27 replies to this topic

[Soultraxx]

One measly pixel movement on a PSR change whether I had an 800 dmg/3 kill game in an Ebon or a 17 damage "got goosed by a Streak Crow" match in my KFX.

Maybe it was better not knowing my Tier because now its like a nagging itch that is difficult to scratch.

Anyone else feeling this?


(Im Tier 3 by the way)

[Chuck Jager]

somebody said "psr is not a badge or achievement to be earned"

if you want to move do what ever it takes to win a match and then add damage and kills. You move faster over time with consistant wins versus sporadic wins with high damage. This is to get a sample group of how the player does over time especially in very close matches where every action counts.

Basically keeps folks who spam lrms, farm dcs, and hide to snipe damage from getting big bumps when they are carried

[Random Carnage]

You got a whole, entire pixel? Damn boy, you da lucky one!

I have a dead pixel on my screen. I think it died of old age and bordom waiting for the tier bar to fill it.

[Paigan]

Soultraxx, on 27 September 2015 - 11:11 AM, said:

One measly pixel movement on a PSR change whether I had an 800 dmg/3 kill game in an Ebon or a 17 damage "got goosed by a Streak Crow" match in my KFX.

Maybe it was better not knowing my Tier because now its like a nagging itch that is difficult to scratch.

Anyone else feeling this?


(Im Tier 3 by the way)

All I say is: PSR insanity

Your problem goes in the same direction.
Current PSR does not really rate a pilot's performance in the match, it just rates if your team won or not, just like ELO. Plus some teeny-tiny pilotic-specific corner-case-addon that makes up like 1% or so.

Edited by Paigan, 28 September 2015 - 01:39 AM.

[Joseph Mallan]

LOL you expect to see rapid results without knowing what the tolerance is per tier?

[Paigan]

Joseph Mallan, on 28 September 2015 - 01:39 AM, said:

LOL you expect to see rapid results without knowing what the tolerance is per tier?

Ah, good old Joseph.

As almost always, you fail to see the point.

What he said was:

Soultraxx, on 27 September 2015 - 11:11 AM, said:

One measly pixel movement on a PSR change whether I had an 800 dmg/3 kill game in an Ebon or a 17 damage "got goosed by a Streak Crow" match in my KFX.
[...]

It's about the comparison of "super high performance match" and "lousy match" resulting in the SAME gain.

Let me guess: you just stopped reading after "one pixel", right?

[Joseph Mallan]

Paigan, on 28 September 2015 - 01:42 AM, said:

Ah, good old Joseph.

As almost always, you fail to see the point.

What he said was:


It's about the comparison of "super high performance match" and "lousy match" resulting in the SAME gain.

Let me guess: you just stopped reading after "one pixel", right?

Paigan how large is each tier?

100 points
1,000 points
10,000 points?

Depending on how many points from bottom to top will determine how much the level bar grows/falls from our efforts.

Edited by Joseph Mallan, 28 September 2015 - 01:50 AM.

[Signal27]

Soultraxx, on 27 September 2015 - 11:11 AM, said:

Anyone else feeling this?

I did at first. But I just took another good look at it and realized how far I've inched closer to the right side of the bar in the past few days.

[Aggravated Assault Mech]

Paigan, on 28 September 2015 - 01:38 AM, said:

All I say is: PSR insanity

Your problem goes in the same direction.
Current PSR does not really rate a pilot's performance in the match, it just rates if your team won or not, just like ELO. Plus some teeny-tiny pilotic-specific corner-case-addon that makes up like 1% or so.

Win rate should be used because past win rate is the best predictor of future win rate. Even if a good player gets carried with 0 damage, or loses with 800 dmg once in awhile, they'll still win more often on average than a bad player (that's fundamentally how you identify them as a good player, obviously).

It's not even clear how great the correlation is between score and win rate. Being that it's further removed from the actual objective of the game, there is zero reason to use it as the biggest contributor to PSR. The only reason would be to just placate people that don't understand the implications that system would have, based upon how it has played out in similar games.

If there's any problem, it's that PSR should probably be a rolling average of the past 500 games or something like that, rather than this cumulative system.

[Joseph Mallan]

vnlk65n, on 28 September 2015 - 01:58 AM, said:

Win rate should be used because past win rate is the best predictor of future win rate.

So if you get stuck with a lot of bad teams that should affect the effort you put in? Victory and defeat are done by the team as a whole not just you or me.

[Aggravated Assault Mech]

Joseph Mallan, on 28 September 2015 - 02:03 AM, said:

So if you get stuck with a lot of bad teams that should affect the effort you put in? Victory and defeat are done by the team as a whole not just you or me.

You are a contributing element of that team, and over a statistically relevant number of games that level of contribution manifests itself as winning more or less than average.

You can't win every game, but being able to pull a win out of a seemingly hopeless match is part of what separates great players from the plebes.

It is important that the sample size for PSR is large enough that bad (or good) streaks aren't enough to dump a player from a tier that they're actually well suited to.

[El Bandito]

Joseph Mallan, on 28 September 2015 - 01:49 AM, said:

Paigan how large is each tier?

100 points
1,000 points
10,000 points?

Depending on how many points from bottom to top will determine how much the level bar grows/falls from our efforts.

Don't know the exact value but after playing about a dozen matches per day, this is how much I have progressed in five days. So around 60 matches at least. I won slightly more than I lost--about 55-60% WLR.

Five days ago.

Today.

.

Edited by El Bandito, 28 September 2015 - 02:23 AM.

[Random Carnage]

Ha! I'm virtually identical to El's rating, to the percent. Same relative gain too.

[Satan n stuff]

Joseph Mallan, on 28 September 2015 - 01:49 AM, said:

Paigan how large is each tier?

100 points
1,000 points
10,000 points?

Depending on how many points from bottom to top will determine how much the level bar grows/falls from our efforts.

I imagine the tiers are all the same size, otherwise we wouldn't get the bell curve distribution the forum poll indicates. Given my own progression, it doesn't have to take very long to move up a tier, maybe a few weeks if you're climbing slowly.

[Alistair Winter]

The progression is a bit slow, but it's also worth considering that most of the people who play regularly will have gone up a tier in a few weeks or a couple of months. And when you hit Tier 1, that's it. That's the peak of the mountain. It's not like WoW, where they will raise the level bar from 60 to 70 in the next expansion.

My biggest concern is the fact that PSR increases so dramatically fast when you're playing in a group. It doesn't seem to account for the fact that you're a part of a larger group. If you wanted to make some sort of scientific estimate of skill, then you'd want to adjust the value of a match score to account for certain variables, such as being part of a 12-man group. In other words, getting a 400 score in a 12-man group would not be as meaningful as getting a 400 score while you're on the losing team of a 0-12 stomp in the solo queue. However, I don't think the PSR accounts for such considerations.

Edited by Alistair Winter, 28 September 2015 - 03:17 AM.

[Kyynele]

I think it's a good thing the PSR doesn't have super high rewards for super high match scores. 1 good match is no indicator that you're a god in your tier. Everyone gets those every now and then, and getting a lucky game shouldn't have a huge impact on one's rating. If you're literally capable of doing that in every match you play, you're bound to win way more matches than you lose, and you won't lose any rating on losses either, which means you're going up, fast and steady.

[Lugh]

Kyynele, on 28 September 2015 - 04:02 AM, said:

I think it's a good thing the PSR doesn't have super high rewards for super high match scores. 1 good match is no indicator that you're a god in your tier. Everyone gets those every now and then, and getting a lucky game shouldn't have a huge impact on one's rating. If you're literally capable of doing that in every match you play, you're bound to win way more matches than you lose, and you won't lose any rating on losses either, which means you're going up, fast and steady.

Only it doesn't seem to consistently reward consistency either.

I've been watching closely as I tend to score 300-500 damage a game win or lose with a few outliers here and there, and the 300-450 range on a loss is still a downgrade.

The outliers of less than 300 damage on a win is still an increase and even a 50 damage 1 kill 60 Match rating on a win is still an increase.

So I am not sure how one would lose rating on win....

But they should if they are consistently less than 300 damage with no kills and few assists.

[Mechteric]

Progress Quest: The hunt for miniscule experience gain and marginally better loot!

[MechWarrior5152251]

PGI better make tiers secret before half of the population commits suicide out of depression at their underhive status....

[RoboPatton]

It's an EXP system... Not a measure of how good you are.

It's more a measurement of how good you should be.

Maybe it's better that less weight is put on a quantifiable value of the player by PGI, who, let's face it, I don't want trying to rate me the one they do a game-score. The variables required to determine a "good player" vs an "experienced player" is huge, and filled with nuance that I don't trust them to do accurately anyways.