149 replies to this topic

[Star Dust]

Thank you 7ynx for taking the time to write up this information for the community. My experiences have been consistent with your findings.

Right now this system is too heavily weighted for wins for a 12v12 game. Wins matter more the lower the team size. Obviously it is possible to improve PSR on a loss, but consider the recent patch changes where the games are no longer as close (solo queue I am referring to here). Last patch or the patch before I was getting great games most of the time, but from what I understand, some players queue times were suffering. I never noticed it. That said, when games are no longer as close, the matches end quicker and therefore there will be less opportunity to make the score requirements to improve on a loss. Hopefully this will be considered and balanced out accordingly with time.

Perhaps additional factors should be used for both wins and losses, or maybe even just losses. On a quick loss damage dealt per time is a good metric, but falls short of being the perfect measure because it favors tonnage. Factors could be introduced to take this into consideration though. Certainly the developers have a large pool of data to evaluate damage per tonnage over given match times. Number of scouts could be good for lights, uav kills, uav launches and many other items that play into match scores could and should be more heavily considered.

Don't get me wrong, if you have the ability to be vocal on comms, and can lead a team as a solo player it is currently possible to consistently improve your PSR by guiding your teams collective decision making... even that though depends on a team to listen. If you don't have the ability to vocalize (no mic, have to play quietly for kids sleeping nearby, etc..) positioning importance, the essentials of establishing lines at given locations, or announcing flanks you cannot sway the difficultly matched games enough to extend the match time on quick losses to improve PSR.

Edited by raarz, 03 October 2015 - 06:15 AM.

[Not A Real RAbbi]

raarz, on 03 October 2015 - 04:53 AM, said:

Perhaps additional factors should be used for both wins and losses, or maybe even just losses. On a quick loss damage dealt per time is a good metric, but falls short of being the perfect measure because it favors tonnage. Factors could be introduced to take this into consideration though.

Consider this: Damage for a mech of a particular weight class should be weighted based on the weight class of the mech damaged. It's far more impressive for your STK-4N to hit a Locust with that LPL than a Dire Wolf, for instance. And so on. Consider possibly, as a first step, weight class of the piloted mech and that of the mech damaged, when awarding points for the damage. Say, mechs of the same weight class and the adjacent one are worth a certain amount of points per point of damage. A light gets 0.5 PSR points per point of damage to another light or medium. Maybe ramp it up for a 2-class difference (0.625 against a heavy), and even more for the light-on-assault and assault-on-light, say to 0.725.

Then take the time to study accuracy numbers for each weight class against each weight class, and adjust as necessary. Maybe lights are the hardest to hit for ANY weight class, so EVERY weight class (including lights) gets a 0.7 PSR per DMG for hitting them. But assaults are still the longest Time to Kill (TTK), so they still get an adjustment to their PSR points for most damage on kill and killing blow, and a BIG boost for any SOLO KILL against an assault.

Something like that.

[Void Angel]

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

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.

That doesn't actually prove anything, since we know the match score is weighted in invisible ways. You're right to ask for confirmation of contrary data - but you're more sure of your own rightness than your data warrants.

[L Y N X]

Void, It's called conviction, and so far no one has proved me wrong in what I have documented.

[Star Dust]

New data point, 250 match score on a loss made my PSR decrease, sorry I didn't think to screenshot.

[L Y N X]

TYVM Raar, I adjusted the OP by the inequality on the boundary watermarks.

Is it just me? or do others notice this as well... this being that it "feels" like PGI might have altered something around how MUCH PSR is lost or gained when it is lost and gained.

When PSR was introduced I could win one match with a match score of 200-ish and my PSR went up, but if I lost a match with a match score below 100 then my PSR went down an equal or more amount. However, that same scenario tonight it seemed that I could lose 3 matches for every match I won to be a net zero PSR. I have no way to measure than to estimate by counting pixels on my PSR bar, but that is how it feels now... anyone else notice this?

Edited by 7ynx, 04 October 2015 - 11:49 PM.

[Tesunie]

Just ended a match. Team victory. My match score was 100, and my PSR remained the same.

Sad part is, it was a match score of 100 because I got team killed (by accident). Not much of an indicator of player skill if you asked me. (And isn't 100 an equal by your data? Either way, more data!)

Figured it might be worth mentioning.

[Not A Real RAbbi]

Just ended a match about 15-20 minutes ago. WIN, scored a 94 (DAFUQ?!?!?!), died in the process. PSR? = I did little for the team, other than soaking up some rounds from the enemy's big mechs (a KGC, 2 TBRs, and a EBJ, all got a good piece of me). No change to PSR. How badly do you have to screw up in a win, in order to go DOWN in PSR? I mean, that was a pretty AWFUL performance on my part. I personally feel that I should have lost PSR for that one, and I'm ashamed that I didn't do more for the team.

Question: Who thinks we should lose PSR points for getting killed in a match? Discuss, if you please.

[Tesunie]

TheRAbbi, on 05 October 2015 - 09:23 PM, said:

Question: Who thinks we should lose PSR points for getting killed in a match? Discuss, if you please.

For getting killed? No.

Reminder that this is a PvP game, and killing and dying are part of that. In order for one team to win, another must lose. If PSR went down for just dying, I think bad things would happen.

I'd rather PSR just be unlinked to winning or losing. This way, a bad performance on a win doesn't get rewarded, and a good performance (without having to be something spectacular) on a loss doesn't go unrewarded. Then, make what would be considered an "average" score as neutral PSR, and a poor score would always bring PSR down, and a good score would always bring PSR up. (This would actually gauge individual player's skills better, instead of team skill/luck.)

Then again, I've said all this before... I think...

[p4r4g0n]

7ynx, on 24 September 2015 - 08:12 PM, said:

-snip-
PSR goes DOWN a little if match score is <=250
-snip-

FYI but this is incorrect as I have had it stay unchanged in a lost match with score of under 200.

Edit: I was piloting an unmastered panther

The reason I remember the result is that I've seen these numbers thrown around and it struck me as not being consistent with said numbers.

Also, looking at the somewhat inconsistent movement in PSR on my screen, I was of thinking that this might be relatively based on total match scores in a specific match. However, this seems to be contrary to the consensus opinion of the forum.

Edited by p4r4g0n, 05 October 2015 - 10:51 PM.

[Tesunie]

p4r4g0n, on 05 October 2015 - 10:43 PM, said:

FYI but this is incorrect as I have had it stay unchanged in a lost match with score of under 200.

Edit: I was piloting an unmastered panther

Wow. I've gotten higher and had PSR go down on a loss.

You must have hit the right actions for PSR to reward you that way.

[p4r4g0n]

It was not a particularly high scoring match IIRC. So far the only conclusions I am drawing is that PSR is only loosely correlated to the match score in EoM.

Edited by p4r4g0n, 05 October 2015 - 10:51 PM.

[Tesunie]

p4r4g0n, on 05 October 2015 - 10:49 PM, said:

It was not a particularly high scoring match IIRC. So far the only conclusions I am drawing is that PSR is only loosely correlated to the match score in EoM.

Been mentioned as a possibility, and then Satan N Stuff posted a little confirmation from the devs. (But, it was before their new PSR explanation post... so...)

Satan 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.

[L Y N X]

p4r4g0n, on 05 October 2015 - 10:43 PM, said:

FYI but this is incorrect as I have had it stay unchanged in a lost match with score of under 200.

Edit: I was piloting an unmastered panther

The reason I remember the result is that I've seen these numbers thrown around and it struck me as not being consistent with said numbers.

Also, looking at the somewhat inconsistent movement in PSR on my screen, I was of thinking that this might be relatively based on total match scores in a specific match. However, this seems to be contrary to the consensus opinion of the forum.

Paragon, Actually it is still correct, I suspect what you saw was a tie result and possibly it felt like a loss. If I am incorrect then please post some screenshots. On a tie, even a 48 match score is a no change in PSR.
Cheers!

[p4r4g0n]

Sorry I didn't get a screenshot of that but it was definitely a loss. Will try to get one if I see that kind of result again but since I have almost mastered all the mechs from the Resistance pack I bought a year ago, it may be unlikely to replicate that.

[L Y N X]

p4r4g0n, on 11 October 2015 - 07:49 AM, said:

Sorry I didn't get a screenshot of that but it was definitely a loss. Will try to get one if I see that kind of result again but since I have almost mastered all the mechs from the Resistance pack I bought a year ago, it may be unlikely to replicate that.

Thanks Paragon! If the result is repeatable then it will be useful, if NOT repeatable, then the usefulness is questionable.

[Tesunie]

7ynx, on 11 October 2015 - 06:30 PM, said:

Thanks Paragon! If the result is repeatable then it will be useful, if NOT repeatable, then the usefulness is questionable.

Trying to get repeat occurrences on an already hard to acquire result is... hard...
(Obvious statement is obvious. )

[L Y N X]

Tenusie, obtaining less than 250 match score on a loss is not that hard.

[Void Angel]

7ynx, on 03 October 2015 - 01:35 PM, said:

Void, It's called conviction, and so far no one has proved me wrong in what I have documented.

You mean except for the contrary data that you are hand-waving away? It's not up to others to prove you wrong - the burden of proof is on you, and you're claiming more than your data supports. Sorry about your loss - but "conviction" in the face of evidence is called "stubbornness." You've got three separate people on this page alone telling you that they've had results that don't fit within your hypothetical framework - and you've basically just told them, "that's nice, but until I see a screen shot, I'm going to assume my data set is inclusive." That's not reasonable; particularly when the part of your hypothesis being questioned is demonstrably wrong.

You might have identified average breakpoints based on typical behaviors in the observed tiers, and that's great - I applaud your effort, and your intent. However, your intransigent insistence on things you haven't proven embarrasses you, and promotes disinformation in a forum dedicated to helping new players.

[Tesunie]

7ynx, on 11 October 2015 - 11:03 PM, said:

Tenusie, obtaining less than 250 match score on a loss is not that hard.

Maybe for you... But beyond that...

It's hard to get the same result, with the same action rewards in match, to not only get the same match score but to also get the same win/lose and to get the same PSR adjustment.

I had a match loss of 330, PSR went down. I didn't screen the match results. I have no idea what action rewards I took to get that match score and still have PSR go down on a lose.

I've had a match near 250 on a lose with PSR going down as well. No screen, and I don't recall the exact number (which is why I didn't present this, and call it unofficial anyway). I have no idea what action rewards I did to possibly get that.

I recall a match (before I crashed into your thread) where a score of around 150-200 resulted in a "no PSR movement" on a loss. Same thing.


I'm not claiming that your system isn't a good estimate, as it seems to follow an average. However, I don't think it's exact either. I still believe that PSR ranks different actions at different weights, which results in "PSR Match Score being different from your actual Match Score. However, this is much harder to prove, because our evidence (which we were not collecting at the time), is a lot harder to get as we have to play and get matches that conflict with your data, and then report what every action we took. (It's a lot harder than simply looking at Match Score and finding an average adjustment point.)


So, it's a lot harder to get a result that conflicts with your data, as we not only have to lose, we have to do well in match score, and we have to preform the correct action rewards to get the odd results. (Maybe we didn't do enough spotting, and only did damage for the match score. Or...)