Jump to content

Psr Tier Metrics - Round 1


642 replies to this topic

#521 Ekson Valdez

    Volunteer Moderator

  • PipPipPipPipPipPipPip
  • 716 posts

Posted 07 August 2020 - 01:50 AM

View PostRRAMIREZ, on 07 August 2020 - 01:42 AM, said:

What ever is in loot caches, it shouldn't be a driver on MS/PSR design




Exactly. Stay on topic, people!



#522 General Solo

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • 3,625 posts

Posted 07 August 2020 - 02:36 AM

Any one know if new Tier numbers has been released, as I'm wondering what the current numbers are for Tier 5 as that was the biggest Tier population prior to the merge and subsikwent PSR reset.

Or did Tier 5 get rekt like some pristine Alaskan coastline due to players leaving before they dropped to Tier 5

I wonder Posted Image

#523 RRAMIREZ

    Member

  • PipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 183 posts
  • LocationIn the Blob

Posted 07 August 2020 - 03:55 AM

View PostOZHomerOZ, on 07 August 2020 - 02:36 AM, said:

Any one know if new Tier numbers has been released, as I'm wondering what the current numbers are for Tier 5 as that was the biggest Tier population prior to the merge and subsikwent PSR reset.

As far as I know, no new data since 15 juily in OP.
Edit:
Wondering also how is this evoving, and if Tiers (out of 3) are growing enough to "tighten" valves rules

Edited by RRAMIREZ, 07 August 2020 - 03:57 AM.


#524 Tarl Cabot

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Tai-sho
  • Tai-sho
  • 7,780 posts
  • LocationImperial City, Luthien - Draconis Combine

Posted 07 August 2020 - 05:20 AM

View PostOZHomerOZ, on 07 August 2020 - 02:36 AM, said:

Any one know if new Tier numbers has been released, as I'm wondering what the current numbers are for Tier 5 as that was the biggest Tier population prior to the merge and subsikwent PSR reset.

Or did Tier 5 get rekt like some pristine Alaskan coastline due to players leaving before they dropped to Tier 5

I wonder Posted Image


Remember, that huge Tier 5 column graph was of players, new accounts did start in Tier 5, who had logged in within the previous 3 months, and most likely played at least 1 game.....though Paul appeared to be shocked about it in his post, even then he failed to really expand on it which is PGI usual M.O. So majority of those "new" previous Tier 5 players are still in Tier 3, they just haven't actually played...

And some wonder why PGI has difficulty with game components.

Edited by Tarl Cabot, 09 August 2020 - 01:08 PM.


#525 morosis

    Member

  • PipPipPip
  • The Sickle
  • 79 posts

Posted 08 August 2020 - 06:47 PM

Update:

Today, as a Tier 2 pilot, i played 10 games.

10 out of 10 were Tier 1 games.

The matchmaker is still not working.

Fix it.

#526 OneTeamPlayer

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • 399 posts

Posted 09 August 2020 - 04:17 AM

View Postmorosis, on 08 August 2020 - 06:47 PM, said:

Update:

Today, as a Tier 2 pilot, i played 10 games.

10 out of 10 were Tier 1 games.

The matchmaker is still not working.

Fix it.


Tier 2 is the tier 1 feeder area. Gonna have to be at least tier 3 to get away from the mess that is T1

Edited by OneTeamPlayer, 09 August 2020 - 04:18 AM.


#527 Nesutizale

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Privateer
  • The Privateer
  • 3,242 posts

Posted 09 August 2020 - 06:10 AM

Beside some player knowing other players Name and Tier...yah makeing a screenshot and checking manualy is the only way I know of.

#528 Nesutizale

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Privateer
  • The Privateer
  • 3,242 posts

Posted 09 August 2020 - 08:16 AM

Well there are such people that actualy go through all the names. Have a friend that plays WoWs and 10 seconds in the match he tells you all the player stats of both teams and if its even worth for him to play the match.

Those kind of people exist. ^_^

#529 Tarl Cabot

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Tai-sho
  • Tai-sho
  • 7,780 posts
  • LocationImperial City, Luthien - Draconis Combine

Posted 09 August 2020 - 09:36 AM

With MWO though, the only real way to see if someone was Tier 1 was by logging into the forums, searching to see if said player had posted, check post to see if said player had setup his profile to show current tier. That was the only way we could confirm that we were able to confirm Tier 1 and 2 players who was avg 171 MS and a 0.83 W/L ratio, and that was the overall avg since the change from Elo to PSR, and then not until PGI started the Leadership boards. And said player, before he stopped playing or not playing 10 games/month had a much lower avg MS from Oct 2019 to March 2020. Prior to that the player was dropping 500-1500 games/month with almost 30K games since July 2016.

And as a reminder, the seeded player determines the Tier before it opens up. So a seeded player could see Tier 1 or 3 when the MM opens up. And Tier 3 games is a complete mess since it also includes Cadets, thanks to PGI collective wisdom....

And hai, I do diss on PGI, though I was glad they were able to bring the game/IP back from the dead and had promise but their own BT/MW fanboyism has gotten in the way so many times, falling on their arses more times than not. Do not get me wrong, I love the game/IP but even then the rules for each format does not convert well, especially when one removes the mech/battlecomputer and replaces it completely with a human in FPS mode, and said rules should be used to flavor the next aspect of it. I think I will stop there.. time for another shot...

Edited by Tarl Cabot, 09 August 2020 - 09:51 AM.


#530 MadcatX

    Member

  • PipPipPipPipPipPipPipPip
  • Big Brother
  • Big Brother
  • 1,026 posts

Posted 09 August 2020 - 02:53 PM

View PostNesutizale, on 09 August 2020 - 08:16 AM, said:

Well there are such people that actualy go through all the names. Have a friend that plays WoWs and 10 seconds in the match he tells you all the player stats of both teams and if its even worth for him to play the match.

Those kind of people exist. Posted Image


I did that up until a week ago, would grab the top 200 players from Jarl's and top 30 units for groups, threw it on a spreadsheet on a secondary monitor and refreshed that list about every week, wanted to see what caliber of player MM was matching me with (or more importantly... who IT SHOULDN'T EVER match me with)

I cannot say with certainty that PSR is "improving the quality of the matches", but based off getting hauled up to fight with/against some very good players, I am starting to question the viability of MM being able to match people of approximately the same skill (give or take) with our current playerbase considering, at times, the 1.7K PSR spread in a PuG team (based off me being at 3.2K PSR on the scale and seeing players that are at or near 5K, basing this off their metrics post-reset.. And this is at peak times.

#531 Nesutizale

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Privateer
  • The Privateer
  • 3,242 posts

Posted 09 August 2020 - 11:28 PM

While I appriciate that they finaly did change the PSR system its something they should have done years ago.
Matches have improved slightly for me with more close games then before but overall it dosn't seam to be that much better.

Well I will stick around till the lights go out but I don't see me doing any kind of investments into the game anymore. I have all the mechs I need, mc, cbills, equipment and there is just nothing new that I would say "Okay I will spend a dime for the effort they made".

I did that with Solaris, hopeing that they see if they add something new people will drop some money again but just beeing there dosn't make me want to spend money. Not with so many new games around the corner that I haven't bought yet.

#532 morosis

    Member

  • PipPipPip
  • The Sickle
  • 79 posts

Posted 10 August 2020 - 10:13 AM

View PostLockheed_, on 09 August 2020 - 05:58 AM, said:

I keep seeing people make these claims that their enemies were Tier[X]. How do you know that except for taking a screen shot at the end of the match and checking every single enemy player on the leaderboard/Jarl's list? Am I missing something? Is there a Tier indicator next to the player or do you guys simply know everyone you play against?


identifying T1 games from the lens of a T2 player is easy. its much harder for the other tiers.

if you've been around the game awhile you already know who the great players are. if you watch comp and toc etc, you know who takes the game seriously. if you spend 5 minutes with the season leaderboard, you know who has been active. If you are into stats, you know what good ms, kda and wlr look like, and how active a pilot needs to be to make meaningful progress. If you use Jarls, you know a pilot's historical quality. You can compare pre-reset Jarls to post-reset Jarls to get a sense of whether the reset has impacted a pilot or not. Honestly, we have pretty good tools for identifying high performers. At this point I feel like I know most of T1 pretty much instantly, since its not a huge cohort and you see the same people over and over.

The other thing that you have to know is that because of the "surprise" matchmaker tightening, all matches will be +/- 1 tier except in unusual cases. This matters a lot because it means that T1 can only play itself or T2. So if you're a T2 player, and you can identify 5+ folks in your game that are T1 players, that means its a 1+2 game, which also means that it is NOT a 2+3 game, since T1 cannot play T3.

The rest is just diligence. Screenshot the score screen if you need to, look up enough pilots to feel comfortable about it. You cant do anything in the client during QP search anyway, so better to do something constructive.

Edited by morosis, 10 August 2020 - 01:56 PM.


#533 Cluster Fox

    Member

  • PipPipPipPipPip
  • 104 posts
  • LocationStuck on a rock in Grim Plexus

Posted 10 August 2020 - 04:47 PM

Cadets and former T5s conveniently ranked mid T3 at 2500 PSR are a great farming tool for T2 players to get to T1. And for T1 players whenever they can't get a match in time.

Meanwhile those poor new players get a beating on their first 50-60 games, if they even keep playing...

We haven't seen any update for round two yet. I hope it's going to be surprising and positive.

In the meantime, if you are interested:
  • Go to Jarl's list
  • Check out your avgMS and WLR
  • Plug it in my PSR prediction worksheet and see how you'll do in X matches.
Cluster Fox PSR prediction

Included as a bonus is both Stable PSRs under my fixes to the current system or NightBird's WLR based MM.

#534 morosis

    Member

  • PipPipPip
  • The Sickle
  • 79 posts

Posted 10 August 2020 - 06:47 PM

View PostCluster Fox, on 10 August 2020 - 04:47 PM, said:

Cadets and former T5s conveniently ranked mid T3 at 2500 PSR are a great farming tool for T2 players to get to T1. And for T1 players whenever they can't get a match in time.



This statement is exactly why i keep calling attention to the current MM issues. Everyone, even those closest to this PSR conversation, assumes that T2 is playing against T3 in a fair percentage of their games, because logically, that's what a balanced matchmaker would do. However, this is not what's actually happening. According to the data that i've collected from my own T2 games, I am being placed into T1 games the vast majority of the time. The remainder, somewhere between 15-20% of games, are spent either intra T2 or against T3, its quite difficult to tell with T3 being such a mess.

To illustrate this point, in the 4 weeks since I got to T2, i have seen exactly one cadet across hundreds of games. And that cadet was actually ASH in an alt account, so I'm not sure it even counts. lol.

Update: I played 12 games tonight. 10 were T1+T2, 1 was T2+T3, and the last was unclear. I saw my second cadet.

Edited by morosis, 11 August 2020 - 01:47 AM.


#535 GARION26

    Member

  • PipPipPipPipPipPip
  • Giant Helper
  • 301 posts

Posted 11 August 2020 - 10:04 AM

View Postmorosis, on 10 August 2020 - 06:47 PM, said:


This statement is exactly why i keep calling attention to the current MM issues. Everyone, even those closest to this PSR conversation, assumes that T2 is playing against T3 in a fair percentage of their games, because logically, that's what a balanced matchmaker would do. However, this is not what's actually happening. According to the data that i've collected from my own T2 games, I am being placed into T1 games the vast majority of the time. The remainder, somewhere between 15-20% of games, are spent either intra T2 or against T3, its quite difficult to tell with T3 being such a mess.

To illustrate this point, in the 4 weeks since I got to T2, i have seen exactly one cadet across hundreds of games. And that cadet was actually ASH in an alt account, so I'm not sure it even counts. lol.

Update: I played 12 games tonight. 10 were T1+T2, 1 was T2+T3, and the last was unclear. I saw my second cadet.

It's doing it off PSR not tier.
If you are a higher Tier 2 you are closer in PSR to low Tier 1s then you are to mid Tier 3's. They use the terms +/- 1 tier but it's actually looking at PSR.

If you are dropping at a time zone with fewer low skill players the available population to drop with may also skew to the higher skill players (i.e. weekday evenings US east coast is more tier 1 players then the weekends where more tier 3 players show up)

#536 morosis

    Member

  • PipPipPip
  • The Sickle
  • 79 posts

Posted 11 August 2020 - 12:09 PM

View PostGARION26, on 11 August 2020 - 10:04 AM, said:

It's doing it off PSR not tier.
If you are a higher Tier 2 you are closer in PSR to low Tier 1s then you are to mid Tier 3's. They use the terms +/- 1 tier but it's actually looking at PSR.

If you are dropping at a time zone with fewer low skill players the available population to drop with may also skew to the higher skill players (i.e. weekday evenings US east coast is more tier 1 players then the weekends where more tier 3 players show up)


Thanks for the insight sir, much appreciated. I have been basing my analysis off of what we were explicitly told: that the MM is allowed to select +/- 1 Tier except in unusual circumstances, and that the PSR delta between highest and lowest player in a given game has been significantly tightened.

Idk - even if you're right that its based on PSR value and not Tier, i still feel like i cant explain what I'm seeing in game given the above two parameters. I'm in the middle of T2, I imagine right around PSR 3500. I should not be seeing elite players frequently, because even a mid T1 player is 1000+ PSR distant from me, let alone the elite players, who are 1500 distant. However, instead of being rare, it is pretty much the standard. I've had numerous days where more than 50% of my games have these outlier elite players in them.

Said a different way: if I'm PSR 3500, I think it would be fair to expect that mean player I am matched up with over a decent span of games would also be around 3500. I feel like it has been around 4000 instead, with half the population in each game being T1 and the other half T2. and the presence of so many elite players, who i imagine must be PSR 4500+, just doesn't make sense in any way. The only explanation that seems plausible is that the MM has been told to reduce wait times, so it is opening up the gates for mid T1 games that it cant fill, and finding players like me to plug in.

It is entirely possible that there is a better explanation for what i'm seeing, but given what we have been told, I cant see what it could be.

Edited by morosis, 11 August 2020 - 01:07 PM.


#537 Landragn

    Rookie

  • 5 posts

Posted 12 August 2020 - 02:44 PM

My first Post. So...I'm a six year member. I'm kind of an old nasty buzzard. I've never joined any of the groups. I'm a loner. (my choice) My abilities run my avg match score between 200-300 with the odd 400-600 game scores to keep feeling good about playing. I have avg 300-500 dmg matches, up to the odd 1000 dmg and its been mostly fun. I don't know any hitbox/special tactics stuff and I mostly go for piloting bigger Mechs because the small Mechs make me feel like I'm playing HALO which does not appeal to me. I like the "TANK" sight/sound/feel of the bigger Mechs. I've been a loyal Battletech fan since the board game and no one was more happy when the first MW computer games came out years ago. Played them all.
My point. I'm struggling to have fun since the groups joined Quickplay and the tier/matchmaking patterns have changed. Change...I know...we all have to accept change. Everything changes. I've played through quite a few MWO changes since 2014.
I guess my sadness is from playing through all the stomps now whether its win or lose. I don't see the stomps getting less. I'm wandering through tier 2 now but wow is it chaotic. Either the team is very random acting (mostly) or surgically precise. Either way its a stomp most matches.
To be honest and fair, once in a while I'll be matched with one of the well known capable group teams and it is fun. I know I can push because I know they will push and it is high scores for most everybody. My point? Somebody gets stomped most of the time. If one side is a herd of cats then shouldn't the other side be a herd of cats too? If that is the plan then I'm not seeing it take form yet and it has been a while now.
I have no demands. I know our player base is shrinking. I don't want anything any certain way. I don't cheat...I won't cheat. I just want to enjoy MWO at some minimum or other. I am finding it difficult to achieve this and I guess I am lamenting on what I'll need to do if I can't figure something out. I'm trying to pilot smaller mechs, using LRM boats (not my favorite weapon) and calling the odd match but it is not happening. Not a fearsome dude so any flamers need not put me in my place. I'm not challenging anyone. In all the math and statistical talk in the forum I thought I'd throw in some subjective commentary to voice how I am feeling about MWO right now with the new changes. I hope it is OK.

#538 Cluster Fox

    Member

  • PipPipPipPipPip
  • 104 posts
  • LocationStuck on a rock in Grim Plexus

Posted 12 August 2020 - 06:53 PM

View PostLandragn, on 12 August 2020 - 02:44 PM, said:

In all the math and statistical talk in the forum I thought I'd throw in some subjective commentary to voice how I am feeling about MWO right now with the new changes. I hope it is OK.


Subjective feedback is useful man. What you are experiencing is that the current PSR model we have doesn't really help making closer matches. Sadly this is about as good as it's going to get. Larger spread of W-L = larger amount of Stomps. Because Stomps are more likely to happen the less the teams are balanced.

Check this out: Same players, same simulation, same mechanics, same MM logic, different ways to calculate PSR.

Narrow and High = closer matches and better team balance.
Posted Image
Current system is the orange line. The player base deserves better.

More details here for the full story on that simulation

Edited by Cluster Fox, 12 August 2020 - 06:54 PM.


#539 ERSmurf

    Member

  • PipPip
  • Knight Errant
  • 20 posts

Posted 13 August 2020 - 03:52 AM

48 (Jul'20) 15247 10% 67 0.43 10% 0.12 118 116.3 -29% 22% 28% 49% 0%

47 (Jun'20) 12929 18% 129 1.08 18% 0.22 168 163.7 15% 11% 51% 33% 0%

On this account above and this below on my main (ERescue)...

48 (Jul'20) 12983 23% 211 0.77 18% 0.29 158 154.1 -19% 17% 7% 76% 0%

47 (Jun'20) 10243 35% 82 1.28 29% 0.40 187 190.9 23% 48% 18% 32% 0%

Both of my accounts took a massive nose down in results. Even more interestingly, without having executed any precise count, it feels both extremely balanced (10 or 11 vs 12) AND stomp (0-3 to 12) games have gone up and the so to speak average (5-7 vs 12) games are the rarest of all.

This account has already dropped to T5 and my main account will likely follow suit soon. I was at T5 before the reset so I am not worried as such, but I do wonder are some people at T1 shooting up to very maximum of PSR already?

What I am slightly worried about is the fact there have been no more updates on changes of numbers of people in various tiers.

#540 Tarl Cabot

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Tai-sho
  • Tai-sho
  • 7,780 posts
  • LocationImperial City, Luthien - Draconis Combine

Posted 13 August 2020 - 06:04 AM

View PostNightbird, on 10 August 2020 - 08:56 PM, said:


Speak for your self... WLR will make it as hard as possible to stat pad by grouping, since after a point, your group of 4 will always be teamed with the 8 worst players out of the remaining 20.

If you can still rack up wins with that, great, you deserve it. Most 4 mans can't and therefore will reach some equilibrium. We'd see very few people with over 3WLR under my system. Under Jay Z's, 30WLR, 50WLR, the sky's the limit. The same will be true under your system because you cap PSR and so under-estimate the impact of grouping and high WLR players.


Pulled this from another thread. Simply pointing out, as per Nighbird's post, of how a MM setup would work towards keeping low W/L ratio, for both sides of a drop. So in this drop, one side would have a killer group with "drones" or meatshields/cannon fodder, players with the lowest W/L ratio of that drop, and potentially the lowest Matchscore, Damage, preferably accurate dmg > followed by kills/kmdd, etc wins most games, even more so while working as a team.

And this is one of the first times I have seen where Nightbird posted an example, general example since weight class, etc is not notated.





4 user(s) are reading this topic

0 members, 4 guests, 0 anonymous users