Jump to content

Is Psr / Tier Accurately Reflecting Skill Level?


56 replies to this topic

#1 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 24 September 2015 - 04:24 AM

Hi All,
My PSR/Tier seems to be accurately reflecting my skill level. However, I took a quick survey of my unit and the Tiers seemed "off" by a little for some pilots and "way out of whack" for one or two pilots in particular.

The end result is that I have some skilled veterans playing in Tier 4 against new pilots.
The end result is that I have a skilled veteran who has been on hiatus / infrequent play who is in Tier 5 with the steering wheel brigade.
The end result is that I have one pilot who plays often in the group under one account and is Tier 2, but his alter-account (he plays infrequently when he wants to play alone) is Tier 3. Same pilot skill, same mechs... different result.

We are drawing a conclusion that your PSR seems to be linked to the frequency of play as much as skill.

These Tier misalignments have lead to some of my middle-of-the-road pilots serving out some whopping scores against Tier5's... no fun for either party there.

My question to you guys, what have you found in your units? Are your infrequent pilots accurately ranked/tiered? Are your most active pilots your highest rank? Is this reasonable?

Since the PSR algorithm is hidden and PSR values are not public... it is really difficult to accurately tell if PSR is broken already.

-- NomadicCanuck.

#2 Ironwithin

    Member

  • PipPipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 2,613 posts
  • LocationGermany

Posted 24 September 2015 - 04:35 AM

Every player's initial PSR was calculated using matchscore data reaching back to january 2015.
So people that have been inactive since then or inbetween then and now may have an inaccurate score.

The system is not magical, it cannot know how "good" a player is simply from them logging in.
It needs data to make calculations. That data comes from matchscores.

The inaccuracies come from it being a new system (with a set cut-off date for archived data), with time they will shrink away. There is no way around that.

#3 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 24 September 2015 - 05:11 AM

All players in my post above have played since January. The difference is some of us play daily, some play a few times a week and some only play a few nights per month. In any case, all pilots have active profiles.

I am aware, as you said, of the January 2015 date which is why we were all shocked at some of the Tiers. I have a pilot that was ranked in the upper 20 for the last tournament; He has been playing for years and is a solid contributor to any match he's in although he only plays about one night per week. He is ranked Tier 4.... and has been destroying the solo queue. That seems "off" to me.

Seriously, ask in your unit to see if your infrequently playing pilots are ranked below their station and your frequent pilots ranked above. That's the question.

#4 Josef Koba

    Member

  • PipPipPipPipPipPipPip
  • Wrath
  • Wrath
  • 527 posts

Posted 24 September 2015 - 05:11 AM

I can only report my own (biased) experiences. What I can say is that ELO must have put me in a lower bracket, since prior to the implementation of PSR I was routinely at the top of my team in kills, damage, and match score. Since its implementation, I've become a comparatively atrocious pilot. Routinely in the absolute BOTTOM of the match. Heck, I sometimes have trouble breaking 200 damage. Perhaps PSR has me where I'm supposed to be, but if so...man that's a frustrating thing to realize I suck so badly in tier 3... (laugh out loud)

#5 Thorqemada

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 6,389 posts

Posted 24 September 2015 - 05:59 AM

Playtime and success... ('# of matches + W/L).

#6 n4sty

    Member

  • PipPip
  • Mercenary Rank 5
  • 31 posts

Posted 24 September 2015 - 06:26 AM

The problem with PSR is very simple: it does not change quickly enough to refelect the player's actual skill. Very good (bad) players who played much CW in 2015 and/or only infrequently on standard maps have a too low (high) rating... that's it.

I think there are many solutions to this and PGI will calibrate the system in future. E.g. a "winning (loosing) streak" system could help

Edited by n4sty, 24 September 2015 - 06:26 AM.


#7 Anunknownlurker

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 362 posts
  • LocationBetween here and there

Posted 24 September 2015 - 07:22 AM

This has become a major topic amongst my MWO friends.

Now, let's get this over with; I am tier 5 (about half way up to tier 4). I cannot deny this surprised me. I always viewed myself as an average pilot, over 2500 games my win/loss is running around 50% and, whilst my kill/death is negative overall, on most of the mechs I play now, I have a low positive result (it took me a long time to learn the game and find the mechs I liked, sorry!). My average damage on my most commonly played chassis hovers around 300 and I, like most average players I am certain, have had the odd 1000 pointer as well as the occasional total derp. I've even demonstrably won games myself.

OK, so, either I am not an average player; maybe I am deluding myself, maybe those figures above are not as good as I think they are and I am a poor player, that's fine, I can accept that; or tier 5 is where average players are meant to live, which strikes me as an odd place to put average!

Don't get me wrong, I don't really care where I am, this is not an epeen exercise for me and I am all in favour of some kind of skill-based match making in a game like this; but, I need to have confidence that I am actually facing players of a similar skill level otherwise, frankly, what's the bloody point?

Just my 2 cents.

Good hunting!

Edited by Jimmy DiGriz, 24 September 2015 - 07:23 AM.


#8 Modo44

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • 3,559 posts

Posted 24 September 2015 - 07:49 AM

PSR reflects your match score, not how much you contribute to wins. It is close to player skill, but not as direct as Elo. Literally nothing is as good of a skill indicator as Elo because other methods always bias towards counting whatever the author thought was important.

Edited by Modo44, 24 September 2015 - 07:56 AM.


#9 Commissar Aku

    Member

  • PipPipPipPipPip
  • Urban Commando
  • Urban Commando
  • 195 posts

Posted 24 September 2015 - 09:08 AM

As far as I can tell your tier rating is based almost solely on your W/L ratio. If you get a great score ie 3+ kills and 500+ damage with a match score of 300 or so but your team still loses you rating still goes down, no matter how many games you play, as long as you lose you more or less go down. Same with winning you can have a total derp game 54 damage or something, your score doesn't go up or down. If you win and do 150+ damage your score will go up. Seems to me that this punishes good players for having bad players on their team, and rewards bad players for having good people on their team. This used to be how MWO choose its games in the match maker and it was a total fail, to see them try it again is a slap to the face to anyone who has played this game for over a year.

#10 Anunknownlurker

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 362 posts
  • LocationBetween here and there

Posted 24 September 2015 - 09:15 AM

@ Modo44 - yes, the current iteration of PSR reflects your match score but, in my opinion, is too heavily weighted for the win. E.G. a win with 120 damage will mean your PSR will go up, where as a loss with 400 damage and it will stay the same (I think that's what happened to me last night). Now, I can't remember the exact details of component damaged, spots etc but I do remember I did next to bugger all in the win and worked my socks off on the loss!

I believe your description of how it works is how it was intended to work, it simply doesn't do it. To take an extreme view, whether you win or lose is actually very little to do with your own, individual skill, there are 23 other players who have something to do with that, I accept that is not the whole truth but it's close; there aren't many players who are capable of carrying a weak team.

Elo is great for 1 v 1 games (as it was designed) but is pretty much useless for team games.

What I would like is to see the current PSR shifted slightly to weigh individual contribution far more highly than win/loss.

#11 NomadicCanuck

    Member

  • PipPipPip
  • The 1 Percent
  • 53 posts

Posted 24 September 2015 - 04:37 PM

As I mentioned before, one of our pilots has two accounts one for social play, one for solo play... same pilot skill, similar (same?) mech chassis, vastly different tier.

As Jimmy says above, this is about matching with a similarly skilled group of players. Those of us who are experienced and have a long history with the game, want to fight against the same to provide a challenge. New pilots do not want to fight against us as they end up in scrap. The steering wheel brigade wants to jump in and play against others that share their struggle as they learn to pilot etc.

So far, the PSR tiers feel like the ELO episode (remember this?) where the more you played the higher your elo rose? Same thing. I think the PSR / tiers is a good idea but I'm not sure the implementation is as expected since my observed results seem to be all over the map.

(also, wouldn't a 6 tier system where pilots are bell curved with each standard deviation being a tier, make more sense? I digress, that's another thread...)

#12 Ironwithin

    Member

  • PipPipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 2,613 posts
  • LocationGermany

Posted 24 September 2015 - 04:51 PM

Maybe just de-coupling the PSR evaluation from winning/losing might help.
High matchscore ^ PSR goes up, mediocre matchscore = PSR stays, low matchscore v PSR goes down, no matter the outcome of the match.
I'd like to know if those numbers are relative to the team (they should be) or just hard nurmericals that are the same for everyone in every situation.

#13 Alzarns Fire

    Member

  • PipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 221 posts
  • LocationNew Zealand

Posted 24 September 2015 - 04:54 PM

Short answer... no.

First of all you cannot go down during a win... no matter how badly you perform. But despite some good results during bad losses (I.e. top damage, top score) my PSR still went down... illogical.

Also in my unit we quite often have fun nights, or group up with the children of pilots so we naturally get nights of lots of bad matches... this lowers PSR.

The way around it is to only play competitive builds with serious teams to build up PSR... but takes away community and fun factor.

So I've told my unit we are ignoring PSR for unit purposes, we will continue to look at pilots based on match results when considering promotions and competitive teams.

#14 Fahr

    Member

  • PipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 114 posts
  • LocationBeeville, Tx

Posted 24 September 2015 - 06:46 PM

feeling similarly - one of my teamates plays infrequently and is relatively new to the game, he's tier 4 - I on the other hand play way more games, but tend to run light builds - I'm tier 5 (near the teir 4 end by the R in Tier). team play is paramount for me to succeed, but it is nearly impossible for me to get the "good match score" on a loss, most games I can't even manage the loss break even scores. this turns the PSR system into really a W/L counter for me, and since I lose as much as I win, and Losses count more (when you have reasonably average match scores for a light 150-300) I will trend to the bottom tier. while someone who sits in a heavier mech has more Damage potential and will have an easier time getting the match score needed in a loss to break even, they will tend towards the top tiers.

since the system is supposed to give you even matches where your W/L is 50% I think this will end up striating players by metagame factors like build efficiency and mech choice.

#15 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,586 posts
  • LocationSeraphim HQ: Asuncion

Posted 24 September 2015 - 07:16 PM

PSR is suppose to "reward" good team play and pilot skill. However, it rewards high match scores (sorta a decent indicator of skill, but is bias on damage too much) and wins.

I've been RAISING my PSR by fielding one of the most team hindering (depending upon how it's used) and selfish (depending upon how it's used) builds I have, an Arctic Cheetah with ECM and dual CERLLs. I've been getting rewarded for sniping at long ranges, and running away from fights directly. (For the most part, but I often can't stand it for much past the opening of the game, and unless I'm really distracting, I often times will jump in and try to help teammates still.)

However, my PSR was dropping when I was fielding more common place "team" builds, be it LRM support, or brawler/midrange trooper. I worked well with the team, helped to press the enemy, and even lead some charges as the team needed.

I earn ~200 points (between 200-300) and win, PSR goes up. Gain over 300 (I believe it was near the 330 mark) and lost, PSR went down.

Basing a "skill" ranking system on winning or losing isn't indicative of skill. It's just a reskinned Elo system, with one alternative way to go up on a lose really. (Remove the win condition and I think it might better reflect actual player skill.)

#16 rolly

    Member

  • PipPipPipPipPipPipPip
  • Guardian
  • Guardian
  • 995 posts
  • LocationDown the street from the MWO server

Posted 24 September 2015 - 10:21 PM

Link to my post on another thread (covering same topic)

Essentially the same thing, supporting the OP's argument. My friend and clanmate rarely plays (150 matches) vs my 550 matches. Yet he has a higher Tier 4 (3/4th bar) vs my Tier 5 (14/15th bar).

http://mwomercs.com/...-rating-system/

I do a lot more to support my team but apparently even assists don't help.

#17 Kmieciu

    Member

  • PipPipPipPipPipPipPipPipPip
  • Urban Commando
  • Urban Commando
  • 3,437 posts
  • LocationPoland

Posted 24 September 2015 - 10:45 PM

http://mwomercs.com/...37#entry4719837

The two factors to PSR are:
winning
match score

And what gives you best chances of getting both? A meta mech.

A correct Direwolf build can get you a high matchscore even on a loss. Because lots of damage = high matchscore. We all know that after the last tournament.

Incidently, a Direwolf is a mech that also gives you a high chance of winning a match, simply because it is unmatched in 1vs1 and requires a team effort to take down.

Edited by Kmieciu, 24 September 2015 - 10:47 PM.


#18 Meihru

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 378 posts
  • LocationGrafschaft Bentheim/Germany

Posted 25 September 2015 - 03:30 AM

View PostKmieciu, on 24 September 2015 - 10:45 PM, said:

http://mwomercs.com/...37#entry4719837

The two factors to PSR are:
winning
match score


and thats the main problem here, heavys/assaults with normally high dmg outputs (like timber, dire, King crab, Zeus) will get the player rated like tier 1 or 2, while especially light pilots, which normally wont do as much damage will get rated tier 4 or 5 even if they provide their team with the informations needed to win and kill a whole lance on their own by shooting their backs simply because they dont do as much damage this way
in fact, in my Unit every heavy/assault pilot is rated tier 1-2 and some 3 now, while our Mediums and Lights are ALL rated 4 or 5, with only one who barely made tier 3

in short: you get rewarded for spreading your damage all over the map instead of ensuring a win and killing efficiantly

and thats why no one is using gauss or PPc anymore, they are "too precise" to make a lot of dmg, everyone uses Lurms now because of the spread

View PostKmieciu, on 24 September 2015 - 10:45 PM, said:

And what gives you best chances of getting both? A meta mech.

A correct Direwolf build can get you a high matchscore even on a loss. Because lots of damage = high matchscore. We all know that after the last tournament.

Incidently, a Direwolf is a mech that also gives you a high chance of winning a match, simply because it is unmatched in 1vs1 and requires a team effort to take down.


well i can **** your dire with any light you give me by cracking your back, but the psr wont reward that because it isnt worth anything to kill a dire in a locust as long as i dont do tons of damage

#19 Koshirou

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 827 posts

Posted 25 September 2015 - 03:58 AM

View PostKmieciu, on 24 September 2015 - 10:45 PM, said:

The two factors to PSR are:
winning
match score

In my experience, there are five factors:
1. Winning
2. Winning
3. Winning
4. Winning
5. Match score... which is also partially influenced by winning

Edited by Koshirou, 25 September 2015 - 03:59 AM.


#20 Inflatable Fish

    Member

  • PipPipPipPipPipPipPip
  • Rage
  • Rage
  • 563 posts

Posted 25 September 2015 - 04:04 AM

the major flaw in the system is that it doesn't reward good play, instead it rewards primarily doing damage.





11 user(s) are reading this topic

0 members, 11 guests, 0 anonymous users