikeyness, is that they make it take such a long time to get anywhere (tier wise). Not only that, but last time I checked - we want to reduce the amount of rofl stomps... So, yay, you get placed on the stronger team, but the keyword there is -stronger-. Why is there even a stronger team? This is called a match-maker for a reason, not "a-chance-you'll-be-on-the-stronger-team-later" maker. As I said, and will continue to say, win/loss carries too much weight, and your actual contribution to the match is over-looked.

If it's called match score - why isn't it relevant to the match that it's coming from?

Elo eventually placed you much the same way. The only problem PGI solved was so that you couldn't get carried up by being in a group of really good people. But it takes forever, and PSR is still heavily influenced by win/loss, and not relative performance.

"unless one improves personally, he will have hard time climbing later on." I'm confused. Not sure how this relates to PSR...
I'm going to take a wild guess - are you saying that fair matches aren't grounds for improvement? My response is simply this: If someone wants to be better, they'll play better. You don't need a rating to tell you that. ... was I even close? lol i tried *shrug*

Edited by MoonUnitBeta, 29 September 2015 - 10:07 PM.

[Wintersdark]

Yeah, I had assumed that PSR wouldn't measure against fixed values but instead using relative values across your team. I assumed that, because it was the only way PSR made sense as an alternative to Elo. As it stands, PSR is just more exploitable Elo.

Clearly, I was wrong. Sadly. So, now, even if you're a great player doing everything right, if your on a team that gets omgwtfbbq rolled, you're going to lose rating. Sure, maybe you got 200 match score... But the rest of your team was 0-50! You got 200 match score in a 12v1 situation, and that's a damned good showing.




(Or, you hid in the back and nerfed LRM's uselessly until they got around to killing you last, but that's another issue with PSR).

[Jelan]

imho the win/loss calculation needs to be removed completely as its irrelevant and the PSR change needs to be based on your score solely.

In theory the new scoring system is based on team play, (notwithstanding the fact that its still to highly damage focused) so if you get a decent score you have been a team player and should be rated as such.

It seems a no brainer to me, low score = psr loss, medium score = psr stays the same, high score = psr goes up. Win or Lose should have no effect on it whatsoever as that is based on your teams performance as a whole and not yours

[Doman Hugin]

I understand what the OP is saying:

If you win with a good score, but were riding the coat-tails of those that caused the win, ( say 7-8th highest score on that team).

Or loose with an average score, but it was still more than double anyone else on the team.

These things are not taken into account.

Would this make PSR more accurate, i believe it would, but in the long run these types of games are not frequent enough to count too much towards your PSR, so while i believe it to be more accurate it probably wont make much differance to your PSR overall.

Oh! and Jelan, i believe that nebulous OP variable called teamwork, is why win / loss is a major factor in PSR.

[Ex Atlas Overlord]

Jelan, on 30 September 2015 - 01:31 AM, said:

imho the win/loss calculation needs to be removed completely as its irrelevant and the PSR change needs to be based on your score solely.

This is a TERRIBLE suggestion.

This would result in people simply throwing teamwork away to pad their stats even more.

[El Bandito]

The Atlas Overlord, on 30 September 2015 - 02:06 AM, said:

This is a TERRIBLE suggestion.

This would result in people simply throwing teamwork away to pad their stats even more.

Yep, there will be so many guys staying behind to pad their precious score and no one will be willing to expose himself first due to current state of alphas.

Edited by El Bandito, 30 September 2015 - 02:34 AM.

[Jelan]

The Atlas Overlord, on 30 September 2015 - 02:06 AM, said:

This is a TERRIBLE suggestion.

This would result in people simply throwing teamwork away to pad their stats even more.

Not at all, if the base scoring system rewards you for teamwork, which it allegedly does, then people can pad their stats by doing teamwork, its a win win scenario

edit: And tbh if people want to try and pad their stats to get to a tier higher than they are capable of then thats up to them but i would think it would lessen their gameplay experience

Edited by Jelan, 30 September 2015 - 02:38 AM.

[El Bandito]

Jelan, on 30 September 2015 - 02:35 AM, said:

Not at all, if the base scoring system rewards you for teamwork, which it allegedly does, then people can pad their stats by doing teamwork, its a win win scenario

edit: And tbh if people want to try and pad their stats to get to a tier higher than they are capable of then thats up to them but i would think it would lessen their gameplay experience

And why would you support lessening the gameplay experience?

[Jelan]

El Bandito, on 30 September 2015 - 02:38 AM, said:

And why would you support lessening the gameplay experience?

I said i think it would lessen their gameplay experience not the gameplay experience, which is entirely up to them

Edited by Jelan, 30 September 2015 - 02:45 AM.

[El Bandito]

Jelan, on 30 September 2015 - 02:41 AM, said:

I said i think it would lessen their gameplay experience not the gameplay experience, which is entirely up to them

Still not convinced. I have played enough challenge events in MWO to know that not including win/loss as a factor is a very bad idea. Players go to despicable lengths to earn their precious score.

Edited by El Bandito, 30 September 2015 - 03:03 AM.

[MoonUnitBeta]

El Bandito, on 30 September 2015 - 02:50 AM, said:

Still not convinced. I have played enough challenge events in MWO to know that not including win/loss as a factor is a very bad idea. Players go to despicable lengths to earn their precious score.

Win/loss is still being included. The chart Paul posted is still being included, but niether of them would be de-facto ways to tell you how much you advance or not. There's other factors involved that affect how good and how well you do in a match, and those are not being considered.

As mentioned earlier, if you have a really good 11-12 game, where everyone did really well, the current PSR rating only sees each one as their own individual, and assumes that everyone else performed the way they were supposed to (if you can tell me what that means, be my guest!). The fact that it's easier for everyone to get a higher match score in a well matched game vs a poor 2-12 game is ignored. The fact that the game was even matched at all is ignored.
How are those not important? This is a team based game. Why the team, and why the match, are not considered are beyond me. You're being rated against fixed values in a table, not against how relevant you were.

The Atlas Overlord, on 30 September 2015 - 02:06 AM, said:

This is a TERRIBLE suggestion.

This would result in people simply throwing teamwork away to pad their stats even more.

Not at all. Please explain why would it result in padding. Please explain HOW you can pad your stats?
If it results in people "throwing away teamwork", then there's problems with PGI's match score formula, not the PSR rating.
If everyone in your team collectively threw out team work and performed equally, everyone one your team would be getting the same match score.
So the new way would it would look at your match score, it would then look at your teams match score, and it would then look at your opponents match score to gauge the overall quality of the match. And those variables will also affect the amount of change your match score has on your PSR. Rather than Paul sitting in the background saying "Oh you got 300 match score on a lose? Yeeeeeeeeeah it's ok, but not good enough." HUH? Who says?! Did you see the miracle that me or my team just performed?, etc etc.

Right now it's just arbitrary values with no meaning except for paul saying that there's a meaning.

With the new way, all we're doing is adding better ways to keep the rating accurate by adding relevancy to your match score for that match... We're taking Paul's arbitrary values and giving them a meaning inside of a match.


To both of you:
If there's confusion or something, please explain it because I'm really not seeing how this is throwing away team work... Would a picture help? Am I not explaining this clearly enough? Are you fixating on a singular scenario that you see abuses the system? Is it a lack of understanding? I'd like to know.
Thanks!

Edited by MoonUnitBeta, 30 September 2015 - 07:25 AM.

[Vlad Ward]

I think the real question is why the OP would only use 1 significant digit for an easily expressible rational number.

[Appogee]

Are you certain there are fixed match score levels which determine on a loss whether your PSR goes down a lot, a little, stays the same, or goes up?

If so, what are those levels?

[Screech]

Tom Brady is only 9% of his offense, I wonder if he ever thought he didn't have any impact on the outcome of a game. I wonder if he ever thought as long as he threw 4 touchdowns who cares who won the game. Until they come up with a Solaris mode wins and loses should be the main objective and rewarded accordingly.

[Chuck Jager]

Learn to math
bad team but good performance PSR stays same or slightly drops
Good team good performance bigger jumps in psr than drop in losses (why people are upset with just playing to move up)

Over 1000 games you should draw close to equal team quality for wins and losses. Variance in random number is offset by bigger reward for win.

Only constant factor guess what - YOU. IF you constantly play well you move up. (note) there are ways to inflate numbers that help epeen, but they let the team fall short. Number padding by never being the person in front will result in consistently letting the team down. You will get more losses.

(total wins) adjusted by (consistent good play in losses) over (extended period of time) = movement up in psr

Your performance in the loss does affect how you move up even a negative arrow can still be less than a worse performance. Do not just look at the arrow and make assumptions.

This is a math problem with multiple parts

Edited by Chuck YeaGurr, 30 September 2015 - 08:23 AM.