Pilot Rating

[Som31Else]

I know, I know, its been said many times before but I just wanted to ask how tf pilot rating actually works, I just came away from a round with 756 damage given, time spent capping in dom, many components destroyed, team only killed 4 in total and I had assists for all of them.

I walk away with a no change pilot rating, wtaf

[salamatmahal]

There's equations and math you can do, but the hand-wavey version is...

The higher your match score relative to the other players in the match, the more your PSR moves up.

The lower your match score relative to the other players in the match, the more your PSR moves down.

But if you just love reading and math and you've already finished your SATs...

PSR change when your team wins = PSR_W = 20*((0.6*my_MS/avg_my_teamMS)+(0.4*my_MS/(avgALLMS))-15

PSR change when your team loses = PSR_L = 20*((0.6*my_MS/avg_my_teamMS)+(0.4*my_MS/avgALLMS)) -25

You do get match score for various 'helping the team' things but unless you are doing a low amount of damage, your damage dealt will tend to be the most significant contributor to your match score.

You can PSR up with a very very low match score in the 'win on basecap without a fight happening' scenario, but this is somewhat rare. (most people don't play the stompy robots with guns game to not shoot things)

If your contributions legitimately contribute to your team winning in a statistically significant way, you are likely to rank up your PSR over time simply due to the bonus from being on the winning team even if you don't do much damage. That said, it's probably easier to rank up your PSR by consistently doing damage, which is also likely to contribute to your team winning if you do enough (effective) damage.

Oh, also the mech class really doesn't matter much as far as dealing damage goes. Mechs in every class are quite capable of doing 1k damage in a quickplay match. Lights are less likely to do it, but not massively so.

[Som31Else]

then 2 rounds later, skirmish, 275 damage done, similar "help" stats, win, go positive, I don't feel damage matters as much as believed.

Not blowing a trumpet or anything, tier 4 but have been tier 2 in the past, I'm pretty much top 3 match score every round and regularly top, think I run about 280 average on the jarls.

[spartanspud]

Pilot rating, despite what many people will tell you, is a 0 sum game which is relative to the performance of everyone else in the match. If you get 300 damage in one match where everyone did 200 you will go up.
If you get 700 but everyone else got 800, you will go down.
There's also preference given to winners. But you can still go up if you lose or vice versa.

[Shiro]

This is an interesting perspective, because I'm used to "elo" ratings showing one and one thing only: capability to win.

This means, rating goes up if you win, and goes down if you lose. Only round result and difference between your rating and server average matter (if you were, for example, below server's average, you would lose less on lose, and gain more on win).
But how you won? It doesn't matter. You can deal damage, steal secure kills, scout and flank, even just attract enemy fire and attention for as long as you can... as long as what you're doing helps you and your team, your win chances will increase, and with them your average rating gain. Meaning your rating - after it stops growing as you reach 50% winrate - literally shows your capability to win by any means.

But, looks like MWO's matchmaking is a different, and tries to take various stats (such as kills or damage dealt) into account. Problem, of course, happens when their current idea of how those stats affect the end result does not 100% correlate with how they really affect the end result in reality/current meta.
But, to be fair I'm not sure how well elo matchmaking would work with current playercount and with 12v12 being main gamemode. Maybe stats-based evaluation is what simply somewhat works for MWO?

[spartanspud]

For the most part, the match score is made up of how much damage you dealt.

[de.lorca]

going down with 700 damage and loosing rating probablie means it was a very close game with a lot of assault mechs

[Som31Else]

If only that were what I was experiencing, the 700 damage round was the enemy walking over us, think the end scoring was in the region of 12-2, I do try to support teamies so some games I do decent damage and earn no kills but usually >75% of total kills in assists.

Its clearly some arcane system whose technology and inner workings have long been lost, incantations and liberal sprinklings of holy mech water are the only things keeping its functions intact but hidden deeply behind gatekeepers and clouds of steam, 2 or 3 different takes on its working in this thread alone.

I guess it makes zero difference but I like progression and to not have it when expected is rather chafing.

Of course, appreciate the discourse on the matter. ;)

[isawaawasi5]

There are 24 players in a match. Top 11 highest match scores get Pilot Skill Rating up. Middle 2 are unchanged. Lowest 11 get PSR down.

[Zyrrashijn]

Here is an educative video that explains how PSR works.
https://youtu.be/JFqlZXOVtdw?si=evNWB_GhvhjVT8wc

[Fastspectre]

Originally posted by Shiro:

This is an interesting perspective, because I'm used to "elo" ratings showing one and one thing only: capability to win.

This means, rating goes up if you win, and goes down if you lose. Only round result and difference between your rating and server average matter (if you were, for example, below server's average, you would lose less on lose, and gain more on win).
But how you won? It doesn't matter. You can deal damage, steal secure kills, scout and flank, even just attract enemy fire and attention for as long as you can... as long as what you're doing helps you and your team, your win chances will increase, and with them your average rating gain. Meaning your rating - after it stops growing as you reach 50% winrate - literally shows your capability to win by any means.

But, looks like MWO's matchmaking is a different, and tries to take various stats (such as kills or damage dealt) into account. Problem, of course, happens when their current idea of how those stats affect the end result does not 100% correlate with how they really affect the end result in reality/current meta.
But, to be fair I'm not sure how well elo matchmaking would work with current playercount and with 12v12 being main gamemode. Maybe stats-based evaluation is what simply somewhat works for MWO?

elo only works for 1v1... it is really bad in every team enviroment to "RATE" a single player
so no team based game should ever use elo rating
if you wanna raise your psr all you need is an average of 300 or so matchscore and you will hit tier one at some point (so it is just a glorified xp bar if you know what you are doing)
if they did it right it would take all the helping things and KMDD more into account then just pure DMG, the would also NEED to tie it to every specific mech (the specific build) since meta builds do way better then less meta'y builds

[Shiro]

Originally posted by Fastspectre:

Originally posted by Shiro:

elo only works for 1v1... it is really bad in every team enviroment to "RATE" a single player
so no team based game should ever use elo rating

Utterly wrong. There's no reason whatsoever why elo wouldn't work in team environment to rate each individual player.

It's simple. If a given player wins his games repeatedly, then he clearly deserves a higher elo. And after he is given higher elo. in his games teams will be balanced such as that player will face competition of other players with high elo, or he will be balanced by placing low-elo teammates to his team while enemy's elo is all average... end result is the same, creating a balance between teams where that player's chance to win is as close to 50% as possible.

Now, it's possible that the player does better in some team compositions than the others. For a rough example, if some Assault Joe was an excellent shooter and tank but suffered from snipers, he would do well each time his team had lights or countersnipers to distract snipers, and die fruitless if snipers stood unopposed. However, remember that each team is made of twelve such Joes, and so the real win chance of thar team will fluctuate around the stated difference of teams' "average elo" values. If Joe is team-dependant but wins 50% of the time, then he will simply keep his current elo. Turning round outcome to a gamble of sorts perhaps, but still it will ensure team's win chance of ~50%... at least until Joe figures out how to evade snipers on his own, bumping his winrate and elo, at which point he will be given weaker teammates and will still lose 50% of the time but to different reasons.

P.s. there is one downside in that of you achieved your current level by pushing a certain playstyle only, changing that playstyle will weaken you and therefore make you lose for a while. This might necessitate an "offset" for each given playstyle. For example, Deadlock has an offset per each hero. However, for MWO identifying what "role" you have could be challenging beyond weight class split (either light/medium/heavy/assault, or a 20/100 ton offset that affects your final elo the more you move away from 60 tons).

[de.lorca]

Originally posted by Shiro:

Originally posted by Fastspectre:

elo only works for 1v1... it is really bad in every team enviroment to "RATE" a single player
so no team based game should ever use elo rating

Utterly wrong. There's no reason whatsoever why elo wouldn't work in team environment to rate each individual player.

yeah loosing games because the rest of the team can't carry there own weight doesn't change the fact players that are an asset to the team will win more games than those that aren't, it really isn't that complicated, people are just very bad at handeling the frustration of loosing and linger in those negative emotions to mutch to look at the bigger picture which is either that they had a great game in spite of loosing or they shud focus on there own failures instead of blaming them on others

[Shiro]

Originally posted by de.lorca:

people are just very bad at handeling the frustration of loosing and linger in those negative emotions to mutch to look at the bigger picture which is either that they had a great game in spite of loosing or they shud focus on there own failures instead of blaming them on others

Yes the attitude is the main problem, and will stay so regardless of how you rate players. Some will always blame the others instead of trying to git gud (if not by improving their skills and metamaxing their mech, then at least by getting some sleep and entering the game calm, rested and focused). But, that's the problem of those humans I guess... best way to avoid it is by raising above low rating "hell" which traps those who keep proving "noob" and "newbie" should have different definitions.

[Fastspectre]

Originally posted by Shiro:

Originally posted by Fastspectre:

elo only works for 1v1... it is really bad in every team enviroment to "RATE" a single player
so no team based game should ever use elo rating

Utterly wrong. There's no reason whatsoever why elo wouldn't work in team environment to rate each individual player.

because ELO would always rate the TEAM, yeah there might be one player on there that carries hard but since he has to carry that hard his rating wouldn't be correct
and the people that got carried..... their rating would be utterly overated/distorted

ELO is completly useless to rate single players in everything other then 1v1 cenarios

Arpad Emmerich Elo (né Élő Árpád Imre August 25, 1903 – November 5, 1992) was a Hungarian-American physics professor who created the Elo rating system for two-player games such as chess

two players 2
1v1
get it?

btw in relation to MWO multiplayer how do you ELO rate a guy on the losing team that did 1500 dmg when the highest one on the winning team did 600?
and the only thing that counts for ELO is win or lose
that's right the guy who did twice as good as the best player on the enemy team loses rating... because it's stupid to use ELO for team rating and even more to use it to rate single players in team games

ELO was invented for 1v1

[spartanspud]

It's not so black and white. If you had PSR (OSR?) in Siege for example, it would encourage poor play. With elo, your goal is to win. That works better in some games than others.