7 replies to this topic

[Seelenlos]

Hi,

PSR is good, but:

1. every time such features get active there must be a reset.
With ELO every one was in the middle and we got the suck of ELO in worst manner mthematicaly
Now we had the chance to slowly populate the tiers but with this approach you got us again in ELO ranks (methodicaly and mathematicaly wrong - it is a new pool)

2. the score are too high
before that PSR the highest scores where 157 or so, now 500?
This will not help to seperate the poor from the rich (yes this is it at the end)
For 500 steps the tiers are to few. Then we had to have 10 to 12 tiers!

3. it is not clear which tier fighting which tier.
more tiers than 5 for more separation from 3 level deference or only 2 level difference fights

4. statistics:
how fast are they implemented, RT, every 3 matches, Day statistic, weekly ????

5. what a re exceptional scores in a losing team?
I must know that for at least make more scores from the rest of the idiots to ake myself a little better than the losers in that team.
You must exactly show the numbers which are sucking and numbers which separates me from the losers to reach them in a match !

I think we must all optimize on that, then it will get a nice curve and learning and challenge for all !

Edited by Seelenlos, 21 August 2015 - 08:19 AM.

[Seelenlos]

The only fast adjustment PSR needs is redusing the damage bonus.

Damage is all to get the kill, but in my example 720 match score for my crab with 1020 damage is too much bonus toward the damage.

For exceptional damage scores above the 600 mark this must also reflect that the dealing this damage is only possible because others who became damage shielded the damage-dealer so he/she was not that much under fire as the other mechs geting the 200-400 damge and dead in match.

This is my observation, So giving me with my KC 720 match score 5 kill, 10 assists and 1020-1050 damge / but still losing 7 friendly mechs means i was not the one who was got that much damage, cause others took the damage helpign me deliver my damage.

There for A. some amount of my PSR must go to the others as a team-bonus or B. my damage is not that holy to give it that much PSR bonus, cause it was not realy ONLY my work.

I prefer A. for the winning team, so the participating players are rewarded through their teamplayers to become the possibility delivering their superior damage.

Any other alternates i have overlooked?

Edited by Seelenlos, 27 August 2015 - 01:40 AM.

[Seelenlos]

I specially come to this conclusion, when i think of meta-DWs.
I can not correctly put them in PSR schema, as sometimes is "feel" this mech-eater 4x gauss sniping mechs with help of his ecm-cover from far away and on the other hand i see some really browler DWs in middle of the pack dealing the heaviest damages and kills.
For my sense the browler DWs are earning more PSR than the Gaus-DW as they risk they are also taking to go in front of enemy.
On the other hand the Gaus realy makes his sniper role, but the risk of losing is higher if the team in front is decimated fast.

So how to handle that?

LOL i am not talking about LL-ECM-Ravens at all.

Thanks to PSR and the game mechnics they are getting obsolate more than ever (their pilots only are so silly, did not mentioned it)

And by the way, who the f...un was the first id..oll to implement that build?
(Such a lowest performance mech-build for any match and team, near the 4x Lrm 10 Jaeger)

[Lily from animove]

tripleposting? because edit button is lostech?

1. a start form the elo matches as we have it now is still more accurate than a complete restart, only a few people would have been compeltely judged wrong.
2. scores are not too high, they just have a different scale now.
3. its just not visible. Except that T1 is nto meeting T4/5
4. As soon as Tiers are visible, we will instantly know this. It's alos not that important if it happens daily, per hour, PGI's magic 15minutes or realtime.
5. when its visible you will see this, givne its Realtime.It's not necessary to know, but would be nice to know. Also, when you see many idiots, you are probably one of them. otherwise you would not constantly meet idiots.

how much score we need to drop and raise is what we soon will know, because once the PSR is visible we can check it after every match and easily make estmations.
What currently is a bit sad is indeed damage having a too high influence on scoring. So either this must be lowered, or they need to raise the other Score bonuses

Dealing exceptional damage does also not always imply someone tanked for you, sometimes buildings and rocks do that job or just finding a good location where the others can not easily reach you. So when i poke at corners and hills I cna also "soak damage" but not getting damage. This distraction helps me and my team to deal damage as well without anyone in the own team receiving damage. actually, taking "tanking" into account is a polarised way to judge "efford"

imagien 3 examples:

Probob meets Randomaimrandy. He shreds radomaimrandy fast, low damage, some Solokill scores, medium scorings for Probob sicne he killed fast and efficient. Low scores also for "tanking low amounts fo damage, since he killed fats and efficient. His result will be some medium scorings form this encounter. Low scores for Randoaimrandy.

Randomaimrandy meets Randomaimrudolf both craplaod damage on each other, one kills the other sooner or later. One will get the Solokill and lots of scores for many inefficient damage done, AND lots of scores for "tanking damage" by the other being too stupid to properly aim.
Bth will get good scores for bad gameplay.

Probob meets ProkarlThey mostlikely end up with both doing only medium scores, because they will focusfire on critical locations and not spread much damage, even with twisting, because they won't shoot at nonsense locations as the randomaimguys. And so they will also tank less damage.

ScoreabusingSam vs Randomaimdudes and produde:He will try to strip as much as he can while staying hidden and finish opps when possible to grab solokills. He will strat "tanking" at the end to increase score just for the sake of increasing score. He will msotlikely get the most scores and do the leats to help the team winning.

Both, taking and delivering damage are bad skillindicators, because their judgement is numberbased, yet the circumstances of those numbers can be the same while the skill leading to those circumstances cna be entiely different. Whch in the end will "judge" skill incorrectly.

Way more interetsing would be an entire differently formula to judge kills their effort and efficiency.

Damage done on a mech should be worth 1/8th in Score. Every HP a mech has left when being destructed should be worth 1/4th in score. Thsi way an effiently killed emch by coring and destroying the CT or ST would grant more score than Stripping and finishing it. Thsi rewards tactics, such as sneakign up, coring form behind, or targetet destruytion of both legs or CT way better than what we have now. Also, it takes counterskill into account. When a pilot os good at spreading damage, he can by this lower the score granted to an opponent. compared to what an fats and efficient coring kill would be worth.
Further, the Scores granted by "HP left" should be distributed to everyone involved in the kill. and distributed by the %damage they were involved to kill the mech. 10% should go to the killing blow delivering mech ontop. (this is to prevent someone Stripping" a cored mech to get more part of the booty, instead taking the last 1HP and only getting 0.5% score from the kill).

Edited by Lily from animove, 27 August 2015 - 03:35 AM.

[MechWarrior849305]

Forget about PSR or ELO ffs, just play. Bads with LURMz with high ELO/PSR/Whatever-other-rank will be the saem bads with LURMz

[IraqiWalker]

Seelenlos, on 21 August 2015 - 08:10 AM, said:

Hi,

PSR is good, but:

1. every time such features get active there must be a reset.
With ELO every one was in the middle and we got the suck of ELO in worst manner mthematicaly
Now we had the chance to slowly populate the tiers but with this approach you got us again in ELO ranks (methodicaly and mathematicaly wrong - it is a new pool)

With PSR, we did not use a reset, because there was a lot of data present. PSR uses your stats recorded since January of 2015. SO it gives a decent approximation to start off with, and allows all buckets to be populated, instead of starting everyone in tier 4.

Seelenlos, on 21 August 2015 - 08:10 AM, said:

2. the score are too high
before that PSR the highest scores where 157 or so, now 500?
This will not help to seperate the poor from the rich (yes this is it at the end)
For 500 steps the tiers are to few. Then we had to have 10 to 12 tiers!

That is an assumption. We simply have no idea how big a tier is. For example, does a player need a PSR score of 5000 to get from tier 4 to tier 3?

We don't know that. We also don't know how the match score is used by PSR, it could use a formula such as MS/100 . So a score of 500 gives you 5 PSR points.

We don't know the inner workings of the system, PGI stated they won't reveal that, so no one abuses it, or games the system.

Seelenlos, on 21 August 2015 - 08:10 AM, said:

3. it is not clear which tier fighting which tier.
more tiers than 5 for more separation from 3 level deference or only 2 level difference fights

That is very true. what we do know, is that you only tier 3 players are exposed to the full spectrum. Since it's a 3 tier variance in either direction. However, I think that if we want to make it only a 2 tier variance, we would need to deal with much slower matches. Since the MM has to sort based on tier, server, and then game modes selected.

If we had a bigger population, I think a 10 tier system would be more suitable.

Seelenlos, on 21 August 2015 - 08:10 AM, said:

4. statistics:
how fast are they implemented, RT, every 3 matches, Day statistic, weekly ????

I believe this one is within 15 minutes of the match ending. It's the same with challenges. You'll notice that it takes up to 15 minutes for the system to record your progress in a challenge. So it's safe to assume your stats update within such an interval.

Seelenlos, on 21 August 2015 - 08:10 AM, said:

5. what a re exceptional scores in a losing team?
I must know that for at least make more scores from the rest of the idiots to ake myself a little better than the losers in that team.
You must exactly show the numbers which are sucking and numbers which separates me from the losers to reach them in a match !

I think we must all optimize on that, then it will get a nice curve and learning and challenge for all !

Considering the scale of the match score right now, I'd estimate (this is my own guess) you'd need a match score in the 350+ maybe even 400+ range for it to qualify.

The thing is, that score gives you a very tiny increase. I would assume it's almost negligible. We know that it's much less than what you would get for doing well in a win.

---

PGI have stated that they won't reveal the inner workings of the system, or the formulas, specifically to avoid cases where people would just gear up to try and break the system. This means a lot of what we have to work on is conjecture, and extrapolation.

[Lily from animove]

IraqiWalker, on 27 August 2015 - 05:01 AM, said:

This means a lot of what we have to work on is conjecture, and extrapolation.

Or field testing science.

[Willard Phule]

Look, I only know what I see and what happens in the games I play in....which tend to be solo PUG matches.

When Elo/MM was used, you had to carry at least 10 times your weight in new players....because PGI decided that instead of coming up with a meaningful tutorial, everyone else would be expected to teach them how to play.

Now that we've got this PSR thing and everyone is divided into tiers....you have to carry at least 10 times your weight, if not more.

And, to top it all off.....we just had a wave of ECM mechs released into the wild, followed by yet another PGI balancing act with the ECM (meaning it's working as intended...totally screwed up)....and then more LRM mechs added to the trial list.

Yep. All well thought out by the people that have been bringing you busted patches and ridiculous nonsense since closed beta.