63 replies to this topic

[Ashvins]

So in the Town Hall meeting Russ mentioned that the PSR (Pilot Skill Rating) needed a few "tweeks". Well to be blunt about it, Win or Loss factors too much into the PSR. You can have a Good Match (not Great just good) like this one and loose. and have your PSR stay equal or Go down.

On the other hand you can have a total garbage match and have your PSR go UP as shown Here.

I intentionally sabotaged myself in the 2nd match shown here to prove the point. Terrible damage, but a win and my PSR goes up. This should not happen. The win/loss factor for the PRS is what is out of wack. My PSR should have gone UP in the first match thanks to a good performance. In the 2nd it should have stayed the same or gone down due to the poor performance.

Any given pilot can only do so much to influence the win. In a situation where your team is not performing well yet you do, you should not be penalized through your PSR. Your already penalized in the pocket through the loss and lower C-bill earning, penalizing the PSR on top of it is adding insult to injury.

4X the damage yet 1/2 the C-bills is enough of a penalty for a Loss. Stop the reduction of our PSR because our teammates could not carry their weight. And most definately Stop the increase of PSR for terrible performance on a win.

Luck should have nothing to do with the PSR and with the way it is a terrible player can get lucky by having a good team and increase his PSR. Increasing his rank and then hindering people of higher rank with his poor play when he no longer get's lucky.

Let's look at the XP gained as well (BROKEN) 4X the damage, a kill a KMDD, Comp Dest, and 2/3 the XP of the win with terrible stats. I understand the C-bills being less for the loss but XP and PSR should not be affected by the Win or loss in any given match.

So there you go WHAT IS WRONG WITH PSR.

[El Bandito]

For the umpteenth time, Win/Loss MUST be the main factor in PSR to prevent people staying back and farming points!

Otherwise what's the point of leading a charge and allowing your team to win if your PSR goes down due to low damage done? What's the point of baiting the enemy lance to follow you while your team mops up the rest of the enemy? What's the point for scouting/Narcing/TAGging the enemy to help winning, if they do not give enough points for your PSR to go up?

Skilled players win more, hence win/loss has to be the chief mover of PSR. It could be harsher than current value when losing, but that's the gist of it.

Edited by El Bandito, 30 January 2016 - 04:33 AM.

[MrMadguy]

PSR is way too W/L dependent (and W/L, which is team performance measure - isn't appropriate measure of personal skill) and way too biased towards increasing. PSR CAN'T drop on Win at all, so the only way to descend - is to have average Match Score < 100, which is terribly wrong. Requirement to have Match Score > 400 on defeat in order to climb - is ok, as I think. Match should be decent for climb to be reasonable on defeat - no change is reasonable for simply good match.

Clarification:

There are two ways to climb/descend: 1) Win more/less 2) Have bigger/lesser Match Score

#1 doesn't depend on you: one player can neither win a match, nor lose it. It depends solely on MM, cuz W/L - is measurement of team performance and teams are being assembled by MM. You may deal 100500dmg, but lose. You may go AFK, but win. Using W/L as personal skill measure is proven to have large measurement errors. The larger team is - the larger this errors are. W/L is best for 1vs1, it may be acceptable for 4vs4, but for 12vs12 - it's just a mess. You can affect W/L only in some extreme cases - being much more skilled, then other 23 players, is most probable one. In most cases you stabilize at W/L = 1, so result of match - is just random.

About #2. Current MS thresholds:
0-100 MS - slow descend
100-250 MS - W/L dependent
250-400 MS - moderate climb
>400 MS - fast climb

So, if we'll assume, that PSR is "real" skill rating, used for matchmaking, then we'll see, that lower skill players (those, who due to some reasons can't reach 100% Tier 1) are doomed to sit at 100-150MS cap or constantly lose forever! For such players game turns into constant loss or stomp streak. This is unfair and unacceptable.

In order for MM to be fair and balanced, we should have stabilize MS cap value to be shifted upwards to at least 300 or even more. And this will require one simple fix - almost complete removal of bias towards winning. I.e. MS thresholds for winning should be exact the same, as for losing. I just don't think, that rising defeat MS thresholds even more is reasonable and good idea.

So MS thresholds will be following ones:
0-100MS - fast descend
100-250MS - moderate descend
250-400MS - W/L dependent
400-450MS - moderate climb
>450MS - fast climb

Edited by MrMadguy, 30 January 2016 - 03:32 AM.

[Alek Ituin]

PSR is bad. End of story.

It's basically ELO 2.0 and it's just as useless for determining actual player skill. It fails to account for the tactical differences between weight classes, fails to account for non-physical contributions, and relies entirely on random chance. The second biggest contributor to PSR is damage. So we have W/L and damage done... You know what that caters to? Heavies and Assaults. Skirmisher Mediums and scouting Lights get shafted because they don't provide a direct contribution to the match through sheer damage output. It has effectively alienated 50% of the weight classes by being a poorly constructed system.

What I propose is this: Change the foundation of PSR to take in to account weight class. If you play your weight class appropriately, you'll get rewarded appropriately. Damage farming in a Light wouldn't mean much to your match score, but targeting assists, TAG/NARC assists would get you huge points. On the other hand, TAG/NARC assists and targeting assists would net you nothing in an Assault, because your job is damage support.

Note: I don't mean to determine PSR per weight class. I mean using 1 of 4 scoring systems depending on the chosen weight class for the match.

Edited by Alek Ituin, 30 January 2016 - 03:12 AM.

[Ghogiel]

The PSR adjustnments just need to be harsher. Need more down on losing and winning. W/L is not random.

Though starting an alt and having 40:4 W/L and 50 KDR with no end in sight until past T3, does mess with PSR, but I can't see a solution to that. .

[Chuck Jager]

If you win more over 100 games you are better. Individual matches mean nothing. Single matches just allow any player to betterif the game is stacked in there advantage. You will always have games stacked against you too.

If you tend to help a team win in the mech you are in, I want you on my team. If you tend to lose more but get a ton of damage, I want you on the enemy team. PSR is not about epeen it is about how much you help win over 100+ games.

L2MATH

Edited by Chuck Jager, 30 January 2016 - 03:15 AM.

[C I L L I P U D D I]

@Ashvins

If you look at the top image, you have 415 damage, 1 kill, and some assists, in addition to some minor other bonuses. 415 damage is fairly....average, if that's your dead BLM on the ground. BLM's easily pull over 700, which in a loss, would give you a positive bump.

The second image has a long list of team oriented bonuses; UAV, Scouting, Capture, Spotting, etc. If you notice, you received doubled the amount of cbills and did 1/4 the damage you did in the top image. Even though you may not have been melting mechs, you were a more valuable asset to the team doing your scouting thing.

It's not all about damages and kills. As I'm sure most of you have seen, you can do less damage and have far less kills, yet have a higher match score.

If climbing PSR is what tickles your fancy, then focus on component damage, UAV usage/kills, scouting, KMDD, and Capping during Conquest. Yes, winning helps greatly and you don't have to carry as hard as you would in a loss but there are way more factors in PSR then jsut Kills, Assists, and Damage Dealt.

Edited by Cillipuddi, 30 January 2016 - 04:42 AM.

[pyrocomp]

The only problem with PSR is (as with lots of things in this game) in it's poor naming. PSR was introduced as a measure of pilots value to the team. To some point it is players skill with the game (not a measure of his reflexes or mouse movement speed and accuracy or damage dealing ability - see daily LRM whine threads).

So, in oder to clarify things and stop the stream of daily whine threads PGI (I think it's Russ personally) should rename Pilot Skill Rating to Pliot Value Rating and change nothing more.

Moreover, to preserve epeeen challenges and other forms of length/size/mass/I-know-how-to-use comparisons it is a good idea to introduce a handfull of new XP skill bars for brawling (based on brawling rewards, KMDD, DD and CD, health at the end of match and so on), scouting (scouting reward, UAV scouting, UAV kills and so on), precision (kills per DD, average distanse to the target and so on) and many more, including JJs (with flight distance per match, percentage of fall dagame to average damage recieve, airboune kills and so on).

PS: anyone claiming that W/L is too havily weighted in PSR and damage done should be rewarded more should explain why the light pulots are less skiled than assault pilots and should be put in their place.

Edited by pyrocomp, 30 January 2016 - 05:19 AM.

[Ted Wayz]

El Bandito, on 30 January 2016 - 12:59 AM, said:

For the umpteenth time, Win/Loss MUST be the main factor in PSR to prevent people staying back and farming points!

Otherwise what's the point of leading a charge and allowing your team to win if your PSR goes down due to low damage done? What's the point of baiting the enemy lance to follow you while your team mops up the rest of the enemy? What's the point for scouting/Narcing/TAGging the enemy to help winning, if they do not give enough points for your PSR to go up?

Skilled players win more, hence win/loss has to be the chief mover of PSR. It could be harsher than current value when losing, but that's the gist of it.

My position is well known. Win/loss should be a factor, but not the main factor. The best player if on the losing team should not go down while the worst player if on the winning team should not go up. PSR currently has nothing to do with skill.

As I have mentioned, the only people who defend PSR are people who ride coat tails in group queue. PSR protects their ego and their "skill".

[Ted Wayz]

Cillipuddi, on 30 January 2016 - 04:08 AM, said:

@Ashvins

If you look at the top image, you have 415 damage, 1 kill, and some assists, in addition to some minor other bonuses. 415 damage is fairly....average, if that's your dead BLM on the ground. BLM's easily pull over 700, which in a loss, would give you a positive bump.

The second image has a long list of team oriented bonuses; UAV, Scouting, Capture, Spotting, etc. If you notice, you received doubled the amount of cbills and did 1/4 the damage you did in the top image. Even though you may not have been melting mechs, you were a more valuable asset to the team doing your scouting thing.

It's not all about damages and kills. As I'm sure most of you have seen, you can do less damage and have far less kills, yet have a higher match score.

If climbing PSR is what tickles your fancy, then focus on component damage, UAV usage/kills, scouting, KMDD, and Capping during Conquest. Yes, winning helps greatly and you don't have to carry as hard as you would in a loss but there are way more factors in PSR then jsut Kills, Assists, and Damage Dealt.

I wish he had shown a SS of the player scores. His average match might have been above average given the match. PSR is not relative to performance in a match, it just picks an arbitrary bar that applies to all matches.

So what if his score in this match was double the closest person on his team and higher than any on the winning team. The resulting PSR is not the measure of his skill now is it?

PSR is garbage. Better ways of actually measuring skill, and exposing those that do not have it, are available and should be implemented.

[C I L L I P U D D I]

No matter how they change the PSR Ted, bads will be bads and the top tier guys will stay on top.

[Pariah Devalis]

That I can perform, personally, amazing but get my PERSONAL rank get dragged down due to an entire lance being terrible is a bit unsettling. On the other hand, the system does allow for your rank to go UP on a loss. However, it requires an unreasonably high output to do so when considering your team is falling apart around you and you are quickly becoming outnumbered:



*in a GAR no less. Suck it, Trebek!

**screenie from a video taken before the CB and XP values fully added up on the sides.

Edited by Pariah Devalis, 30 January 2016 - 07:01 AM.

[Mystere]

Ted Wayz, on 30 January 2016 - 06:50 AM, said:

My position is well known. Win/loss should be a factor, but not the main factor. The best player if on the losing team should not go down while the worst player if on the winning team should not go up. PSR currently has nothing to do with skill.

As I have mentioned, the only people who defend PSR are people who ride coat tails in group queue. PSR protects their ego and their "skill".

So the winning team's worst player who had a match score of 1001 (with the best player having 1012) should not advance in PSR, right?

And what do we do about people who vanguard, TAG, spot, NARC, bait, and distract in lieu of killing to help their team win? We should make their PSR drop to the floor anyway?

[Ted Wayz]

Cillipuddi, on 30 January 2016 - 06:58 AM, said:

No matter how they change the PSR Ted, bads will be bads and the top tier guys will stay on top.

Not true.

I do not see many of these Tier 1 players on leaderboards. Base our tiers on leaderboards and that will separate the men from the boys. Not only that PSR is blind to what you pilot. By basing skill rating on leaderboards you could have a skill rating for each mech.

[Ted Wayz]

Mystere, on 30 January 2016 - 07:04 AM, said:

So the winning team's worst player who had a match score of 1001 (with the best player having 1012) should not advance in PSR, right?

And what do we do about people who vanguard, TAG, spot, NARC, bait, and distract in lieu of killing to help their team win? We should make their PSR drop to the floor anyway?

Err, here we go.

If you read all that was said and the ideas presented to replace PSR you would know my position is the basis for your rating should be based on skill. So no, someone with a good score, on the winning or losing side, should not go down.

And if you have read all of what I have said, we should increase the rewards for ancillary activities to be more in line with damage and kills.

Edited by Ted Wayz, 30 January 2016 - 07:15 AM.

[Pendaelose]

Win Loss ratio is the only number that accurately identifies everything a player contributes to his team's performance.

Kills, Damage, Assists, even spotting don't mean anything without looking at "Do you help your team win more often than not?"

If I'm being matched with team mates, I'd take the guy with a 2.0 WLR over the 5.0 KDR any day of the weak, because you can score 5 kills in a match and still lose, but our team is literally twice as likely to win the match if Mr. 2.0 WLR is on the team. Here's the thing, he may not even shoot a single weapon all match. Maybe he's just awsome at caping points, maybe he spots everything and calls out the best targets over Voip? Maybe he's a ******* bullet sponge and can tank 700 damage for the team... I don't care what he does, he's going it right if his WLR is better than 1.

[C I L L I P U D D I]

Most of the top players dominate the leaderboards. Not knowing who they are is more of the issue. Psicoholic, Zeleglok, JayZ, just to name a few. All top tier comp players, who consistently top the boards.

[Mystere]

Ted Wayz, on 30 January 2016 - 07:14 AM, said:

Err, here we go.

If you read all that was said and the ideas presented to replace PSR you would know my position is the basis for your rating should be based on skill. So no, someone with a good score, on the winning or losing side, should not go down.

And if you have read all of what I have said, we should increase the rewards for ancillary activities to be more in line with damage and kills.

How do you accurately measure someone playing vanguard? What accurate method do you use to detect someone baiting, distracting, or delaying the enemy?

Isn't it much simpler to just use winning as an approximation?

Edited by Mystere, 30 January 2016 - 07:20 AM.

[Pariah Devalis]

Mystere, on 30 January 2016 - 07:19 AM, said:

How do accurately measure someone playing vanguard? What accurate method do you use to detect someone baiting distracting, or delaying the enemy?

Isn't it much simpler to just use winning as an approximation?

Surely they can track NARC/TAG use, as well as scouting bonuses, UAV usage, and actually hitting R on untargeted or unLOSed mechs, and we know they can track accuracy with weapon systems. There must be a way (or a way might be able to be made) to track one's ability to soak up damage and armor roll effectively. It already tracks sticking with your teammates, soon will have a reward for supporting with AMS, and already tracks ECM shielding friendly mechs. Hell, while not always a great indicator, there is probably a way to determine how long you lasted relative to the duration of a fight, indicating you stayed combat functional and influential for longer.

There are many metrics to determine a good player from bad beyond KDR or flat WLR. That is not to say WLR is not important. It absolutely is an important factor. It just should not be the single defining primary factor, here.

[Ted Wayz]

Mystere, on 30 January 2016 - 07:19 AM, said:

How do you accurately measure someone playing vanguard? What accurate method do you use to detect someone baiting, distracting, or delaying the enemy?

Isn't it much simpler to just use winning as an approximation?

Nope.

Currently PSR only regards killing and doing damage as factors in winning. If you play conquest and you insta cap what happens to everyone's PSR? The winning team stays even and the losers, who might have been a milli-second behind on the cap, all go horribly down.

The activities you mention you could do effectively in a win or a loss and neither are rewarded in either case. So what is your point?