Jump to content

Psr Update And Hold On Patch.


713 replies to this topic

#581 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 11:55 AM

View PostNightbird, on 14 June 2020 - 11:51 AM, said:


What's more common, players with hundreds of games and 10 WLR not being close in skill, or players with hundreds of games and the same avgMS and not being close in skill?

https://leaderboard....&d=DESC&page=10

One page with a 5 point difference in avgMS and WLR from 1.2 to 6. avgMS sucks.


Because both AMS and WLR are flawed systems that is not representative of a players skill level

#582 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 14 June 2020 - 11:58 AM

View PostZerex, on 14 June 2020 - 11:55 AM, said:


Because both AMS and WLR are flawed systems that is not representative of a players skill level


AMS has 11% accuracy, WLR has 37% accuracy, and we don't have anything else better. Also it's impossible to use both because they are collinear.

Edited by Nightbird, 14 June 2020 - 11:58 AM.


#583 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 11:59 AM

View PostGozuri, on 14 June 2020 - 11:53 AM, said:


Only the teams collective ability is considered. Hence it does not translate into mwo, where players are not drawn from preformed teams.



Ahh, then correct, your personal stats are being defined by the team effort and not by you.

#584 Nameless King

    Member

  • PipPipPipPipPipPipPip
  • The King
  • The King
  • 692 posts

Posted 14 June 2020 - 12:08 PM

If solo players and groups are playing together, then Win Loss needs to be removed from the discussion.

Edited by Nameless King, 14 June 2020 - 12:16 PM.


#585 D U N E

    Member

  • PipPipPipPipPip
  • Bad Company
  • Bad Company
  • 131 posts

Posted 14 June 2020 - 12:10 PM

View PostZerex, on 14 June 2020 - 10:52 AM, said:


More over its open to be abused by players win farming by on game modes like Assault, Conquest and Incursion. Which could lead to increased numbers of base race matches instead of mech fights, which i think you'll see players leave the game at that point due to it not b a Mechwarrior game but a racing game in mechs.



If you are able to win via Assault, Conquest or Incursions main game modes, then congratulations, you are winning. If the enemy doesn't kill your mechs that are farming, you do this a lot and people still don't stop you, perhaps it will push against nascar meta as *Surprise Pikachu Face* running in circles is a bad idea.

View PostZerex, on 14 June 2020 - 10:52 AM, said:

Also looking at stats good players that always perform well but on the losing team, i believe i counted 5 players in the top 40 medium pilots (arranged by AMS) this month that have a +400 AMS but less than 1 WLR.

Also looking at this months stats you have players that have high WLR (top 40 players arranged by win/loss) that have really low match scores, like AMS in 1 case was 165 with a 11 WLR.

With this taken into account using win/loss only leads to exploitable mechanics, good solo players kept in low tiers and bad players rising through the tiers t what i can only guess is their team mates carrying them to wins.


I believe you mean arranged by MS - AMS stands for "Anti Missile System", while MS stands for "Match Score"
There are currently imbalanced groups - and with GQ being merged means they are likely to face talented players in groups. But notice something with these players, look at the amount of times they have died, and the amount of times they have lost.
A good player will try to fight until their mech is dead, but when you are living in quite a few matches that you have lost, the chances are these are players in mechs such as the PPC Shadowcat - If you play it right, you can get high damage numbers (meaning high MS) - though it doesn't have much DPS, meaning you aren't dealing enough damage at the right time. These players are therefore likely not helping to kill the enemy as much despite having a high MS. Not to say these mechs are innately bad, but for the meta in solo que these players are not as useful as they could be. They also allowed the enemy to complete the objective, meaning instead of attacking the enemy while they were completing the objective, they ran away for whatever reason.

Edited by D U N E, 14 June 2020 - 12:11 PM.


#586 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 12:12 PM

View PostNightbird, on 14 June 2020 - 11:58 AM, said:


AMS has 11% accuracy, WLR has 37% accuracy, and we don't have anything else better. Also it's impossible to use both because they are collinear.


A system that rates player in the match they are in using that matches matchscore as guide will always give better results, if the top performing players in that given match do well they increase PSR, if they perform badly they go down PSR, it has nothing to do with MS other than ranking how well those players performed in that 1 match.

Edited by Zerex, 14 June 2020 - 12:18 PM.


#587 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 12:17 PM

View PostD U N E, on 14 June 2020 - 12:10 PM, said:


If you are able to win via Assault, Conquest or Incursions main game modes, then congratulations, you are winning. If the enemy doesn't kill your mechs that are farming, you do this a lot and people still don't stop you, perhaps it will push against nascar meta as *Surprise Pikachu Face* running in circles is a bad idea.



I believe you mean arranged by MS - AMS stands for "Anti Missile System", while MS stands for "Match Score"
There are currently imbalanced groups - and with GQ being merged means they are likely to face talented players in groups. But notice something with these players, look at the amount of times they have died, and the amount of times they have lost.
A good player will try to fight until their mech is dead, but when you are living in quite a few matches that you have lost, the chances are these are players in mechs such as the PPC Shadowcat - If you play it right, you can get high damage numbers (meaning high MS) - though it doesn't have much DPS, meaning you aren't dealing enough damage at the right time. These players are therefore likely not helping to kill the enemy as much despite having a high MS. Not to say these mechs are innately bad, but for the meta in solo que these players are not as useful as they could be. They also allowed the enemy to complete the objective, meaning instead of attacking the enemy while they were completing the objective, they ran away for whatever reason.


MS is matchscore, matchscore is only used for single matches, AMS is Average Matchscore and is used on the PGI leaderboards and Jarls list, I'm not the first person to you AMS as Average matchscore in this topic and i bet i'm not even the last, this might go a little way to explaining why your don't understand what other people are talking about when you don't even understand the references.

#588 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 14 June 2020 - 12:19 PM

View PostZerex, on 14 June 2020 - 12:12 PM, said:


A system that rates player in the match they are in using that matches matchscore as guide will always give better results, if the top players performing players in that given match do well they increase PSR, if they perform badly they go down PSR, it has nothing to do with MS other than ranking how well those players performed in that 1 match.


That is PGI/Jay Z's system, with avgMS and WLR being other systems.

PGI/Jay Z's system is 5/100 in terms of what a MM can do to make even matches. avgMS is 11/100. WLR is 37/100,

#589 Gozuri

    Member

  • PipPip
  • 25 posts

Posted 14 June 2020 - 12:26 PM

View Postspannerturner, on 14 June 2020 - 10:58 AM, said:


Here's why I brought it up. To translate into MWO terms, Mike Trout would have about a 1.04 W/L stat, which is about equivalent (to what most have said on these last few forums regarding PSR change) to what should be Tier 3. We know he's better than that. There are top Tier players in MWO, who had a less than 1.0 W/L during May's test, that would then be relegated to Tier 3 or lower, even though we know they are better than that. Then there are those that had a higher W/L during May's test that are not Tier 1 players (skill-wise), but whose rating would go up just because they had a >1.0 W/L... This is not skill based, it is luck of the draw based (at least for the solo players).

Skill based matchmaking ensures that you are matched against those with skill levels similar to you.
W/L based matchmaking ensures that you are matched against those with the same W/L% as you.



Ok, I'm gonna try to make this clear. Mike Trout can be a very good player, best in the world, but he isn't going to be put in a team of random players. Instead he is going to be playing for his team and his team only. So, when a tournament organizer matches teams against each other, they don't go, "hey, Mike is in this team, it should be seeded as the top team". They take the team's performance as a whole and seed them accordingly. So in the context of matchmaking, his personal score carries no weight whatsoever. Correct me if I'm wrong (I don't know much about professional sports).
Mike's personal score may be used in building teams to prevent one sided stacking, and that is fair. Because they don't pit individual players against each other, and said individual players cannot switch teams on a whim, a measure is needed to score a player and so they take the next best option, that is their arbitrary performance on the field.
In MWO, we don't form a full team of 12 people, then go find other teams to play against. Instead, we are dropped into a team of random people. Because you and your team have never played with each other as a team prior to this, there is no way to rate the whole team based on its performance, so we take the individual players skills based on their past performance. The discussion here is should we use MS in our consideration.
My argument is no, because MS does not predict a person's winning odds as accurately as their past history of wins and losses. Also, while more wins will give you an upward trend of rating gain (in a W/L system), it is not necessarily so.

Red Avg psr: 1000, 1010, 990
Blue Avg psr: 1000, 990, 1010
R Potential chg: +10/-10, +8/-12, +12/-8
B Potential chg: +10/-10, +12/-8, +8/-12

If you're consistently matched against stronger opponents, you lose less for each loss and earn more for each win. It is possible to have a negative WLR and still rate highly. I think this is the part missed from my original post. I never advocated using a ratio of wins vs losses, merely a removal of MS from the equation and thus relying purely on the outcome of the match, i.e. win, lose, or draw, taking your opponent's ratings into account.

So yes, if there is a Mike in MWO, he will end up carrying his team on most occasion, and be rated highly.

Edited by Gozuri, 14 June 2020 - 12:48 PM.


#590 D U N E

    Member

  • PipPipPipPipPip
  • Bad Company
  • Bad Company
  • 131 posts

Posted 14 June 2020 - 12:29 PM

View PostZerex, on 14 June 2020 - 12:17 PM, said:


MS is matchscore, matchscore is only used for single matches, AMS is Average Matchscore and is used on the PGI leaderboards and Jarls list, I'm not the first person to you AMS as Average matchscore in this topic and i bet i'm not even the last, this might go a little way to explaining why your don't understand what other people are talking about when you don't even understand the references.


Fair enough, That said, while you countered that - how about you look at the rest of the argument. (And not the mild correction- as I have seen AMS as in, Anti Missile System been discussed in depth in terms of Relative MS gained for standing their doing nothing.)
The top people you describe are likely not assisting with the win as much as you believe they are, and therefore should be in a lower skill bracket, as winning is not just abusing a mechs speed and jumping around out of range.

If we look at the highest people in W/L ratio, in the top 20, I have not checked the other pages, they die more than they have lost. Cause they put their entire mech forward in securing the win, as opposed to securing the highest K/D

Edited by D U N E, 14 June 2020 - 12:32 PM.


#591 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 12:35 PM

View PostNightbird, on 14 June 2020 - 12:19 PM, said:

That is PGI/Jay Z's system, with avgMS and WLR being other systems.

PGI/Jay Z's system is 5/100 in terms of what a MM can do to make even matches. avgMS is 11/100. WLR is 37/100,


Its closer to PGI

OK lets take 2 example players and work on the PSR gain/loss is 10 points max per match, these players are not in the same match.



Player 1 wins 10 out of 10 of their matches and places mid table in both team and over all match performances

player 2 losses 10 out of 10 of their matches but places as the top performer in all ten matches in team and overall

Your WLR system

Player 1 gains 100 PSR

Player 2 loses unknown value of PSR

Ranked system

Player 1 PSR remains unchanged

Player 2 gains 100 PSR

At no point does Average match score ever come in to it

Why would a player that is clearly in the right tier as they are hitting middle table almost every match and performing as PSR and MM predicted gain PSR in your system?

#592 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 12:43 PM

View PostD U N E, on 14 June 2020 - 12:29 PM, said:


If we look at the highest people in W/L ratio, in the top 20, I have not checked the other pages, they die more than they have lost. Cause they put their entire mech forward in securing the win, as opposed to securing the highest K/D



That is one way at looking at it, or they were playing outside of their skill zone died and the team went on to win the match, amazing how the same info can read as 2 different stories.

As i have said, you need to rank that players performance in that match against every other player in that match so understand how much they contributed to the win and until LURM damage farming and AMS (Anti Mssile System) farming is reduced its not as easy to tell as much as one would like.

WLR, KDR, and MS can be wildly affected by other players.

#593 Gozuri

    Member

  • PipPip
  • 25 posts

Posted 14 June 2020 - 12:57 PM

View PostZerex, on 14 June 2020 - 12:35 PM, said:


Its closer to PGI

OK lets take 2 example players and work on the PSR gain/loss is 10 points max per match, these players are not in the same match.



Player 1 wins 10 out of 10 of their matches and places mid table in both team and over all match performances

player 2 losses 10 out of 10 of their matches but places as the top performer in all ten matches in team and overall

Your WLR system

Player 1 gains 100 PSR

Player 2 loses unknown value of PSR

Ranked system

Player 1 PSR remains unchanged

Player 2 gains 100 PSR

At no point does Average match score ever come in to it

Why would a player that is clearly in the right tier as they are hitting middle table almost every match and performing as PSR and MM predicted gain PSR in your system?


Then either player 2 is extremely unlucky, consistently matched against stronger teams, or he isn't actually doing things to win the match but just scoring high.
Either way, his rating change would be justified.

View PostGozuri, on 14 June 2020 - 09:15 AM, said:


If you need a system, just get the average ratings of both teams and determine potential gain/losses in ratings based on the difference.
E.g.

Red Avg psr: 1000 1010 990
Blue Avg psr: 1000 990 1010
R Potential chg: +10/-10 +8/-12 +12/-8
B Potential chg: +10/-10 +12/-8 +8/-12


Edited by Gozuri, 14 June 2020 - 01:02 PM.


#594 D U N E

    Member

  • PipPipPipPipPip
  • Bad Company
  • Bad Company
  • 131 posts

Posted 14 June 2020 - 01:01 PM

View PostZerex, on 14 June 2020 - 12:43 PM, said:


That is one way at looking at it, or they were playing outside of their skill zone died and the team went on to win the match, amazing how the same info can read as 2 different stories.

As i have said, you need to rank that players performance in that match against every other player in that match so understand how much they contributed to the win and until LURM damage farming and AMS (Anti Mssile System) farming is reduced its not as easy to tell as much as one would like.

WLR, KDR, and MS can be wildly affected by other players.


I like how you keep ignoring key points.

Let's do this again, but slower.

The main players you noted - with a negative W/L ratio have died less than they have lost.
The players I noted, with a significantly positive W/L ratio, have died more than they have lost.

To really show this point, in the first 40 mediums. By a quick look, the 37/40 people died more than they losts.
3/40 people therefore lived during a lost/draw. 2/40 people have a negative W/L ratio - who also happens to be two of the 3 people that lived during losses.

Now what does this represent? I think you can answer that yourself.

Now lets look at people with the highest W/L ratio (Which are very likely to be boosted by groups) - Once again, looking at mediums rated by W/L ratio. 0/40 people sortied by W/L ratio in medium mechs have died less than they have lost. 40/40 have died more than they have lost. These people are therefore, dying. They can die on a win, they can die on a loss, sometimes they might livie during a loss. Though the norm for people that are not K/D farming, is that in a loss they also die - as QP is Nascar. Currently Nascar is "Every Game-mode is Skirmish and racing"

So with all of this information overtly described to you - it is safe to assume, my initial argument, regardless of "Well you can die doing 0 damage and still win" - (which is an outlier that if a person persisted in participating in, would also end up having a negative W/L ratio.) likely stands firm.

Edited by D U N E, 14 June 2020 - 01:03 PM.


#595 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 01:08 PM

View PostGozuri, on 14 June 2020 - 12:57 PM, said:


Then either player 2 is extremely unlucky, or he isn't actually doing things to win the match but just scoring high.


Just an example, without breaking out screenshots and charts its the simplest way explain it in quick simple motions.

Take those 10 matches and scatter them in with 90 other matches and you now have those as 10% of the players matches, at what point does this sort of this break Nightbirds system, 1%, 5%, 10%, 30%?

I know that i have more than 1% of my matches that end with me being the top performer in a match but on the losing side.

#596 David Sumner

    Member

  • PipPipPipPipPipPip
  • FP Veteran - Beta 1
  • 470 posts
  • LocationAuckland, New Zealand

Posted 14 June 2020 - 01:21 PM

View PostGozuri, on 14 June 2020 - 10:08 AM, said:


If you're paired with a squad of weaker players, that also means the opponent will have a bunch of deadweight. Carry the game like you're meant to and carry it hard.


You're forgetting.

On average, people are *******.

#597 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 01:24 PM

View PostD U N E, on 14 June 2020 - 01:01 PM, said:


Let's do this again, but slower.



If i am understanding you right you have my point totally wrong, i am saying you can't use Average MS, WLR or KDR as you have just pointed out they give false positives and negatives

Top performance on a loss

Bottom performance on a win

Dying in a win

Surviving a loss

Scoring a low MS in a low tonnage game

Scoring a high MS in a high tonnage

All these things are false data points, a robust system has to either over come them through a formula or it has to score players where these aren't false data sets

#598 David Sumner

    Member

  • PipPipPipPipPipPip
  • FP Veteran - Beta 1
  • 470 posts
  • LocationAuckland, New Zealand

Posted 14 June 2020 - 01:30 PM

View PostOZHomerOZ, on 13 June 2020 - 08:10 PM, said:


...

If you make winning a large part of the formula it will favour grouped pilots and disadvantage solo pilots in terms of PSR movement. Is this really measuring a pilots skill is what I am saying.
Edit: Even Handsome Proton mentioned that maintaining a positive win loss as a solo was hard work or some thing close to it. Correct me if I'm wrong beautiful? Posted Image

Groups don't need more advantages.

...


I would think that "advantaging groups" would result in group players being moved up into a grouping distinct from solo players.
(Which is the divergence everyone is talking about).
And the result of that would likely be that group players would spend more time facing group players and less time facing solo players.

And without a separate group queue, I'm fine with that.

Edited by David Sumner, 14 June 2020 - 01:30 PM.


#599 Horseman

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Mercenary
  • The Mercenary
  • 4,737 posts
  • LocationPoland

Posted 14 June 2020 - 01:47 PM

View PostOZHomerOZ, on 14 June 2020 - 06:45 AM, said:

I thought the MS part is OK, after all, Jarls list is acceptably accurate (not perfect) and is derived from the existing data. And didn't Russ or Paul say to the effect: we just talking about a PSR rejig and not MS rejig but they gave us some data about MS triggers to help make the PSR better. In other words MS twerking is out of scope!
(Otherwise they would of given us the formula not just the triggers and in fact stated they won't give the formula due to concerns about gaming/exploiting/abusing it etc)
Please correct me if I am wrong!

You are wrong.
https://mwomercs.com...ost__p__6337261

Quote

If the community can agree on weighting of these events, we will review them on our side to make sure they're fair to as many play styles as possible. Once that is done, we'll implement the numbers and try them out.


#600 Zerex

    Member

  • PipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 298 posts
  • LocationUK

Posted 14 June 2020 - 01:50 PM

View PostD U N E, on 14 June 2020 - 01:01 PM, said:


Let's do this again, but slower.




Ok to use the example you have given, season 47, mediums arranged by W/L

Posted Image

These 2 players W/R are very close and both very high, Nightbirds system will rate these 2 players at the same PSR

Reading between the lines at the stats, Gurney Hallack is playing far outside skill zone, my guess is they move towards the enemy by them selves or just charge enemy line every time they see them. or it could be an assault being left behind by their team, or they have very little heat management and blow themselves up when the fighting starts, its hard to tel how they play.

1 thing is clear Gurney Hallack doesn't have the same impact on the teams outcome as Ghogiel does, but Nightbirds system will still rate them at the same PSR and treat them as equal skilled players





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users