149 replies to this topic

[Vlad Striker]

I think average team level affects decision bounds. If you play with high tier you must earn more point and so on.

[Yozzman]

That the rating is depending so much on winning or loosing is crap.
Skill ain't based on winning, just plain and simple

WoT uses this...
http://wiki.wnefficiency.net/pages/WN8

Way more advanced, think PGI can learn from this

[Kerensky98]

Satan n stuff, on 30 September 2015 - 10:34 AM, said:

previous posts regarding PSR made it clear that the exact formula will be kept secret,

By all means please provide some evidence of this.

Because the counter evidence this it's match score is PAUL FROM PGI and his clarification post and his graph that says: match score, match score, match score, match score, match score.

I've been comparing my PSR arrow to 7ynx's post and it's solid.

Edited by Kerensky98, 30 September 2015 - 11:51 PM.

[Kerensky98]

Yozzman, on 30 September 2015 - 11:02 PM, said:

That the rating is depending so much on winning or loosing is crap.
Skill ain't based on winning, just plain and simple

It's a team game. If you stand in back lobbing LRMs from your King Crab you get high damage but that 600 points of damage was spread over many mechs and all their entire upper body, leading to no real kills. Meanwhile your team dies because your massive amount of armor would have been better spent in the front line absorbing fire so your teammates could use direct fire weapons to core the enemy.

Skill is more than just match score, you do need to work as a team. The PSR system Paul graphed out reflects that.

[Satan n stuff]

Kerensky98, on 30 September 2015 - 11:43 PM, said:

By all means please provide some evidence of this.

Because the counter evidence this it's match score is PAUL FROM PGI and his clarification post and his graph that says: match score, match score, match score, match score, match score.

I've been comparing my PSR arrow to 7ynx's post and it's solid.

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.

Edited by Satan n stuff, 01 October 2015 - 07:37 AM.

[Tesunie]

Kerensky98, on 01 October 2015 - 12:34 AM, said:

It's a team game. If you stand in back lobbing LRMs from your King Crab you get high damage but that 600 points of damage was spread over many mechs and all their entire upper body, leading to no real kills. Meanwhile your team dies because your massive amount of armor would have been better spent in the front line absorbing fire so your teammates could use direct fire weapons to core the enemy.

Skill is more than just match score, you do need to work as a team. The PSR system Paul graphed out reflects that.

You forget that there are other things besides damage and kills that can affect a team's performance. In your example with the LRMs, it might not be 600 points of damage concentrated, but LRMs also have a scare tactic in them. When people hear the "incoming missile" warning, they hide. 600 damage, even spread out, can cause an opening for direct fire weapons to take advantage of.

And then, this is also when I mention that LRMs should be mid line, not back line, in the team. They also should (my opinion of course) not boat the LRMs and always carry enough secondary weapons to be able to do some direct fire scalpel work if needed. Secondary weapons also make it possible to take some damage as well for the team, as you can, if needed, go where the fighting is. (The enemy is inside that tunnel? Well, here I go with my direct fire weapons.)

As a King Crab, my LRM version has enough backup weapons to be a viable threat. Probably could still use to tweak a few things, but it's worked fairly well for me so far...


I still don't feel that W/L is a very good metric to work off of. I've had matches were I was doing very well and helping my team as much as I could. I see a Direwolf in the tunnel on Crimson shoot out that side entrance at multiple enemies, and back up blocking the Direwolf behind him. Then, he moved forwards again and blocked a heavy mech from being able to shoot, and overheated right there blocking all support fire. Same match, after I had died, the last mech alive was a Direwolf who was hanging out on top of the dock area, sniping with a single Gauss and a mass of CSPLs.

Now, I'm not saying that the whole team was "bads", as it was still a good match, but a few people disrupted the team effort. Why should I (or any other good performer in that match) be penalized by a drop in PSR because we lost?


Then, we have the ever popular bugs department. I can't tell you how many bugs I've been experiencing as of late, or rather just last night. So many times I was stuck on nothing. I might Jump and become stuck in mid air for a while. I shoot, only to teleport away, miss my shot, and be back in front of enemies I had escaped moments ago. Then I also had... (Last night was not a good night for me nor my game.)

Now, I know it's impossible for the system to discredit a match failure by bug, but it's tiresome to be penalized with a lower PSR to myself and my team because "I got a bug guys. Sorry..." It's no one's fault, but I'd rather me be penalized for the bug which affected my individual performance, rather than affect myself and my entire team because I preformed poorly.

Same can be said if there is a TKer on the team. If a teammate decides to damage/kill teammates (besides the obvious ban-able offense), the whole team suffers for their efforts. The whole team's PSR drops because they probably can't win now after being damaged and/or down a couple of mechs.

Same for any Disconnecters. Be it by bug, issue, or purposefully disconnecting. That also affects a team's ability to win, but doesn't affect an individual's possible performance.


I might wish to remind:
- PSR stands for Player Skill Rating. It's not TSR (Team Skill Rating).
- Match Score is suppose to reward players for team actions. One can, in theory, have a high match score from intense team play, but do poorly in damage and kills.
- PSR takes a form of the match score, and rates different actions differently. PSR is not derived exactly from the match score we see, but from an alternative match score that is weighted by the same actions being scored differently. (This is supported by matches where I had a match score of over 300 in a lose, and lost PSR. Then later, had a score just under 300 on a lose and I remained equal. If it was exactly by match score, it should not do that.)
- PSR, if it's suppose to reflect a players individual skill level, should remove w/l from the equation. Of course, this would mean the numbers to go up, down or remain level would have to be altered as well. Do recall, on a loss you already will have a lower match score, which would already make it harder to go up even if w/l wasn't accounted for.


I don't know about other people, but I tend to find that I often work hardest on a loss. I also find close matches to be more rewarding. However, with this currently system, no matter how hard I might fight, or how close the match, if it was a loss I'll probably be losing PSR.

[Anachronda]

7ynx, on 24 September 2015 - 08:38 PM, said:

Tesunie,

All my testing indicates that a losing match and match score of 330 would generate a NO CHANGE to PSR as a result.
Your win is right in line with my testing.

Thanks for sharing.

That's strange because everyone else loses PSR when they score 300+ on a loss. I know I do, every single time. As for losing PSR on a win, you absolutely cannot lose PSR on a win. The worst you can do, on a sub-100 score, is stay the same.

[Anachronda]

TheRAbbi, on 29 September 2015 - 05:55 PM, said:

Yeah, maybe we're all PSR-obsessed right now. Could be. But video games have conditioned that in us for DECADES. Who played PacMan just for the fun, and never paid attention to the Top Score or to raising their own personal best score? Huh? NO ONE. That's who. Now we know if we're on the leaderboard (Tier 1 or not?), and we know what needs to improve to get us there. SCORE. That is, PSR.

I guess for me, the reason I was excited to hear about PSR is I expected it to really be a measure of skill in some way and the new scoring system to provide some metric other than raw damage, which is actually a bad metric. Think about it. One of the nest things you can do to raise match score is find some disconnected/afk mechs and farm them one component at a time while your team is trying to fight live mechs. Doing more damage doesn't really mean you're helping the team unless that damage was done to the right parts of the right mechs to cause kills. So it should be balanced.

The new scoring system rewards lots of actions that help the team, but damage weighs into it more than it did in the old system. FIVE TIMES MORE. I mean, I do want to up my damage, since that is what everyone measures by anyway and hurting the other mechs is what you want to do, but still.

As for PSR, it has nothing to do with skill except for the extreme upper bound. Someone who can do 1000 damage consistently every game, win or loss, will go up, and they should. But someone who gets 100 matchscore or less will go up as long as they group with people who win. Anyone who can't make a 400 matchscore on a losing game (if that number is right) will go down every loss. So it's not what I hoped for; it's not a measure that is going to show me if I am getting better or not, and it's not a metric that is going to encourage people to do things that help the team, especially since team damage doesn't seem to be hurting things anymore.

I think that the numbers should be the same for win as for losses. A score of 100 is abysmal in the current system and you shouldn't be rising from it. And you shouldn't lose PSR when you make a score (300) that has been the GOAL of challenges. From Paul's post, weighting the winning scores the way they did was on purpose with the thinking that it will cause people to rise on average. I think that's kind of dumb, really. People should rise if they do better, period. The current system doesn't reflect that.

[L Y N X]

Anachronda, on 01 October 2015 - 04:33 PM, said:

That's strange because everyone else loses PSR when they score 300+ on a loss. I know I do, every single time. As for losing PSR on a win, you absolutely cannot lose PSR on a win. The worst you can do, on a sub-100 score, is stay the same.

Anachronda, That part you state about everyone else losing PSR on a loss with a match score of 250 to 399 is either an exaggeration or just wrong. Please check that you are noting the MATCH score and not the DAMAGE score. Match score of 250 to 399 on a LOSS is a NO CHANGE to PSR. It has been confirmed on Tiers 2,3,4 many, many times. This is a verified fact.

I very much liked your follow up post and that is also where my feeling/opinion lies. I do not think PGI is done with tuning the PSR settings yet.

[Tesunie]

7ynx, on 01 October 2015 - 07:41 PM, said:

Match score of 250 to 399 on a LOSS is a NO CHANGE to PSR. It has been confirmed on Tiers 2,3,4 many, many times. This is a verified fact.

I probably wouldn't be too quick to call it a solid proven fact... I've had matches that seem to contradict your fact, where my match score was supposedly high enough on a loss to make me remain equal, but instead I lost PSR.

Maybe I'm special. Maybe I misread (don't really think I did) my match score. But, I wouldn't be so quick to call anything an absolute.

[Satan n stuff]

7ynx, on 01 October 2015 - 07:41 PM, said:

Anachronda, That part you state about everyone else losing PSR on a loss with a match score of 250 to 399 is either an exaggeration or just wrong. Please check that you are noting the MATCH score and not the DAMAGE score. Match score of 250 to 399 on a LOSS is a NO CHANGE to PSR. It has been confirmed on Tiers 2,3,4 many, many times. This is a verified fact.

I very much liked your follow up post and that is also where my feeling/opinion lies. I do not think PGI is done with tuning the PSR settings yet.

Tesunie, on 02 October 2015 - 07:48 AM, said:

I probably wouldn't be too quick to call it a solid proven fact... I've had matches that seem to contradict your fact, where my match score was supposedly high enough on a loss to make me remain equal, but instead I lost PSR.

Maybe I'm special. Maybe I misread (don't really think I did) my match score. But, I wouldn't be so quick to call anything an absolute.

The thing is, 7ynx hasn't given us any raw data so we can't really decide whether or not this hypothesis is sufficiently proven. There are users here who have given accounts that would disprove it, but we haven't seen any raw data on that either. As it is now, we have no idea of the volume of data 7ynx may or may not have so calling this explanation sufficiently proven is scientifically BS.

To 7ynx, release your data if you feel you have enough and I do mean release all of it, so we can judge it for what it is.
To everyone else, try to gather as much data as possible and if some of it proves 7ynx wrong, we'll have our definitive answer, if not we'll have one until it's proven wrong.
This is how you do science.

[Tesunie]

Satan n stuff, on 02 October 2015 - 08:40 AM, said:

The thing is, 7ynx hasn't given us any raw data so we can't really decide whether or not this hypothesis is sufficiently proven. There are users here who have given accounts that would disprove it, but we haven't seen any raw data on that either. As it is now, we have no idea of the volume of data 7ynx may or may not have so calling this explanation sufficiently proven is scientifically BS.

To 7ynx, release your data if you feel you have enough and I do mean release all of it, so we can judge it for what it is.
To everyone else, try to gather as much data as possible and if some of it proves 7ynx wrong, we'll have our definitive answer, if not we'll have one until it's proven wrong.
This is how you do science.

I'll try to screen shot things, but it's kinda rare for me to "do well" when my team is losing. (I've also been playing light mechs as of late... because I'm crazy!)

You are right though. No matter how much data one has gathered to support their claim, all it takes is one piece of proven evidence saying it is false to disprove all the other evidence's support of the concept.


I will say, I do feel 7ynx has placed a lot of work in this, and he's done great. However, I don't believe that the system is working as he is thinking it is. Even if he is right, or even wrong, I commend his work. It's a good thing to pursue an idea on how something works and seek out the evidence. If nothing else, we find out "one more way to not make a light bulb".

[Void Angel]

Tesunie, on 24 September 2015 - 08:32 PM, said:

Would like to mention... I've had a match score over 300 (I believe it was 330ish) and lost a match. PSR when down. (Then I won a match with a match score of 200ish, and it went up...)

Only note I wish to add, if it helps any.

7ynx, on 24 September 2015 - 08:38 PM, said:

Tesunie,

All my testing indicates that a losing match and match score of 330 would generate a NO CHANGE to PSR as a result.
Your win is right in line with my testing.

Thanks for sharing.

Tesunie, on 24 September 2015 - 08:41 PM, said:

Just saying. Don't now why then, but I had a red down arrow when I lost. I only had the yellow equal symbol once when I had a rare (for me) match score in the 400s. (Which does match your data.)

Just want information to be as accurate as possible, even if I happen to be wrong.


I just don't feel it's right to preform better and lose and drop in PSR, and then perform worse and win and increase PSR.

Remember that PSR ratings changes are not based on your raw match score - the system weights different components of that match score in an attempt to measure behaviors associated with team play. Essentially, this means that your actual match score for the purposes of PSR is secret, too; your raw match score from the end screen can only be used to approximate the "real" one. For this reason, it's not surprising to see an occasional outlier like yours - such results will occur when a player's match score is inadvertently comprised of enough lightly-weighted categories.

The guidelines in the OP may still be accurate as a rule of thumb, but of course they are not based on complete data and are thus almost certain to not cover all possible cases.

[Void Angel]

Kerensky98, on 30 September 2015 - 11:43 PM, said:

By all means please provide some evidence of this.

Because the counter evidence this it's match score is PAUL FROM PGI and his clarification post and his graph that says: match score, match score, match score, match score, match score.

I've been comparing my PSR arrow to 7ynx's post and it's solid.

That post is a list of clarifications - it's not a comprehensive explanation, and is not intended to be. However, since he's referencing the basic explanation (from the same source) of how PSR is going to work, your demand for sources is like asking a classmate to prove that electrons have one negative charge while doing homework. His first response is not likely to be "oh yeah, it's here on page 183," but rather, "didn't you read the textbook?"

[Tesunie]

Void Angel, on 02 October 2015 - 09:34 AM, said:

That post is a list of clarifications - it's not a comprehensive explanation, and is not intended to be. However, since he's referencing the basic explanation (from the same source) of how PSR is going to work, your demand for sources is like asking a classmate to prove that electrons have one negative charge while doing homework. His first response is not likely to be "oh yeah, it's here on page 183," but rather, "didn't you read the textbook?"

7ynx, as much as I appreciate his work here, did start to call all this "verified fact". As a general guideline, it's a good concept. Calling it "fact" with a certainty I think is where people are starting to have problems with him. (Hence, Satan N Stuff's response for sources. Can't call it a verified proven fact without sources.)

There was another reason I posted someone else's quote in this thread, as it backs up what you said in response to me. Match Score is a decent rough guideline, but it appears as though PSR "match score" takes different actions into stronger consideration. (For all we know, Damage may not even be counted in "PSR match score", or it may be rated for 75% of it.)

PS: Did you quote the right person here? Sounded more like you were responding to Satan M Stuff from the written response.

[Void Angel]

Er, no, I quoted the right people. What I'm trying to point out here is that your datum isn't contradictory to the thesis if you remember the hidden weighting system and approach it as a generalization. If 7ynx is insisting that your datum is wrong because his thesis is scientifically proven, then yeah, that's not right.

Edited by Void Angel, 02 October 2015 - 11:41 AM.

[L Y N X]

Guys this is a guide meant to share my findings with others who want to know the watermarks for PSR. you can certainly disagree, but I have plenty of data that indicates the watermarks that I've posted are accurate until PGI changes them. Then they will not be accurate.

Tesunie posted more than once that he may have been mistaken on his match score of 300-330 on a loss and PSR went down. Look just go and test your hypothesis if you think I am wrong. I have already done so. 250 to 399 match score on loss is a no change for tiers 2,3,4 I have no data for tiers 1 and 5, sorry.

[Tesunie]

7ynx, on 02 October 2015 - 04:37 PM, said:

Guys this is a guide meant to share my findings with others who want to know the watermarks for PSR. you can certainly
Tesunie posted more than once that he may have been mistaken on his match score of 300-330 on a loss and PSR went down. Look just go and test your hypothesis if you think I am wrong. I have already done so. 250 to 399 match score on loss is a no change for tiers 2,3,4 I have no data for tiers 1 and 5, sorry.

Still testing. If it comes up again, I'll try to screen shoot it. problem is, I need to not only lose, I also need to do well while I lose. I'm sure no one could understand that one?

Edited by Tesunie, 02 October 2015 - 04:49 PM.

[L Y N X]

Check out the screen posted by Titannium post #46

329 Match score and No Change to PSR... Tier 5 datum.

[Tesunie]

7ynx, on 02 October 2015 - 05:09 PM, said:

Check out the screen posted by Titannium post #46

329 Match score and No Change to PSR... Tier 5 datum.

I don't believe tier ranking has anything to do on the numbers needed to rise or drop your PSR on given results. However, I will say I did seem to gain more PSR on my bar when I was tier 4 than I am now that I'm tier 3. This leads me to believe that the tier PSR "numbers" are different amounts. AKA: Tier 3 numbers are probably a larger set than tier 4's set of numbers. Thus possibly making it easy to raise from tier 5-4 and into tier 3?

Also, it's been noted previously that PSR would be based on a different scaling of match score than the match score itself. Though they are close, PSR might rate the same actions differently. (Then it was just summarized into "match score" in their latest response... So either case could be true.)

Still gathering data myself, but my experience seems to correlate with PSR scores being derived from, but at the same time different from, match scores. (Thus how I could experience a 330 match score with a lose to PSR, and someone else had the same match score and had an equal/stationary result. However, once again, there is a possibility I was mistaken, and I'm waiting to see if it happens again.) Whenever I present data, I always try to give the chance of a mistake or being mistaken. So, I'll keep my eye open, as all data is a good thing.


(And if it doesn't happen again, I'll just alter a screen shot to prove myself correct... Oh, wait. You didn't read this.)