Jump to content

Is Psr / Tier Accurately Reflecting Skill Level?


56 replies to this topic

#41 IraqiWalker

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 9,682 posts
  • LocationCalifornia

Posted 25 September 2015 - 10:34 PM

View PostNomadicCanuck, on 24 September 2015 - 05:11 AM, said:

All players in my post above have played since January. The difference is some of us play daily, some play a few times a week and some only play a few nights per month. In any case, all pilots have active profiles.

I am aware, as you said, of the January 2015 date which is why we were all shocked at some of the Tiers. I have a pilot that was ranked in the upper 20 for the last tournament; He has been playing for years and is a solid contributor to any match he's in although he only plays about one night per week. He is ranked Tier 4.... and has been destroying the solo queue. That seems "off" to me.

Seriously, ask in your unit to see if your infrequently playing pilots are ranked below their station and your frequent pilots ranked above. That's the question.

Basically some players have many data points for the match maker to figure out where they belong, while others don't. I really don't see a problem with that. It was the same when Elo was implemented, it was the same when we reset the stats, and it is how the world functions.

Only way for someone's skill to be measure accurately, is by them playing the game often, so the matchmaker has enough data points to figure out where they are.

I can guarantee you that the inconsistencies increase proportional to how rarely the player plays.

#42 Mochyn Pupur

    Member

  • PipPipPipPipPipPipPip
  • The Patron Saint
  • The Patron Saint
  • 521 posts
  • LocationDerby, England

Posted 26 September 2015 - 01:08 AM

Excellent match for me last night win a CTF-4X - 3 kills, 5 assists, 691 damage and a team loss - - Tier score went down ? Later, 0 kills, 3 assists, 121 damage and a Team win - Tier score went up!

Yes, this is working a expected :)

#43 Anunknownlurker

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 362 posts
  • LocationBetween here and there

Posted 26 September 2015 - 01:51 AM

@PeppaPig - that is a perfect example of the flaw with this system; your personal contribution to the game was significantly larger in the loss (imagine how much more of a stomp it would have been if you hadn't been there?) than it was in your win yet this is not reflected in the PSR. In which of those games were you "playing for the team" as PGI is trying to encourage?

Any match making system which is predicated on the, let's be honest, pretty random nature of win/loss is not fit for purpose; of course I accept that a truly skilled player can carry a game but that is a realm inhabited by the MechWarrior gods not the vast majority of players.

Now, if this system is designed to get the top 1% out of the PUG world then, great, let them slap each other about until Doomsday and leave us potatoes alone but it clearly isn't, a 2 tier spread does not do that, therefore, to me, whilst PSR is more suited to this game than Elo it still needs a lot of tweaking to make it a reliable, trustworthy match making algorithm.

Peace.

Edited by Jimmy DiGriz, 26 September 2015 - 01:57 AM.


#44 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 26 September 2015 - 06:51 AM

View PostNomadicCanuck, on 25 September 2015 - 12:54 PM, said:

I read the update on the PSR and got to thinking... See if there's merit to this theory: If you're a Tier2 pilot, and you play with a friend who is Tier4. Your average is Tier3 and you would play with other tier3's. Lets assume your W/L is 1:1 ... when you win, the Tier2 pilot (in theory) does well and he moves up. The tier4 pilot has a hard time competing against the tier3's and may end up equal. When you lose, the Tier2 pilot likely continues to excel and the tier4 pilot loses points. The combined 'friendship average' remains similar in Tier3 while the actual PSR widens... eventually, It's Tier 1 and Tier 5 playing in an average Tier3 drop. The fundamental argument is that the Tier4 pilot *should* improve, breaking the cycle...I'm thinking that players plateau at a certain point and that improvement may never come. The option? Stop playing with your high-scoring friends. Just a theory...


I think Tiers are not currently considered in Group Queue matchmaker. It's just that better players tend to roll in groups, so that is where you might see them.

#45 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 26 September 2015 - 07:37 AM

View PostTesunie, on 25 September 2015 - 09:22 PM, said:


Damage done to teammates reduces your match score. You can see how much damage you've caused to an ally. Killing a teammate also will drop your match score as well. Not to mention a team kill and team damage's affect on a possible match win, which reduces your chances of gaining PSR...

So in a way, it does. In a way, it doesn't.


ORLY? What about these endgame stats which ended in PSR INCREASE!

#46 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,586 posts
  • LocationSeraphim HQ: Asuncion

Posted 26 September 2015 - 08:38 AM

View PostAnachronda, on 26 September 2015 - 07:37 AM, said:


ORLY? What about these endgame stats which ended in PSR INCREASE!


I can not see his match score, but it does look like he did other things besides attack friendly forces (knowing and intentionally or not, I can't say). The friendly damage he did reduced his match score some, but as he did not kill his teammate, he didn't get the entire negative impact possible. However, with PSR, on a win you need a very small point score to gain an increase.

Right now, PSR is more rewarding on the wins than actual player score. If you win, you basically are going to go up in PSR right now. If you lose and do "insanity" for a match score, your PSR "may" go up then. Otherwise, if you lose, your PSR is basically going down. (Or another rare chance on a lose, your PSR remains the same. I've only had it happen twice. Sad part is, I've had it remain equal on one match score, and then drop down on a higher match score in a loss. I is confused.)


PS: The problem with that thread you linked to... None of them are showing their match score along with their end of match breach down. This makes it harder to gauge the affect of their team damage. However, do note that on the victory results, they typically won and had no change to their PSR. That is because their match score was so low on the win. (Which actually has seemed fairly hard to get in an actual game.)

#47 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 26 September 2015 - 04:02 PM

All of those Team Damage examples meet with my expectations.
The worst you can do in a victory is remain =. That matches the team-damage screen shots.

The screen shot of the defeat was 605 damage with only 22 team damage, with kill and 5 assists ... not a bad match and probably got you high enough to be in neutral territory.

#48 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 27 September 2015 - 05:29 AM

View PostTesunie, on 26 September 2015 - 08:38 AM, said:


I can not see his match score, but it does look like he did other things besides attack friendly forces (knowing and intentionally or not, I can't say). The friendly damage he did reduced his match score some, but as he did not kill his teammate, he didn't get the entire negative impact possible. However, with PSR, on a win you need a very small point score to gain an increase.

Right now, PSR is more rewarding on the wins than actual player score. If you win, you basically are going to go up in PSR right now. If you lose and do "insanity" for a match score, your PSR "may" go up then. Otherwise, if you lose, your PSR is basically going down. (Or another rare chance on a lose, your PSR remains the same. I've only had it happen twice. Sad part is, I've had it remain equal on one match score, and then drop down on a higher match score in a loss. I is confused.)


PS: The problem with that thread you linked to... None of them are showing their match score along with their end of match breach down. This makes it harder to gauge the affect of their team damage. However, do note that on the victory results, they typically won and had no change to their PSR. That is because their match score was so low on the win. (Which actually has seemed fairly hard to get in an actual game.)


I think you need to look again. In several matches he did not have any kills and had a few assists. The ones where he had kills he only had one. I can tell you right now from experience that having that few kills will not give you much in terms of match score. Massive damage will. That's always been true, BTW, even before they changed the scoring system. Kills and assists gave a few measly points, but 1/10 of your damage added to match score. Now it is 1/2 of your damage.

Doesn't much matter what his score is, no matter how you slice it. When you imply he had a lower match score, because that just means the system is even more crazy.

I'd suggest you test yourself how the scores turn out. Here is some data for you, though. As you can see, kills give you 4 points. Damage gives you half the damage. Which means that while they may have added a few things that increase score by a few points, all the other numbers stayed the same, except that damage now grants 1/2 toward match score instead of 1/10. They also show 0 rather than a negative for team damage.

View PostNomadicCanuck, on 26 September 2015 - 04:02 PM, said:

All of those Team Damage examples meet with my expectations.
The worst you can do in a victory is remain =. That matches the team-damage screen shots.

The screen shot of the defeat was 605 damage with only 22 team damage, with kill and 5 assists ... not a bad match and probably got you high enough to be in neutral territory.


Yeah except that a game like that would end with your PSR going down in the case of a loss. And that's where it's kind of nuts. 600 damage should be a good enough game to keep you from losing PSR, but it isn't. From what I am hearing, even 1000 damage wouldn't be enough, but I cannot confirm that having only one 1000 damage game ever. Saul or some other tier 1 would be better to ask there.

*EDIT* Here's an example: I was messing around in my Pirate Bane today making money by playing Conquest matches. Capping with a hero mech like that is still the easiest way to make money in regular drop matches. In this game, I did a lot of stuff - captures, assists, scouting, and 1 kill. Look at the score there. That score (151) would most definitely result in a loss of PSR if we had lost. Since we won, it went UP! And thinking back, PGI had said the new scoring was supposed to increase the rewards for infowar participation,stuff like scouting, etc. But the first thing they did was increase the reward for damage while keeping the reward for kills and assists the same. So really, damage is still the greatest contributor to matchscore, yet even with a pretty good matchscore you will lose PSR if you lose. I'd also go back to what they said about PSR:

Quote


With the new PSR system:

If a players' team wins, and the player did well during the match (achieving a high Match Score), the player will rise in skill rating.
If a players' team wins, but the player did not perform well (achieving a low Match Score), the player will not move in terms of skill rating.
If a players' team loses, but the player does exceptionally well (achieving a very high Match Score), the player will go up slightly in skill rating.
If a players' team loses, but the player performed well (achieving a high Match Score), the player will not move in skill rating.
If the players' team loses, and the player performed poorly (achieving a low Match Score), they will drop in skill rating.



So in a loss, you are only supposed to lose PSR if you do POORLY. You are only supposed to increase PSR on a win if you do WELL. Yet on a win, 150 in this case is doing WELL, but in a loss, 300+ is considered POOR performance.

Edited by Anachronda, 27 September 2015 - 07:04 AM.


#49 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 27 September 2015 - 09:56 AM

View PostAnachronda, on 27 September 2015 - 05:29 AM, said:


So in a loss, you are only supposed to lose PSR if you do POORLY. You are only supposed to increase PSR on a win if you do WELL. Yet on a win, 150 in this case is doing WELL, but in a loss, 300+ is considered POOR performance.


http://mwomercs.com/...-tiers-and-psr/

Not quite. According to Paul's clarification, anything other than a HIGH score or VERY high score will cause you to lose PSR in a loss.

I don't know what they consider high and very high because those are relative terms; "High" might be 300 for some pilots or might be 600 for others. I suspect the definition of what high and very high are depend on the tier you are in. A Tier 5 doing a 400 match score game is likely very high compared to a Tier 2 where that is closer to the expected.

Regardless, I think I'm side-tracked a bit. My question was whether the current algorithm is accurately placing your pilots into tiers that are at their skill level. My experience is no.

#50 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 27 September 2015 - 03:43 PM

View PostNomadicCanuck, on 27 September 2015 - 09:56 AM, said:


http://mwomercs.com/...-tiers-and-psr/

Not quite. According to Paul's clarification, anything other than a HIGH score or VERY high score will cause you to lose PSR in a loss.

I don't know what they consider high and very high because those are relative terms; "High" might be 300 for some pilots or might be 600 for others. I suspect the definition of what high and very high are depend on the tier you are in. A Tier 5 doing a 400 match score game is likely very high compared to a Tier 2 where that is closer to the expected.

Regardless, I think I'm side-tracked a bit. My question was whether the current algorithm is accurately placing your pilots into tiers that are at their skill level. My experience is no.


Wrong again, Sam. I can tell you with absolute certainty that a Tier 5 making 300+ points will lose PSR. Nothing in Paul's post contradicts the earlier PGI quote; in fact it CONFIRMS it. And what he left out was the fact, which we have well established, that the same number that is considered "well" or "very well" in a win is "poorly" in a loss. He does not give any numbers. He also does not say anything toward your claim that it is based on tiers. However the numbers Saul, a tier 1, posted, which would have been abysmal, give him the same result that I, as a definitely not tier 1 player, get, a rise on win.

Edited by Anachronda, 27 September 2015 - 03:50 PM.


#51 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 28 September 2015 - 07:19 PM

I don't believe I'm incorrect about my statement above. In a loss, you must do "high" or "very high" in order to increase your PSR.

From casual observation, it appears that mechs that do 300-400 match score seem to be considered "moderate". In the case of a loss, this results in a downward adjustment. Match score over 400 seem to result in a slight rise in a loss. It wouldn't surprise me if there were hidden multipliers on the match score to determine the amount of up or down you go.

Seeing the 4 marks, Low, Medium, High, Very high... how do you tell on the win since Medium, High and Very High all appear the same with a green arrow in a win? Comparatively, Low, medium can be classed together and high and very high can be separated on a loss. It is easier to estimate the match score levels on a loss.

#52 Commissar Aku

    Member

  • PipPipPipPipPip
  • Urban Commando
  • Urban Commando
  • 195 posts

Posted 30 September 2015 - 06:24 AM

Alright just because I wanted to know, I did 254 damage 1 kill and 2 team kills and 208 team damage on a win match score 154 and my PSR went up, I got a match score of 658 4 kills 6 assists 4 team damage on a loss and my score still goes down. Based on team work may ass, the PSR is so broken if it is rewarding clearly anti team play, and punishing people who clearly play as a team. It is stupid to reward you for winning and punish you for losing, we play games to avoid that metric about real life why would we play games to enforce that metric?

#53 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 30 September 2015 - 02:26 PM

Edit: Removed example one as my pilot didn't give me correct information.

Edit: Forgot to add the second example.

Another one of my pilots tells me that they based rushed and even the winning side lost PSR points!

Edited by NomadicCanuck, 30 September 2015 - 02:32 PM.


#54 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 01 October 2015 - 12:16 AM

View PostCommissar Aku, on 30 September 2015 - 06:24 AM, said:

Alright just because I wanted to know, I did 254 damage 1 kill and 2 team kills and 208 team damage on a win match score 154 and my PSR went up, I got a match score of 658 4 kills 6 assists 4 team damage on a loss and my score still goes down. Based on team work may ass, the PSR is so broken if it is rewarding clearly anti team play, and punishing people who clearly play as a team. It is stupid to reward you for winning and punish you for losing, we play games to avoid that metric about real life why would we play games to enforce that metric?


I agree somewhat, but I would be okay with this system if the numbers were more sane. The amount you need to do well should not be more in a loss than in a win. That is kicking someone when they are down.

And reports like yours, the Saul thread, and the matchscore experimentation thread, all suggest that team damage is not deducting from score. Which I think might mean that the damage you do to your team is included in your damage, but not being used as a penalty. If that is correct, it would indeed mean that doing something that really hurts the team is being rewarded. I'd like to try some experiments with that but you kind of need a willing victim and if you don't do it in a private match it's going to mean your team gets screwed that match.

#55 Anachronda

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • 293 posts

Posted 01 October 2015 - 12:24 AM

View PostNomadicCanuck, on 28 September 2015 - 07:19 PM, said:

I don't believe I'm incorrect about my statement above. In a loss, you must do "high" or "very high" in order to increase your PSR.

From casual observation, it appears that mechs that do 300-400 match score seem to be considered "moderate". In the case of a loss, this results in a downward adjustment. Match score over 400 seem to result in a slight rise in a loss. It wouldn't surprise me if there were hidden multipliers on the match score to determine the amount of up or down you go.

Seeing the 4 marks, Low, Medium, High, Very high... how do you tell on the win since Medium, High and Very High all appear the same with a green arrow in a win? Comparatively, Low, medium can be classed together and high and very high can be separated on a loss. It is easier to estimate the match score levels on a loss.


You were incorrect in saying that the numbers vary by tier. They clearly do not. You were also incorrect in stating that 300 would be good enough to stay even. If 400 is enough to rise (couldn't say since on a loss I don't typically score that high), what is the break-even? You'd basically need to do 800 damage to do that. Besides, you said 300 was enough before and it is not, so do you have a reason to believe 400 is enough now? Because I think I have seen higher numbers claimed. I think you're also ignoring how the scoring system now rewards damage even more than it once did. 800-1000 damage may be typical for some, but I would think for most people it is fairly uncommon. Those games where people do that well on a team don't usually end in losses, either, in the regular queue.

#56 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,586 posts
  • LocationSeraphim HQ: Asuncion

Posted 01 October 2015 - 08:38 AM

View PostSatan n stuff, on 01 October 2015 - 07:35 AM, said:

https://mwomercs.com...-1430-18aug2015
The important part:

Quote

The weighted formula and action triggers will not be released publicly. However, we are currently debating the merits of making all players’ respective Skill Tier levels publicly accessible. Part of this potential public Skill Tier listing is the possibility for opting-in or opting-out of listing your Skill Tier publicly. Tina will likely be putting up a poll to gather feedback about which option the community feels would be the best choice for handling public Skill Tiers.



#57 Commissar Aku

    Member

  • PipPipPipPipPip
  • Urban Commando
  • Urban Commando
  • 195 posts

Posted 01 October 2015 - 09:01 AM

View PostNomadicCanuck, on 27 September 2015 - 09:56 AM, said:


http://mwomercs.com/...-tiers-and-psr/

Not quite. According to Paul's clarification, anything other than a HIGH score or VERY high score will cause you to lose PSR in a loss.

I don't know what they consider high and very high because those are relative terms; "High" might be 300 for some pilots or might be 600 for others. I suspect the definition of what high and very high are depend on the tier you are in. A Tier 5 doing a 400 match score game is likely very high compared to a Tier 2 where that is closer to the expected.

Regardless, I think I'm side-tracked a bit. My question was whether the current algorithm is accurately placing your pilots into tiers that are at their skill level. My experience is no.

If their PSR or "ELO" is working correctly there should be an average match score you are achieving in each match. If you are continually doing 1000 points of damage in a match there is something wrong and you need to be moved up until your match score is more even with the rest of your team, because that would mean you are more or less fighting people in the same ability level as you. If you are regularly getting match scores 100 to 200 more than the rest of the people around that means there is something wrong, watch a comp team game sometime and look at their damage and scores, if anything they are lower than a lot of average scores in lower tier games, but it is because they hit more and don't do as much damage due to better aim and teamwork, however their match scores are about even across the board, and not any higher than a standard pug game.





2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users