106 replies to this topic

[Kubernetes]

If they switched it to look only at personal performance, that would seriously handicap lights. Lights will never do as much damage, but you're kidding if you think they can't make a huge contribution to a team's win. I used to run a Narc Raven 3L... the thing rarely ever broke 200 damage but its WLR is >2. That wasn't random. Giving your team a complete picture of the enemy position contributes hugely to wins. If a person's contribution to victory is random, you'd expect most people to hover right around the 1.0 mark. But good pilots are normally way above 1.0. That's not random, that's doing things that consistently contribute to victory.

[ExplicitContent]

If someone renamed themselves to 'PSR System' and then reported this thread, what would happen? I could see someone having fun with that, since 'PSR System' is the commonly used term and normally spoken of disparagingly.

[wanderer]

Fixing PSR would be relatively simple.

You increase the match rating needed to go + or = as your tier ranking goes higher, and you also increase the penalty for low match ratings as you do as well.

It should take better performance as you go past rank 4 to even hold your head above water, and T1s should be constantly hoping for exceptional performances to maintain their ranking, leaving them few and far between. As it stands, tiering is literally nothing but an exp bar and does little to give players opponents they're competitive against.

[Reptilizer]

Rizzwind, on 26 March 2016 - 05:22 AM, said:

I normally try stay out of the Forum PR BS but I have to agree with this. It's a random team based rank system and has nothing to do with your lone player skill. I have tested this even if you use 1 sl once and your team wins your rank goes up.

No. If i you literally only used your LS once, your match score probably will be well under 100. You will not move in rank then.

[KHETTI]

PSR doesn't work very well currently, simply due to laser vomit + overquirked crutch mechs.
Want PSR to actually mean something? then PGI need to remove the skill tree and all quirks and go back to base values and balance from there.
No more eze mode, one weapon group, point and click insta kill mechs will help players settle into their true skill bracket.

Its pretty simple, balance is sh*t, therefore PSR is sh*t.

[Solarise03]

Its way more fun and relaxing in lower tiers then tier 1 XD

[adamts01]

Kubernetes, on 29 March 2016 - 06:00 PM, said:

If they switched it to look only at personal performance, that would seriously handicap lights. Lights will never do as much damage, but you're kidding if you think they can't make a huge contribution to a team's win. I used to run a Narc Raven 3L... the thing rarely ever broke 200 damage but its WLR is >2. That wasn't random. Giving your team a complete picture of the enemy position contributes hugely to wins. If a person's contribution to victory is random, you'd expect most people to hover right around the 1.0 mark. But good pilots are normally way above 1.0. That's not random, that's doing things that consistently contribute to victory.

Nailed it. There's no computer program that can account for half the things that make a player useful. W/L tells all. sure there are games where you did everything you could do but your team let you down, but it all averages out. If it was all about personal numbers, you'd see much more yolo rambos. So what if you get 5 kills and top damage if they were all gimped squirrels that led you away from the front line and priority targets? Meanwhile, the sub 100 light might have spotted, called weak components, observed all enemy movement, ran back to stall a base cap...... Numbers don't matter.

KHETTI, on 30 March 2016 - 02:35 AM, said:

PSR doesn't work very well currently, simply due to laser vomit + overquirked crutch mechs.
Want PSR to actually mean something? then PGI need to remove the skill tree and all quirks and go back to base values and balance from there.
No more eze mode, one weapon group, point and click insta kill mechs will help players settle into their true skill bracket.

Its pretty simple, balance is sh*t, therefore PSR is sh*t.

I agree with ditching the skill tree. I want everyone to be on equal footing. No magic speed boost or cooling because grindy reasons.
About your other argument though, it's not "eze mode" if your competition is running comparable equipment. You can limit the game to stock mechs and you'd still see the same thing, people picking the creme of the crop. It's the nature of a competitive environment. As for "1 click insta-kill mechs," how is doing well in that not skill when half your targets can insta-kill you right back? Yeah it takes more skill to do well in a Mist Lynx than a Cheetah, it also takes even more skill to do well in Commando with no armor, and yet more skill a War Hawk with no armor. You can play that game all day.

[Saltychipmunk]

Is there a good way to lower psr without purposefully losing? high level play is pretty boring and i miss the gloriously silly low tier matches where you get some good old fashion noob brawls (god i love those noob brawls, you know the ones where both you and your enemy make clearly stupid builds and use them in clearly stupid ways to have little laser cuddles at 200 meters... awesome)

all these mid teir matches are clan stacks vs clan stacks vs clan stacks vs clan stacks vs clan stacks vs me vs some other unlucky dude

Edited by Saltychipmunk, 30 March 2016 - 03:38 AM.

[Podex]

Meh, it's a non issue for me. I still feel like it's the same old match maker and PSR is just some visual indication of where I am in the greater scheme of things. For personal standing, there's always KDR and the rest of your stats, which nobody really cares about. It's the same for PSR, except on a team level.

[PyckenZot]

Wintersdark, on 26 March 2016 - 05:58 AM, said:

What the good Mr. Winter wrote.

What's important to keep in mind is that PSR isn't really a good personal rating system match to match. You may as well just use match score for that (though it's not very good either, but is just the best of limited tools).

In the OP's case, erroneous results happen - up (a little bit) on a poor performance with a win, or down (a little bit) on a decent performance with a loss, but keep in mind these are small changes to rating that don't move the bar much. This is to account for times where you actually did contribute to a win but didn't do much raw damage (say you're a great scout/spotter/leader/whatever) or that you're actually contributing towards losses despite getting mediocre scores (poor team play).

After all, it doesn't take a stellar performance in a loss to prevent rating loss entirely, and a decent performance in a win causes your rating to increase much more than a poor performance does.

Finally, keep in mind the One True Purpose of PSR and ANY matchmaking/rating system PGI uses: the protection of new players from experienced/veteran players. PSR is positively weighted for this reason - you'll tend to always progress upwards overall unless you're really dragging your team down, but slowly (very slowly at higher tiers). This serves to move players away from the kiddie pool as quickly as possible, so the more experienced you get the less you'll encounter new players.

This says it all.
Well done sir!

[adamts01]

Saltychipmunk, on 30 March 2016 - 03:37 AM, said:

Is there a good way to lower psr without purposefully losing?

Nope. I ran no armor LRM boats for about 40 hours and managed to drop my XP bar to just a hair under maxed. I don't think PSR is really that bad, it's the way tiers are mixed that's to blame. I say there should be 3 tiers and everyone plays against their own tier. That way middle of the road guys don't get grouped against the comp guys or the new players. We have basically 3 matches anyway: Tiers 5-3, 4-2 and 3-1.

[Wintersdark]

adamts01, on 30 March 2016 - 04:12 AM, said:

Nope. I ran no armor LRM boats for about 40 hours and managed to drop my XP bar to just a hair under maxed. I don't think PSR is really that bad, it's the way tiers are mixed that's to blame. I say there should be 3 tiers and everyone plays against their own tier. That way middle of the road guys don't get grouped against the comp guys or the new players. We have basically 3 matches anyway: Tiers 5-3, 4-2 and 3-1.

You understand that tiers are only mixed when there are not enough players at a given tier to make a match, right? They're not just mixed because "haha lol!"

So, no, they can't just make single tier only matches all the time. It's not an option. Sometimes, sure, matches are predominantly single tier and occasionally +/-1 (busier times, and the odd match when everyone else is in game).

Oh. I see you 3 tiers thing. That's fundamentally no different than stretching to +/-1, really. And the MM is only going +/-2 tiers after 2 minutes trying to build the match.

The MM still picks people based on rating (Russ has discussed the absolute rating difference in matches) rather than solely via tier, so it attempts to get similarly rated people within a tier as well.

[adamts01]

Wintersdark, on 30 March 2016 - 05:14 AM, said:

You understand that tiers are only mixed when there are not enough players at a given tier to make a match, right? They're not just mixed because "haha lol!"

So, no, they can't just make single tier only matches all the time. It's not an option. Sometimes, sure, matches are predominantly single tier and occasionally +/-1 (busier times, and the odd match when everyone else is in game).

Oh. I see you 3 tiers thing. That's fundamentally no different than stretching to +/-1, really. And the MM is only going +/-2 tiers after 2 minutes trying to build the match.

The MM still picks people based on rating (Russ has discussed the absolute rating difference in matches) rather than solely via tier, so it attempts to get similarly rated people within a tier as well.

That 2 minute wait is for any single player. A single tier 1 waiting 2 minutes could have 23 tier 3s insta-dropped in to his match, a very dramatic example but absolutely possible. I know for a fact that I've been insta-dropped with tier 3s very frequently. I'm fully aware we don't have the players to only get matched in our own tier with 5 tiers, but I think 3 larger tiers would be a very good compromise. I think it would be great for everyone. Take the average tier 3 for instance, probably not using a meta build, has played a couple hours a week for a year. He shouldn't be in a match with comp players or players just getting out of the tutorial, just like a fresh Steam player should be matched against a 1 year vet. I'm a little more pissy about MM than most because of the very small player base on my half of the world, and NA primetime is considerably better, but I still think the system could be dialed in a little.

[Dar1ng One]

The only Tiers PGI cares about, are the tears streaming down your cheeks.

Like for instance, if your a really good, selfish, independent player, then this tier system is not for you. It rewards positive society rainbows and unicorn team based synergy play.

And lets not get started on the whole, I will carry you little noob says the team, and grant you Death Star, and tier 1, not because your any good, but because your likable, pander to my e peen ego, and suck my **** dry with your pretty lips.

And then you get the 2 x account players that load their prime account, and then the old stationary account via wireless dongle laptop account in team drops and CW.

My advice, play CW, get slammed, grind your mechs, and sleep in the knowledge that CW stats, wins or loses will not affect your tier or KDR whatsoever.

[Podex]

adamts01, on 30 March 2016 - 05:30 AM, said:

That 2 minute wait is for any single player. A single tier 1 waiting 2 minutes could have 23 tier 3s insta-dropped in to his match, a very dramatic example but absolutely possible. I know for a fact that I've been insta-dropped with tier 3s very frequently. I'm fully aware we don't have the players to only get matched in our own tier with 5 tiers, but I think 3 larger tiers would be a very good compromise. I think it would be great for everyone. Take the average tier 3 for instance, probably not using a meta build, has played a couple hours a week for a year. He shouldn't be in a match with comp players or players just getting out of the tutorial, just like a fresh Steam player should be matched against a 1 year vet. I'm a little more pissy about MM than most because of the very small player base on my half of the world, and NA primetime is considerably better, but I still think the system could be dialed in a little.

Again, I still don't see much difference between PSR and the old MM. I still have winning streaks followed by a streak of total annihilation because of mismatched teams. While I'm not in a unit myself, I can often tell how the match is going to play out by the ratio of unit tags on both sides, sometimes seeing around 2:10 differences when I'm doing too well. In these matches you'll see my team chasing a spider around the map while getting picked off one by one. My theory is that nothing has changed on the backend, but PGI was trying to placate the people crying that they wanted to see what MM bracket they were in.

[adamts01]

Podex, on 30 March 2016 - 05:41 AM, said:

Again, I still don't see much difference between PSR and the old MM. I still have winning streaks followed by a streak of total annihilation because of mismatched teams. While I'm not in a unit myself, I can often tell how the match is going to play out by the ratio of unit tags on both sides, sometimes seeing around 2:10 differences when I'm doing too well. In these matches you'll see my team chasing a spider around the map while getting picked off one by one. My theory is that nothing has changed on the backend, but PGI was trying to placate the people crying that they wanted to see what MM bracket they were in.

That's funny and quite possible that things never changed, now it's sad. I honestly wasn't that good back in the elo days so I can't comment much on that. I started playing again after this PSR thing after a year long break and I've noticed games gradually getting worse as more people squeeze though the cracks in to a tier they don't belong in. I do feel you on the unit tag thing. It's not at all foolproof but most people who go out of their way to join a unit have some thought about being part of a team. On the other hand, I consider myself a very good team player and I've been in a unit for about 3 months since beta. Idk, PGI sucks.

[Black Ivan]

PSR is simply a skill bar nothing more. It does not judge the skill of the pilot in a meaningfull way. Match Making could be so much better if PGI would have made a better ELO or BV value system.

[Podex]

adamts01, on 30 March 2016 - 05:50 AM, said:

That's funny and quite possible that things never changed, now it's sad. I honestly wasn't that good back in the elo days so I can't comment much on that. I started playing again after this PSR thing after a year long break and I've noticed games gradually getting worse as more people squeeze though the cracks in to a tier they don't belong in. I do feel you on the unit tag thing. It's not at all foolproof but most people who go out of their way to join a unit have some thought about being part of a team. On the other hand, I consider myself a very good team player and I've been in a unit for about 3 months since beta. Idk, PGI sucks.

I think the MM does and has always worked toward keeping people at thresholds, in may case it's a 1:1 KDR and 1:1 WLR. It just seems like it always wants me floating around these ratios. I could be totally wrong, and it's just some subliminal handicap that I give myself in game, but I have statistically been in the same spot for over a year and a half. It's fair.

Edited by Podex, 30 March 2016 - 06:16 AM.

[Lugh]

Triordinant, on 26 March 2016 - 06:44 AM, said:

Maybe they should have separate PSR for the group queue and solo queue.

Too hard. Remember MATH is PGI's downfall always. And they aren't especially good at IF THEN ELSE statements either.

A simple
IF Pilot Lugh = Grouped THEN use PSR rating Gx(variable for group PSR)
Else use Pilot Lugh = Solo [not grouped] THEN use PSR Rating Sx(variable for Solo PSR).

Would likely be well beyond their capabilities.

Edited by Lugh, 30 March 2016 - 06:32 AM.

[Remillard]

I don't have much of a gripe with PSR on the whole. Some method is required to keep folks playing in brackets that are challenging (relatively speaking) to them while providing some minimal buffer to keep new folks from being clubbed by more experienced folks.

However OP's post did sort of spark an idea. PSR does have a damage done component. For the portion of match score that is based on damage, would it make sense to scale that based on the mech's actual damage capability in some rough measure? If I do 100 points of damage in a 5 LPL Wubshee versus doing 100 points of damage in a light with a couple of MLs, that does show a different amount of work.