Jump to content

- - - - -

The Player Skill Rating (Psr) System Explained... (As Best I Can)

Guide

149 replies to this topic

#121 L Y N X

    Member

  • PipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 629 posts
  • LocationStrana Mechty

Posted 14 January 2016 - 04:20 AM

1/14/2016 Made Tier 1 today.

Edited year... was a late night!

Edited by 7ynx, 14 January 2016 - 01:01 PM.


#122 Sky Hawk

    Member

  • PipPipPipPipPipPipPip
  • 700 posts
  • LocationDeep Periphery, aka Hungary

Posted 14 January 2016 - 05:00 AM

Congatulation!... And.. Ahmmm.. Welcome in 2016!

Edited by Sky Hawk, 14 January 2016 - 02:27 PM.


#123 ztac

    Member

  • PipPipPipPipPipPipPip
  • 624 posts

Posted 14 January 2016 - 11:49 AM

People always forget that individual performance in a team based game is meaningless. You can not rate player ability based on luck or team composition as these are random factors .....

Fact is there are just too many variables in a game like this .... from team composition to mech builds or even how far along skilling up your mech you are! Essentially it is a rating based on random factors.... and to make matters worse some of this can be gamed.

#124 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 14 January 2016 - 12:01 PM

View Postztac, on 14 January 2016 - 11:49 AM, said:

People always forget that individual performance in a team based game is meaningless. You can not rate player ability based on luck or team composition as these are random factors .....

Fact is there are just too many variables in a game like this .... from team composition to mech builds or even how far along skilling up your mech you are! Essentially it is a rating based on random factors.... and to make matters worse some of this can be gamed.

This may be true to some degree but since there are those that have not been able to increase their tier since the system was implemented, I'd have to say it is doing something right.

#125 L Y N X

    Member

  • PipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 629 posts
  • LocationStrana Mechty

Posted 14 January 2016 - 01:08 PM

View Postztac, on 14 January 2016 - 11:49 AM, said:

People always forget that individual performance in a team based game is meaningless. You can not rate player ability based on luck or team composition as these are random factors .....

Fact is there are just too many variables in a game like this .... from team composition to mech builds or even how far along skilling up your mech you are! Essentially it is a rating based on random factors.... and to make matters worse some of this can be gamed.



The STRENGTH of the chain lies within the WEAKEST link.

Point being, improving individual performance increases the team's likelihood or chances at winning. Regardless, improving one's own performance increases the PSR progress at a faster rate than not improving. Empirically speaking, this is not a skill rating but rather an experience rating as I have previously stated. It is not perfect, however it is better than anything PGI implemented to date. My matches rarely consist of players who do not know what they are doing anymore. At Tier1, PGI has said that I do not get matched up against or with Tier 5 players in the solo queue. That should be incentive enough to want to improve one's own experience and performance on average in any given match.

Edited by 7ynx, 14 January 2016 - 01:11 PM.


#126 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,585 posts
  • LocationSeraphim HQ: Asuncion

Posted 14 January 2016 - 07:07 PM

How's this one?

I loaded into a match. As soon as I was in the ready screen... I lost connection. (As in, internet was no more.) By the time I loaded into the match, I was already dead.

I somehow did 1 point of damage and got one kill assist. Match score of 1. (Duh. I wasn't there.)

PSR goes down. *Sigh*

Not MW:O nor PGI's fault I seemed to have lost my connection. (No idea what happened.) But it still sucks... Posted Image

#127 Hexenhammer

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,729 posts
  • LocationKAETETôã

Posted 14 January 2016 - 08:23 PM

Only thing I learned when I hit t3 was how nasty t2, t1 players can be in VoIP.

#128 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,585 posts
  • LocationSeraphim HQ: Asuncion

Posted 14 January 2016 - 08:32 PM

View PostHexenhammer, on 14 January 2016 - 08:23 PM, said:

Only thing I learned when I hit t3 was how nasty t2, t1 players can be in VoIP.


Oh? Does your team always seem to suck too? Don't worry, I'm told my team sucks all the time. Posted Image

#129 L Y N X

    Member

  • PipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 629 posts
  • LocationStrana Mechty

Posted 15 January 2016 - 03:15 PM

If you improve your performance on a consistent basis (on wins as well as losses, one will always have power outages and such), would you expect the average match score to decline, remain the same or increase over time?

If the average match score increases over time, do you expect PSR to increase faster or slower or decrease?

If PSR increases faster, do you expect you will eventually cross a tier boundary?

Remember that one's average is the worst of their best and equal to the best of their worst! Strive for more!

#130 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,585 posts
  • LocationSeraphim HQ: Asuncion

Posted 15 January 2016 - 08:46 PM

View Post7ynx, on 15 January 2016 - 03:15 PM, said:

If you improve your performance on a consistent basis (on wins as well as losses, one will always have power outages and such), would you expect the average match score to decline, remain the same or increase over time?

If the average match score increases over time, do you expect PSR to increase faster or slower or decrease?

If PSR increases faster, do you expect you will eventually cross a tier boundary?

Remember that one's average is the worst of their best and equal to the best of their worst! Strive for more!


I'm going to take this is an open ended question for anyone to respond to...

If you improve your performance, which means your average goes up, eventually PSR (or any proper ranking and match maker system) goes up enough to drop you with tougher and more skilled opponents. This, by nature, should drop your "average" back down till your skill catches back up and you improve some more. Tougher opponents means harder to get the same match score. (In theory.)

Average match score should remain average, and shouldn't "go up" if you get ranked with players of appropriate skill. However, I expect PSR to try and keep players within a ranking, unless they continually perform "above average" for that ranking/rating. If they typically perform "bellow average" then they should drop in ranking/rating. This should happen until they start to perform average again by more appropriate match average skill levels.


My concept is a system that wants players to get an average score. It should adjust players until they achieve that average as an average, and then keep them there (until their skills improve again or decline). It should be harder to move in tier overall, and easier to remain within a tier (unless you start to seriously perform better or worse drastically, which would indicate you are not in the proper tier).

Tougher opponents should mean harder games. Harder games means harder to get the same average you once use to get. One example of this would be like if each difficulty in a timed match, your targets gained more health. Eventually, you will get to a point where you achieve whatever is "average" clear times because of improved skills (presuming damage output doesn't change). This is because the difficulty to achieve your target goal got harder. (Think of the turret attack mode in the training grounds. Each wave it gets harder and harder because the turrets have more health. Eventually, it gets so hard you fail to deal enough damage. Over time, even with the same build, you can get a little better scores and farther along the waves. Eventually, your average wave max should increase as you get better, till you hit a cap that you can't go farther. This would be, in PSR terms, your rank/rating.)


I believe I've mentioned these ideas before here. If so... ignore? Or if confused, mention where I need to elaborate? (Sometimes I don't state things very well...)

#131 Hexenhammer

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,729 posts
  • LocationKAETETôã

Posted 15 January 2016 - 09:01 PM

lol thats every tier.

But it wasn't until I hit T3 did I start hearing people on comes complain about how lousy someone's build is.

The only diff between t5 and t1 play is the steam rolls happen faster.

Edited by Hexenhammer, 15 January 2016 - 09:02 PM.


#132 L Y N X

    Member

  • PipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 629 posts
  • LocationStrana Mechty

Posted 16 January 2016 - 02:56 PM

View PostHexenhammer, on 15 January 2016 - 09:01 PM, said:

lol thats every tier.

But it wasn't until I hit T3 did I start hearing people on comes complain about how lousy someone's build is.

The only diff between t5 and t1 play is the steam rolls happen faster.



Hex, Generally speaking T1 Rolls happen faster because, in general, T1 players are more more aggressive... Have more efficient builds for damage output, and have learned to focus fire whenever possible so that the mechs just die faster which leads to faster rolls and shorter matches.

Ten, please be clear when you mean the tier average vs the players own average. I think I agree with you thinking, but in a couple places I was unsure whose average you meant.

#133 Tesunie

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Seeker
  • The Seeker
  • 8,585 posts
  • LocationSeraphim HQ: Asuncion

Posted 16 January 2016 - 03:19 PM

View Post7ynx, on 16 January 2016 - 02:56 PM, said:

Ten, please be clear when you mean the tier average vs the players own average. I think I agree with you thinking, but in a couple places I was unsure whose average you meant.


Player average. There should be an "expected average" of a player every match, probably considered around 200-300 (250?). No matter the tier (in theory), that average should still be a player's average match score (with some matches being under and some over). The purpose of the tiers is to place players in matches with equally skilled players, which should mean that the matches are harder which makes it harder to get "average" with a previous tiers skill/performance.

In theory (how it works in practice... I don't know, would need testing), a player should be getting an average match score unless they are inappropriately matched for their skill level. A T1 vs a T5 should (once again in theory) result in the T1 winning and having an average match score. However, a T1 vs a T1, each should be equally skilled, which means that either one could win, and each of them (over time as match averages) should be able to get a similar match score.

AKA: The higher in tier that you go, the more difficult your matches should be, the harder it is to get the same average match scores. Thus meaning that you should still be challenged to get an "average" match score. If you are over ranked, then you should be getting an above average match score, which should boost up your ranking until you face opponents that are more skilled, resulting in you getting a more average match score again.

(I'm not sure I can explain any clearer? It's not something easy to describe via text. Posted Image )

#134 Rogal Thorne

    Rookie

  • Giant Helper
  • 1 posts

Posted 24 January 2016 - 07:41 AM

Hi people! First post after 9 days of playing and 250+ matches... I played a ton of starcraft2 before and quite a lot of dota2 and this skill rating thing always fascinated me. Right now i think mwo has a good rating system, just a few points:

1) 12 vs 12 is a great mode, fun for all but of course much more difficult to extract meaningful data for psr. take a slightly skewed 12 sided die. How do you determine how much it favours a '12'? you either put a full lab with equipment and a supercomputer on it... or you throw the die. A lot of times, 100 times will not suffice. Thats the same with mwo, you represent 4% of the overall 'skill pool' in any given match, and thus your influence on the outcome is minimal. However your influence on the outcome of 100 matches is significant.

2) The emphasis on match outcome vs a personal metric achieves two things: Firstly it incentivizes team friendly play and secondly it means all those 'soft' skills like communication and scouting etc. factor into your psr without PGI putting a metric on it.

3) The worst thing imo that could happen would be if PGI divulged how exactly they compute their psr changes.. that would quickly lead to some 'metabuilds', making cues worse, decreasing variance in builds, shifting the focus of many players on maxing out their psr rank instead of enjoying themselves.

4) We all want that pat on the back when we performed great, but it does not happen in every game in this system. If I play this game for another couple of months I will probably want a rank advancement to reflect my increase in skill. But considering that i guess 99% of the playerbase thirsts for more xp/credits, being undervalued in this system is pretty much a free premium bonus, since you will get those kills so much faster

5) The skill number is just a number, and there are many factors that require suboptimal play, like leveling certain mechs.. the way i see it right now people who want to climb the ranks should look at reliable and stable builds that get those deeps applied to the enemy, probably fast heavys with a good alpha.

I think that 'matchmaking hell' syndrome is found in any team game, your teammates faults will always be more apparent to you than your enemies'.. when in fact they are both -on average- the EXACT same skill level. Asking for a 1 to 1 representation of your performance is human as well, but in a 12 vs 12 game it is not really possible, inevitably causing frustration. PSI have wisely put very little emphasis on PSR level, it gives zero benefits but of course it immediatly becomes a very important topic here nonetheless.

#135 4chtun9

    Member

  • Pip
  • Liquid Metal
  • 11 posts

Posted 15 February 2016 - 07:09 AM

View PostTesunie, on 06 January 2016 - 06:27 PM, said:

Then, I'm not the only one here...

So... if we all have it wrong on how it works (as it was explained to us by PGI), care to enlighten us? Or you going to sit there and just continue to say how flawed everyone else is and never give any real response?

Did I miss anything major? Did I mess something up here?


No you are not; unfortunately it would seem the majority of posters have simply no understanding of ELO.

Ignoring that that is not even close to how PGI explained it, he seems to be one of the few that actually understands ELO and did attempt to 'enlighten' you.

Yes you missed a few major points and yes you seriously messed some things up. No worries though, you are not the only one.

#136 L Y N X

    Member

  • PipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 629 posts
  • LocationStrana Mechty

Posted 02 March 2016 - 12:22 PM

2chtun9,

Care to explain? The way I understand it, the PSR system PGI has put in place still has influences from Elo. All the emphasis on winning for example to make the PSR go up is Elo-esque. However, Elo was not intended for multiplayer team ratings.

#137 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 09 March 2016 - 02:28 PM

View PostTesunie, on 06 January 2016 - 06:27 PM, said:

Elo as I understand it:
- Is based on W/L.

Elo ratings are derived from the results of playing the game, using win or loss as the only input. That's not the same as saying it is "based on W/L."

Quote

- It gives a rating/number to a player based on their individual W/L rate.

Sort of. It isn't based on the player's W/L rate, it's based on their results against other players. That sort of sounds like W/L, but it actually isn't.

For example, you can't look at a player's stats and say "this player has a W/L rate of .500, so his Elo rating is 1600." His Elo rating will depend on the ratings of the players he beat and lost to, not simply on his W/L rate.

Quote

- It then tries to pair that player with another player (single player related) who has a similar if not exact rating. (Or for teams, tries to create teams that all player Elo scores equal a certain total per team.)

False. Elo is a rating system only, it doesn't have anything to do with matchmaking.

A matchmaker can use Elo ratings as input into its matchmaking process, but that's not at all the same thing.

Quote

- If there is a disparity of their Elo rating, then the formula predicts (by what PGI said of their own formula) which player (or Team for MW:O) is likely to win.

An expected outcome can be derived from relative Elo ratings, yes. But that's not really "predicting" the outcome of the match like people think. It simply gives you a percent chance that side A will beat side B.

PGI's matchmaker attempted to keep the aggregate Elo ratings of the two teams equal. If the two sides' Elo ratings are equal, then the expected outcome is 50/50. I.e. both sides have an equal chance to win.

Quote

- If the player who is predicted to win wins, then there is little change in their rating. If the player predicted to win losses though, then changes happen accordingly.

This is where PGI deviated from a proper Elo system, and their deviation made the system less responsive. Their change still works, it just works even more slowly than normal.

In a proper Elo system, your rating will change after every match (with some minor corner-case exceptions) in line with the expected outcome.

What that means is that if you and I play, and I have a 90% chance to win, then if I win there will be very little change in either of our ratings. But if I lose there will be a very large change in both of our ratings. If I only have a 50% chance to win, then win or lose both of our ratings will change by a moderate amount. The winner's rating always increases while the loser's rating always decreases. (Again, except for some corner cases that aren't worth worrying about.)

Quote

- As stated before, is a system designed best for solo PvP (as in, one player on each side, not a team of players per side) styled games, such as chess (or checkers, or boxing even, or Battleship, or...). In that area, it is very effective.

It was first used for chess, but it isn't correct to say that it was "designed for Solo PvP" in any way. Elo ratings are simply a mathematical formula that works in any 2 sided system. Each side can have as many players on it as you want.

Quote

- Doesn't work so well for Multi-player, as each player could have vastly different actual skill levels within the (randomly) assigned team (for PUGs).

False. Elo works just as well for multi-player as it does for 2-player, it just takes longer to reach equilibrium. PGI's Elo ratings took even longer because of the change they made. However, players with thousands of games easily had accurate Elo ratings even with PGI's modified Elo ratings. (The math says that a couple hundred games at most would be plenty, and possibly as few as 50.)

Quote

Did I miss anything major? Did I mess something up here?

You fell into the typical gamer trap, that's all.

Elo ratings are a very simple, very accurate mathematical formula. All of the rest of that baggage really belongs to the PGI matchmaker and it's use (or misuse) of Elo ratings as input.

#138 PFC Carsten

    Member

  • PipPipPipPipPipPipPipPipPip
  • Little Helper
  • Little Helper
  • 2,188 posts
  • LocationOn your six

Posted 09 March 2016 - 03:29 PM

View PostRogal Thorne, on 24 January 2016 - 07:41 AM, said:

3) The worst thing imo that could happen would be if PGI divulged how exactly they compute their psr changes.. that would quickly lead to some 'metabuilds', making cues worse, decreasing variance in builds, shifting the focus of many players on maxing out their psr rank instead of enjoying themselves.

No magic here. It's match score, pure an simple. With fixed amounts for win and loss respectively needed to go up and down.

#139 S 0 L E N Y A

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 2,031 posts
  • LocationWest Side

Posted 09 March 2016 - 05:22 PM

View PostPFC Carsten, on 09 March 2016 - 03:29 PM, said:

No magic here. It's match score, pure an simple. With fixed amounts for win and loss respectively needed to go up and down.


This, so you could argue the primary factor is win/loss, however that is largely boosted or off set dmg.

However, when you look at the system as a whole, it is much much much harder to go down than it is to go up.
IE: anything is possible on a loss, including moving up, however it is not possible to go down on a win. If you put up just the tiniest positive effort (generally damage) on a win you will move up.

#140 bar10jim

    Member

  • PipPipPipPipPipPip
  • 352 posts

Posted 09 March 2016 - 06:02 PM

I've always felt that the PSR needed to be "adjusted" for matchmaking purposes based on the mech that you are piloting. Now, I don't mean having a separate PSR for each variant, chassis, or weight class. Merely and adjustment considering if the mech you are piloting has completed Basic or Elite levels of training.

For example: as a player you have a PSR of 30% of tier 3. If you have not finished Basic in your current mech, you enter the queue a full tier lower (30% tier 4) for matchmaking purposes ONLY (your base PSR stays the same, and is adjusted as normal following a match. If you have completed Basic, but not Elite, you enter the queue 1/2 tier lower (80% tier 4 in the above example).

PGI could vary the "spread" or percentage of tier reduction as result of performance metrics to better tune the matchmaker results and match balance.





11 user(s) are reading this topic

0 members, 11 guests, 0 anonymous users


  • Facebook