277 replies to this topic

[Kubernetes]

The system seems fine. The bigger factor in moving up or down is wins and losses. Do you really think your win-loss record over time is completely random? The steering-wheel trash-build player will not advance up to Tier 1 because his teams will always be at a disadvantage and he will lose more often than "tryhards." As for the complaint about Light mechs not scoring as high, it's still possible to contribute to a win without doing massive damage. My Lights have lower damage than my other classes, but they have a higher win/loss percentage.

[50 50]

@Mors.

That is certainly true.
However the levels were reset and now with the visibility we are able to guage what is going on.
We are no longer needing to guess or assume what tier we may be or how we are tracking.

@Kira
It is assumed that just by playing enough you will increase in tier as your skill level should be increasing.
Provided you are then winning more matches than you are losing and are achieving a regular match score of 100pts every time you win, then yes, eventually you will increase your tier..... but it is going to take a long time as the increment you raise needs to be greater than what you are losing by.

I believe this system is setup in a fair and balanced way rewarding both good teamwork and personal skill.

Edited by 50 50, 24 September 2015 - 07:32 PM.

[TLBFestus]

Just to clarify my position on this breaking news;



I don't care.

[Alzarns Fire]

What is ridiculous is that in order to go UP when the team loses you have to get a super score.... try that when you are dropping constantly against 8 man tier 1 competitive teams. Essentially lot match, despite highest damage an score, means drop in PSR.

As far as not dropping Tier 1 against tier 4/5s... doesn't work. We had 3 tier fives in our team last night and we were dropping against tier 1 teams. Why... the MM can't find players for the tier 1s in a reasonable time so just drops them against whatever meat is going.

I agree with the suggestion that regardless of win/loss, if you do better than your teams average you go up, you are average you stay, you are worse than average you go down. Simple.

[Xmith]

Nightmare1, on 24 September 2015 - 05:29 PM, said:

I rather thought he did a good job with the clarification. I dislike that it's so biased towards wins since I've done spectacularly well on a few losses recently, but oh well...

Since when did winning become a bad thing?

[Tesunie]

Alzarns Fire, on 24 September 2015 - 07:19 PM, said:

What is ridiculous is that in order to go UP when the team loses you have to get a super score.... try that when you are dropping constantly against 8 man tier 1 competitive teams. Essentially lot match, despite highest damage an score, means drop in PSR.

As far as not dropping Tier 1 against tier 4/5s... doesn't work. We had 3 tier fives in our team last night and we were dropping against tier 1 teams. Why... the MM can't find players for the tier 1s in a reasonable time so just drops them against whatever meat is going.

I agree with the suggestion that regardless of win/loss, if you do better than your teams average you go up, you are average you stay, you are worse than average you go down. Simple.

Group queue is not affected by PSR. Only Public Solo queue is affected. (Mostly because you can have T1s teaming up with T5s as a group of their own.)

[Nightmare1]

Xmith, on 24 September 2015 - 07:24 PM, said:

Since when did winning become a bad thing?

Winning itself isn't bad, I just wish that PSR was more performance based rather than winning based. Wins are often luck-based since the RNG MM determines who your teammates will be. You may get a team that works well together or you may not; that's outside your sphere of influence. How you perform in-game, however, is directly within your control. Thus, I simply think it would be better if the Performance Bonuses carried more weight than your W/L ratio.

[Tesunie]

Xmith, on 24 September 2015 - 07:24 PM, said:

Since when did winning become a bad thing?

Winning isn't a bad thing. It's winning to make your "Player Skill Ranking(level)" go up (or losing to make it go down).

PSR is suppose to be a singular player skill rate. A player can be skilled, and still lose because there are 23 other players in the match. Right now, you can preform your same average (say, 250 match score), and go up or down solely based on if your team won or lose. This isn't indicative of your personal skill (which 200-300 match scores is probably "average"), but just if your 11 other players could pull a win, or the opposing 12 other players couldn't.

PSR needs to be unlinked with winning or losing. If you do well, it should go up because you did well. If you did poorly, it should go down. Not, if you did poorly, but the 11 other players on your team could still get a win, your PSR remains the same or even goes up.

[Jman5]

Wins where you do jack-squat should lower your PSR or at the very least not move it. Moving a player's Skill Rating whenever they win will just allow players to ride the coat tails of a high performing premade to tier 1. The entire point of PSR was sold to us on the premise that this would not be the case.

There is some miscommunication going on at PGI because now they're saying well actually this isn't what "we" intended. Patch notes and Russ say one thing, then Paul comes in and asks: Why did you think that?

Players want good players matching against good players, mediocre players matching against mediocre players, and bad players matching against bad players. Simply pouring tons of manhours into a game while being in the top 51% or playing exclusively on a Tier 1 type premade does not make you a good player.

[Be Rough With Me Plz]

50 50, on 24 September 2015 - 06:51 PM, said:

DEFEAT
I am not sure what is needed to still achieve an increase in PSR when you are on the losing side but let us assume it may be up around 600pts.

I'm going to assume a matchscore of 500+. I had a high 500-600 match on a loss where my PSR went up with 5k/2a/811dmg. I just had another match with a score of 499 5k/1a/411dmg and my PSR remained the same.

[Vlad Ward]

Tesunie, on 24 September 2015 - 07:24 PM, said:

Group queue is not affected by PSR. Only Public Solo queue is affected. (Mostly because you can have T1s teaming up with T5s as a group of their own.)

Group queue is affected by PSR.

Just not by as much. Because it's working with 1/3 the population and ridiculous team size and 3/3/3/3 buckets, so it allows larger gaps in PSR.

Also mixed tier groups are just treated as an entity with tier equal to the unweighted mean tier of its members.

Edited by Vlad Ward, 24 September 2015 - 07:41 PM.

[50 50]

@Be Rough With Me Plz.
Ok, that is quite interesting to hear.
Thank you for putting up that information, I will update my earlier post.

@Jman @Nightmare1 @Tesuni
The point is, you will be rewarded for good personal effort, even in a loss.
You will not automatically raise your PSR just because you were on a winning team.
It is CW that is currently not affected by PSR.
Also keep in mind that with the patch everyone has been changed to tier 5 so for a while you are going to come up against experienced and skilled pilots that may in reality be several tiers above your own level.
This will eventually sort itself out but we just need to keep playing in the mean time.

I will re-write the figures here as I have observed:

VICTORY
Score over 100pts in the match to get an increase.
Expect to get a higher increase with a higher match score.
Score under 100pts and your PSR will not be affected.

This means the more you participate and contribute to the success for your team, the more likely you will win and the more likely you will get over 100pts from the match.
Sit around and not do much... and you are not going anywhere.

DEFEAT
If you score under 300pts in the match, your PSR will decrease.
Expect a greater decrease if you score particularly badly.
Between 300pts and 499pts and there is no change to your PSR.
Score 500pts or more in a losing side and your PSR will increase.

This means that you can have a good game on a losing side and still be rewarded for it.

Give it time.
Exceptional players will be quickly moving into higher tiers as their increase in a win will be higher and they are likely not being as affected or are still improving their PSR even in a loss.

Edited by 50 50, 24 September 2015 - 07:53 PM.

[Tesunie]

Vlad Ward, on 24 September 2015 - 07:40 PM, said:

Group queue is affected by PSR.

Just not by as much. Because it's working with 1/3 the population and ridiculous team size and 3/3/3/3 buckets, so it allows larger gaps in PSR.

Also mixed tier groups are just treated as an entity with tier equal to the unweighted mean tier of its members.

Actually... That's what I meant to say... I just REALLY didn't say it right. (Was trying to say it isn't as affective in group queue. So you are completely right. My bad.)

[Kubernetes]

Tesunie, on 24 September 2015 - 07:28 PM, said:

Winning isn't a bad thing. It's winning to make your "Player Skill Ranking(level)" go up (or losing to make it go down).

PSR is suppose to be a singular player skill rate. A player can be skilled, and still lose because there are 23 other players in the match. Right now, you can preform your same average (say, 250 match score), and go up or down solely based on if your team won or lose. This isn't indicative of your personal skill (which 200-300 match scores is probably "average"), but just if your 11 other players could pull a win, or the opposing 12 other players couldn't.

In one game, yes, but over many games that player's skill will express itself in a good W/L record. W/L will also account for little things that people do to help their team win, but are not easy for the game to identify and score. You'll notice in some of the other threads where people post their stats that higher tier folks have better KDRs *and* W/L percentage. It's not random over time. Better players will win more often.

[Bloody]

it is almost impossible to go down in PSR as well tbh, last 10 games i lost most of them and with sub 100 scores and still my bar never goes down

[sycocys]

So he basically explained nothing new and totally avoided having to discuss the fact that getting rng'd with poor/troll players will do nothing but continue to drag you down no matter how well you play.

Doesn't really matter if you play a million games if MM constantly teams you up with poorly skilled players and matches them against players of much better skill on the other side - you will never advance in tier. Hence the problem of basing a pilots skill on the performance of the other 11 players he is randomly dropped with.

[Tesunie]

Kubernetes, on 24 September 2015 - 07:57 PM, said:

In one game, yes, but over many games that player's skill will express itself in a good W/L record. W/L will also account for little things that people do to help their team win, but are not easy for the game to identify and score. You'll notice in some of the other threads where people post their stats that higher tier folks have better KDRs *and* W/L percentage. It's not random over time. Better players will win more often.

You are 1 out of 24 players in a match. Though sometimes a single player can make the difference, I've won matches (in he new PSR system) with two players disconnected on my team. And it wasn't because the other team took it easy on us either.

Other games, I've fought hard, was top or near top of my team, and my team lost by getting systematically eaten one by one. I tried to help whom I could, but can't save them from being cornered by the enemy. PSR went down anyway.

A win lose bias on pilot ranking is showing the skill ranking of your team, or even luck. It does not show the skill ranking of any individual players performance. So, there ends up being a 1 to 24 impact on a win or a lose being what determines my PSR. This is more luck related than skill. (And don't test my luck. I play a lot of war games that involve dice (warhammer and X-wing to name two). Lets just say, I seem to love my 1s and my blanks. (1s are bad in Warhammer. Blanks are bad in X-wing.) For an example of my bad luck, I had 12 dice to essentially roll over the course of one attack phase in X-wing. Each attack dice has exactly a 50% chance of being a hit or a blank. I rolled 11 blanks and 1 hit.)

I'd rather have something that's more "pilot" ranking, than "team" ranking. Otherwise, call it "TSR" for "Team Skill ranking". Honestly, this is just Elo, but with a few "loop holes" you can use to bypass a win or lose if you are "way over exceptional/deplorable" in your in match performance. Besides those two loop holes, it's still win or lose ":Elo". (Which it's already stated that Elo works great in a 1v1 system, but not so well in a revolving team based system.)

Bloody, on 24 September 2015 - 08:15 PM, said:

it is almost impossible to go down in PSR as well tbh, last 10 games i lost most of them and with sub 100 scores and still my bar never goes down

It is, just it will go up faster than it will go done, from my experience. A single win countered several losses for me.

[GRiPSViGiL]

Wow, it must really be important for people to think this is a ranking system rather than an XP bar that can decay if you go inactive.

[Tesunie]

GRiPSViGiL, on 24 September 2015 - 08:30 PM, said:

Wow, it must really be important for people to think this is a ranking system rather than an XP bar that can decay if you go inactive.

Personally, I don't think PSR should go down for inactivity. However, I do want it to accurately reflect (within as much reason as possible) a players individual skill level. If I'm too high (say because I took a long break and is rusty), then I would want and expect it to go down. If I'm doing well and getting better, I'd like it to go up. If I'm right where I kinda belong, then I'd rather it float evenly where it is.

I just (and people can disagree, as it's my opinion here) don't believe linking wins or losses to the equation is going to give these results.

[Samedi Wretch]

Fundamentally the PSR has to be strongly linked to W/L. This is because its job is not to buff our egos but to make COMPETITIVE MATCHES. And there's no better measure of a winner than winning especially as the data sets get really large. Another bonus of this is it smooths out the imbalance caused by tonnage. If it weren't this way practically all T1+2 players would exclusively drop in assaults and heavies.

However, the systems bias toward rising in tiers is a little troublesome. IMO an ideal ranking system would have the total rise resulting from one match equal the total fall. The question is then how you chose to allocate that to the players of a match. But realistically we have no idea how strong that bias is and its real job is to protect the noobs, not create an ideal ranking system. This is seriously essential for the long term viability of this game.

I'm sorry this is dissatisfying for the tryhardiest of us out there who want this system to be an ideal reflection of their personal glory. It was never going to be that, but what I believe it is is a slick and effective way to make competitive AND fun matches for everyone regardless of skill level.