Jump to content

Psr Update And Hold On Patch.


713 replies to this topic

#161 D U N E

    Member

  • PipPipPipPipPip
  • Bad Company
  • Bad Company
  • 131 posts

Posted 09 June 2020 - 06:08 AM

View PostZerex, on 09 June 2020 - 04:08 AM, said:


That would be a great point and far more believable if they weren't pushing for systems weighted so heavily to W/L in the PSR calc's, the systems i have seen from 2 people who i'm pretty sure consist of these "TOP MEN" are far far more open to exploitation to gain PSR than the present system. Imagine having rubbish players playing as team mates to you because they were carried there off other players skills, i have proved time and time again to these TOP MEN with actual match screen shots proving how flawed there system is, they refuse to even consider removing the major flaw in their system.


If you are exploiting the W/L ratio, that means you are winning. Winning is the objective of the game, if you are winning, it means that you out skilled your opponent, winning is one of the most important characteristics of... Winning...
If you win 10 games and lose 1, that means you should be in a higher tier as you are destroying your opposition.

As for a mech being carried, from what I see of their current PSR system, if you do great on a losing team, you can still go up compared to doing bad on a winning team. So don't worry in that regard.
But if you say you are gimping the system by winning, that is fundamentally the point of the game - to win, if you lose 9/10 times, it means you should go down, if you win that many times, it means you should go up.

That said, I have no idea what other systems you have seen, from what I know, they haven't defined what each attribute (Winning, Losing, Damage, Etc), they have two rough ideas of what MS distribution among players would look like with PSR gain/loss. So instead of being judgmental, when you see the idea actually look at what each one is - then let us know if you think people are being biased, or that all comp players are evil and chase them away with pitchforks. In the mean time, please leave any bias against high skill players.

In the meanwhile, you can see the parameters PGI has given players to control and edit. So what ever opinion you form, also take that into account. Players aren't being given free reign to create a revolutionary system that is perfect. Only to try and create a improved design over the current system that balances teams to make closer matches.

#162 cougurt

    Member

  • PipPipPipPipPipPipPip
  • CS 2023 Silver Champ
  • CS 2023 Silver Champ
  • 691 posts

Posted 09 June 2020 - 06:26 AM

View PostStar Captain Coruja, on 09 June 2020 - 06:03 AM, said:

Thanks for this level of openness.

Please please please weight objective capture more heavily, so that more people actually play the mission instead of "Skirmish with a secondary objective".

Matchscore stays low even if the actions of a few light 'Mechs unequivocally carry the match.

that won't change anything. the reason most matches play out like skirmish with a secondary objective is because, well, that's what they are. killing the enemy team should almost always be the top priority if you want to actually win.

#163 yrrot

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 222 posts

Posted 09 June 2020 - 06:28 AM

Way too many folks here want to place a big emphasis on Win/Loss in the match score calculation. This is problematic for a few reasons.

For one, it's redundant since the winning team is generally going to have a higher match score average anyway. More kills, more assists, more objective points, etc. Would have to look at a large sample of metrics from games, but I'd wager that a winning team with lower match score (even removing this bonus) would be rare. Most likely base rushes, for example, where the winning team didn't do much other than get to an objective faster.

For two, you're rewarding a player that got carried. Do 0 damage on the winning team, but somehow that's better than someone that lost and did some damage? That just doesn't make sense.

Winning the match isn't indicative of skill because it doesn't take into account the expected outcome. If you were outclassing the enemy team, you *should* win, so why would that make you go up in PSR? It's already the predictable outcome. If you're matched against a team of top pilots several games in a row, you're expected to lose, so why would that indicate anything about your skill?

Using W/L as a metric is entirely too dependent on a fair, normal distribution of player skill across matches. I would say that is a bold assumption given both the player count and what we know about the match maker.

#164 Poor-Life-Choices

    Member

  • PipPip
  • The Cyber Warrior
  • The Cyber Warrior
  • 27 posts

Posted 09 June 2020 - 06:29 AM

Restating the goal of PSR and MM is to estimate player skill level so that they can be evenly distributed accross two teams for a fair match.
In order to estimate player skill accurately, the measurements need to be done with an equal benchmark. Comparing players on a winning team with players on a losing team is not an equal bench mark. We all have lower scores when our team loses. So that brings me to two proposals

1) Top 6 on winning team go up, bottom 6 go down., Top 6 on losing team go up, bottom 6 go down, Scoring points shown below

2) Losing team = no PSR change. Winning team, Top 6 go up, bottom 6 go down

Scores
Top 2 players +3 points, and additional bonus point if MS over 500
Next 2 players +2 points and additional bonus point if MS over 500
Next 2 players +1 point and additional bonus point if MS over 500
Next 2 players -1 point and additional bonus point if MS over 500
next 2 players -2 points and additional bonus point if MS over 500
Next 2 players -3 points and additional bonus point if MS over 500

Then take these points and maintain a rolling average, top 5% are T1, next 15% are T2, next 60% are T3, next 15% are T4, next 5% are T5.
New players got to T4, after 20 games they can move based on their average PSR score. and after each 20 games they can move based on their average. Once they get to 200 games they just go where the average takes them. Once you get to 500 scored games then you follow the rolling average of the last 500 games.

#165 WildeKarde

    Member

  • PipPipPipPipPipPip
  • The Corsair
  • The Corsair
  • 487 posts

Posted 09 June 2020 - 06:29 AM

I've not read every post but have a few thoughts on this:
  • Makes wins vs loss less important. Currently on a win even as bad score does well for going up a tier. If the match score is the important factor then multiply the base match score based on result. Say match score x 100% for a win and match score x 90% for a loss as an example. This means good games on a losing team can still be rated over bad game on a winning team. Don't keep them too far apart in multiplier but not making losing irrelevant to the score
  • Damage taken should count to match score. Tanking the damage is stopping it being done to your team. Maybe make each point worth half damage done score.
  • You want to reward activities that support team. Sitting at 1000m sniping er larges for damage is not helping as much as being with your team in the fight to support them. Boost scores for proximity of your team and lance. Also saviour kills and similar being better scored.
LRM's may be seen as "non skill" damage farming but I don't think they have any less skill than a huge alpha build that can one shot a mech. Both reduce risk by simply racking up damage with less risk to the player. Keeping damage as the main scoring factor in matches will only encourage missiles and high alpha builds to be choice for a lot of players.

#166 ThunderKats

    Member

  • PipPip
  • Liquid Metal
  • Liquid Metal
  • 20 posts

Posted 09 June 2020 - 06:36 AM

Posted Image Look exactly like the old system but they just added -100 points to the formula...
Good thing I never care about tier. Only place where I found people willing to have fun (because this is a game) is on the lower tiers (t2-3), where people same as me don't care about tier, they actually use their armor to defend others at the cost of their final score/damage/reward and just try to have fun without melting 2 fast. I keep saying we need a secondary tier system, let people get their cbills rewards by boosting cbills on a win and not by meat-shielding with others to boost their own, let tier be a measure of experience and have a secondary that actually measure their PSR, been this what ever formula you see fit base on average score and I do want to see a dynamic one, using their last 100-300 average matches. This way you get people with experience with their tier 1-2-3 thing but you still can sort them out base on their average scores.
+Add a secondary PSR system to help sorting players base on both experience and piloting skills...
+Make win a rewarding thing by...
reducing reward base on dmg to 75% and add 50K+ for losing and 150K for win similar to conquest or what ever you see fit.
Posted Image No comment about lack of mobility/long-range/indirect-fire - Out of topic, but I do agree on some comments Posted Image

#167 Saint OZ

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,701 posts

Posted 09 June 2020 - 07:09 AM

Guys, sorry for my English.
I think that PGI use matchscore system to estimate player perfomance. There is no strong reason to count win/loss.
Just think: Win/Loss depends on Matchmaking, PSR depends on Win/Loss, Matchmaking depends on PSR.
I Ithink it would better if good matchscore boost PSR, bad - down it (whithout win/loss counting).

#168 Sjorpha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,478 posts
  • LocationSweden

Posted 09 June 2020 - 07:11 AM

There needs to be some emphasis on winning and losing, because it's winning more than losing over time that really captures the subtler details of what makes a player good.

Otherwise I think the suggestion is interesting, it's definitely a good thing that it's zero sum within each match.

You could address the win emphasis by increasing the MS bonus for winning, or you could have a fixed asymmetry between the gains and losses in psr for the winning and losing team.

Andrejz 3rd solution looks pretty good.

#169 Zerex

    Member

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

Posted 09 June 2020 - 07:11 AM

View PostKamikaze Viking, on 09 June 2020 - 04:50 AM, said:


Ok i just went to break it down and i got halfway through and realised this image is cut and pasted and sorted by match score. Hence i dont know who was on which team and cannot apply my system.
can you give me the raw scoreboard pic (names redacted) and then i can address is.

Then again its arguing about an old idea, which is kinda a waste of time, But i wanted to do it to clarify for you.

Edit: here you go. I applied the treatment of that old version to a random match from my screenshot folder.Posted Image


I never said your system didn't work, i just pointed out it has flaws in it, ones that can mean, to a larger degree it can be exploited by a group of (up to 6) players to increase their own chances of gaining PSR and to a lesser degree of players losing PSR while performing better and contributing more to their team than a player who gained PSR for performing less on winning team.

Do i think your system would work: Yes, if each player had a 1000 matches under their belt to iron out the impact of some the flaws in the system

Do i think there are better systems to use? Yes

Do i think this System can be exploited? yes

Out of all systems i have seen Dogmeats actually has the best chance of being a perfect way to grade player rating, think it has flaws that need ironing out with it, but and here is the problem, its a huge huge rework of the system that is in place, there is no way of testing his system until it is up and running (me and them have used example numbers we both agreed on it the tests me and them did, so you can't know how well it works until the system is in place, also the way the they set up the baseline target match score there is a chance that because of elements outside of your control (overall match tonnage being the obvious) you might not be able to reach your target match score before all the enemy are dead. Also it can lock you into a mech or playstyle, on my alt account i play a MPL Crab 27B because i find it the mech i perform best in, on Jarls list i have a average Match score of 412 (me and dogmeat both used Jarls average match score as the example of our PSR), so i now have to reach 412 match score in every match just to maintain that PSR, If MPL's get nerf'd 3 months later i start to lose match score per game and PSR, my skill level hasn't changed, i could still be top player in every match but now i'm losing PSR every game. Alsoi might locked into playing only that 1 mech from now on because if i try new builds out i won't do as well and could drop PSR (This one only really effects ego's and try hards).

As i said on Reddit:

Quote

1, the more complex you make a system the easier it is to be flawed or exploited, while my system isn't perfect at rating you as a player to a 0.001 decimal point, it doesn't need to, it just needs to work, send players that performed well up, players that performed baldy down and players that fall in the middle have a zero gain/loss, if it does that the tier 1 matches will be filled with players that consistently performed well in the matches they played in and players in the bottom tier are there because they perform poorly in the matches they played in.


#170 Kosomok

    Member

  • PipPipPipPipPip
  • Shredder
  • Shredder
  • 187 posts
  • LocationNevada

Posted 09 June 2020 - 07:12 AM

Disclaimer: I played through pretty much the entire beta, was a major contributor/discusser on the forums at that time. Quit shortly after the game "launched" because of real life things (like getting married and not having time and resources to play). Came back recently because I was stuck at home due to COVID and had been playing some MW5.


One of the major issues of the system through its history is that it was made visible. You knew what tier you were in. Psychologically, it became the equivalent of level in an MMO. People expected and wanted it to go up it was the metric of success.

It should have been invisible. It should become invisible, something only the MM sees.

W/L (as a metric) should have essentially zero effect on MS. Winning will be reflected in MS automatically for those players who contributed to the win (in that they did things that were effective and that contributed). No need to add points to MS for winning.

Numerical measurement of skill is quite difficult--and what is measured is not always necessarily the right things--merely the things that the designers THINK are the right things. I think some things in the game are seriously under valued and others are seriously over-valued.

All of this has a knock on effect on actual game play--which results in behaviors that ultimately reduce the fun level (unless you only measurement of fun is how high your MS is and whether you won or lost).

Given the current size of the player base most of this is kinda moot as you are going to continuously end up with low skill versus high skill players in order to make matches. Unless you can grow the player base, this will not change. It will be hard to grow the player base in such a situation because of the stomps and the overall negative environment.

Putting groups and solos into the same MM queue is a serious problem. This was recognized back in beta (I was there). But here we are. It is not a solution to the problem of a lower count player base and reducing MM time--it is part of the problem. The solution is to kill grouping in other than competitive play. It always was the solution. Match size should also be reduced from 12 to 8 per side.

None of what I said bears directly on the issue at hand, but it flows from the issue at hand and it contributes to the slow death of the game. It amounts to re-arranging the deck chairs on the Titanic.

Because, for a lot of players, the game isn't fun. They stop playing.

#171 Nearly Dead

    Member

  • PipPipPipPipPipPip
  • Philanthropist
  • 274 posts

Posted 09 June 2020 - 07:13 AM

I like the proposal that the middle scoring players in a match don't have to move up or down every match. As a player who usually scores 5th, 6th or 7th in TS it would not bother me to stay neutral in some matches. If you are fighting above or below your actual skill level you will still move up or down till you are with your peers but it might take a little longer to settle out once you get close.

I have one question about the variable table. Are the variables for mech type and other factors that affect performance on the table for use? I see variables that use mech class (protected light etc.) but I don't see those variables in the list. It might be helpful to weigh some things like scouting bonuses, TAG, UAV, and NARC damage etc differently for one class than another.

#172 OpticSeagull

    Rookie

  • Liquid Metal
  • Liquid Metal
  • 4 posts

Posted 09 June 2020 - 07:25 AM

As long as participation related achievements overshadow / outstrip, the cost of dying, you will have a game.

If dying is a big loss, then team play will deteriorate and there will be no gain from standing with teammates and dying for a good cause, or for the win.
Make sure the values of team work are high,
Then you can implement all those changes, and the gains from captures and retrieving batteries and stuff.

Also, people who preserve K/D ratio cost the players time and frustration. It's a tremendous cost when you count up all the man hours of non engaging, non participating players. The cycle time between games, from when they are decided to starting the next one, has to be high and replayable. I cannot, however, offer a suggestion as to how to penalize non participants, b/c some are incidental and unfairly disconnected. Perhaps it's another reason to enhance the rewards of participation.

#173 Zerex

    Member

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

Posted 09 June 2020 - 07:25 AM

View PostD U N E, on 09 June 2020 - 06:08 AM, said:


So instead of being judgmental, when you see the idea actually look at what each one is - then let us know if you think people are being biased, or that all comp players are evil and chase them away with pitchforks. In the mean time, please leave any bias against high skill players.



Please name me these high skill comp players and a list of reasons why they are high skill comp players that you say i'm being bias against, I'm very interested to see what they have won

#174 jonomy

    Member

  • PipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 79 posts

Posted 09 June 2020 - 07:27 AM

View PostMordenthral, on 08 June 2020 - 05:32 PM, said:

You're getting entirely off-topic. Your dislike of people that cap (bottom-feeders?) isn't relevant.

My contribution to the suggestions:
Skirmish - Killing the enemy is the mode and should have the most weight
Conquest - Gathering resources is the mode and should have the most weight
Assault - Capturing the enemy base is the mode and should have the most weight
Domination - Holding the circle is the mode and should have the most weight
Incursion - Destroying the enemy base is the mode and should have the most weight

Like if you agree, as mentioned.


This seems in line with the game description, directions, etc. In addition, if matches are won based on stated match goals, players might well be ranked by relative participation in goal attainment (teamwork?). Did you contribute to accomplishing the match goals? 6 kills as a sniper in domination might be hard to rank if the team lost the match unless system sees actual help for teammates who are tighter in the fray.

#175 Warning incoming Humble Dexterer

    Member

  • PipPipPipPipPipPipPipPip
  • Bridesmaid
  • 1,077 posts

Posted 09 June 2020 - 07:28 AM

A as side note I'll say there is absolutely ZERO difference between :
- Going from +24 down to -24 in steps of 2.
- Going from +12 down to -12 in steps of 1.
- Going from +36 billion down to -36 billion in steps of 3 billion.

So just drop the steps from 2 to 1 : I have no idea why you are using steps of 2 for absolutely no effect, reason or impact.

#176 Hastur Azargo

    Member

  • PipPipPipPipPipPip
  • The Angel
  • The Angel
  • 226 posts
  • LocationGloriana class battleship "Red Tear"

Posted 09 June 2020 - 07:29 AM

I just wanted to ask if all of this activity means MWO will get more content and development? I mean when PGI straight out abandoned this game to make a singleplayer game (and an EGS-exclusive one, no less), it was kind of obvious that it's not a sustainable effort. So what's the deal here? Is PGI trying to reanimate this game that they pretty much left to die? Or are there other long-term plans?

#177 ERescue

    Member

  • PipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 203 posts

Posted 09 June 2020 - 07:42 AM

@All: For now I will note this. This is NOT an easy task by any means. I was just in 4-12 loss game... and the best overall player in the entire match was in the losing team. 2nd highest damage (by THREE poins vs an assault missile boat that failed to get any kills), 3rd highest MS (only losing to said missile boat and a pure energy Marauder that did significanly less damage, but managed hit to everyone but me in the losing team) and while my result was nothing to brag about: 2 assists, 85 damage and MS of 116, I still outperformed a Marauder IIc in the enemy team that got 10 assists, but only did a total of 75 damage and ended with a MS of 112. I effectively traded my TDK vs an enemy Nova-S, who wanted too badly to kill me. I went down, but so did he a few seconds later, because he had turned his already damaged back to the rest of my team.

@Yrrot: Team average is a very tricky thing... I have been in several games lately, where the losing team had an average very, very close to the winning one... and in at least one such game, EIGHT of the highest MSs were in the losing team. Game ended at 2 - 3 on base cap, because unfortunately as the only light in our team I died fairly early and others were to slow to prevent enemy from capping after I fell. The enemy light was more interested in winning than points. I think he actually had the lowest score of all 24, but he won the match.

@PGI: I am well aware that you will not put huge amount of work (=money) into this, but PLEASE try to find a solution that covers a wide range of end results... and as I noted in another post, even though ties are very rare, make sure to address that as well... and actually that request goes to all other players as well... I know you all want a functional system, so please present your systems in light of ties as well. Thank you to ALL!

#178 kapusta11

    Member

  • PipPipPipPipPipPipPipPipPip
  • Little Helper
  • Little Helper
  • 3,854 posts

Posted 09 June 2020 - 07:45 AM

View PostKamikaze Viking, on 08 June 2020 - 04:59 PM, said:

Thankyou Russ and Paul for the open communication on this topic.

I think I've hit my limit on how I can affect this discussion with my most recent proposal.
https://mwomercs.com...ost__p__6336540
Posted Image

In my opinion from what Ive seen in recent discussion here and on reddit, Dogmeat and Decency have the best grasp on Matchmaking in general AND the context of how it should work in MWO.

I have also seen very binary arguments about the importance of personal skill vs win loss. In my opinion they are BOTH a factor that must be considered.

And as Russ has said, all of us here are guessing at existing systems. I would really really like it if the Engineer involved would be allowed to come and discuss this with us directly.

To anyone proposing solutions. remember the K.I.S.S. Principle, and consider the work required to implement your solution. Perfect systems are great, but we WILL have to make compromises, its the reality of things.

And I would be fine with addressing PSR now and then letting it settle for a month before addressing Match Score calculations. We have seen in the past that too many simultaneous changes make it hard to understand the effects and what change caused what effect.


Why are so many people upvoting this? In two different games, one filled with competitive players on both sides and one with potatoes on both sides, at the end this system will equalize players by their respective in game ranking. The best competitive player will be as good as the best potato and the worst competitive will be as bad as the worst potato according to their PSR. That's terrible.

It's an exaggerated example but barring top 1% who will always be on top, in practice the rest of the player base will constantly hop in their ranking depending on how many good/bad players queued up at the moment, whether it's prime time or not, whether there's an even or not etc.

Edited by kapusta11, 09 June 2020 - 07:50 AM.


#179 Ridir Semii

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 499 posts
  • LocationPort Torture, Washington, USA

Posted 09 June 2020 - 07:47 AM

I have read through all of this and I honestly believe we are all trying to make a mountain out of a molehill. Let's take a look at all of this in a simple manner:

Point 1: PSR is how MM sorts us into matches.
Point 2: people want to get increases for their performance (when MS would or SHOULD warrant it)
Point 3: people want to see decreases for %$&$%@ off performances (when MS would or SHOULD warrant it)

Those 3 points make this very, very simple. Using a simple formula like the one Paul proposed, PSR is only affected by MS. Not by wins or losses. Since w/l rate is accounted for within the MS system, there is no need for redundancy with it. Your personal performance would become the deciding factor in whether or not your tier increases or decreases.

With all of the factors contained within the MS score calculations, a player who is contributing to their team should have no issues with increasing their PSR. (defined: contributing is marking targets, doing damage, hitting the objectives, killing opponents, protecting friendlies, etc....)

{{Please Note}} I am not saying the MS system doesn't need an overhaul, I am saying that we should tackle one obstacle at a time, start with how PSR is factored, see how the MM handles it... we could then fine tune the MS system then. Adjusting multiple factors at the same time will skew the results. We would not know which factor caused the changes properly

Win or lose, your performance will make or break your tier level. I routinely get in the upper half of MS during my matches, WITHOUT paying attention to the win or loss. According to the current system, I am tier 5, I get stuck in the matches with the "potatoes" and "baby seals" and this tends to cost me a ton of match wins... For starters, I had to take a break from this game for some years because the powers that be kept the system requirements higher than I had, and I was using a cellphone 3g hotspot to learn this game through beta and release... I averaged 15fps-ish during that time... I now have a decent machine and average fps across the servers is about 90.... my performance from match to match has shown this improvement, and showed me that I wasn't the absolute shite player I thought I was... Am I top tier... not on your life..... Nor do I pretend to be.

<<stepping off the soapbox now>>

#180 HolyGrail101

    Member

  • PipPip
  • Shredder
  • Shredder
  • 49 posts

Posted 09 June 2020 - 07:57 AM

I won't claim to get the programing side of the issue but the issue isn't just the lack of player base or Matchmaker for me; It is the off kilter amount of work it takes for Light Mech's to qualify for the same reward level. I generally do way more work in a Commando than most people in Assault Mech's. Paper-Machete armor equals easier death not a lesser pilot. Point being that everyone one is mad about something different, Matchmaker is part of the solution not the problem.

Less diverse Maps and Mode's will only drive away more players.
Drop tonnage may even out when everyone has no choice but to play Skirmish and only 3 maps are available.
Matches that only big Mech's can win.
Broken matchmaker.

I think with the current player base limit I would be most satisfied if the tonnage ratio was just less obnoxiously off.

I'm just brainstorming but:
- Force group tonnage to continually readjust based off of what is available in solo que so Solo que is less likely to be way under or over on a single team

- Give an advantage or two to the team without a group or if they are underweight
A. Better Multiplier
B. Better start locations
C. Better Map or Mode

- Kill off Solaris I suspect it mostly spreads a thinning player base thinner
A. Introduce new Match scoring that includes dueling during a PUG battle & Batchall's during Faction battle you already have the trophies.
B. Killing off Solaris should be fun. Make it a giant planned event
C: If you refuse to kill off Solaris then make it special and only do Solaris as an Event rather than another grind

Edited by HolyGrail101, 09 June 2020 - 09:29 AM.






10 user(s) are reading this topic

0 members, 10 guests, 0 anonymous users