Jump to content

Psr Update And Hold On Patch.


723 replies to this topic

#41 Khalcruth

    Member

  • PipPipPipPipPipPipPip
  • Hero of Steiner
  • Hero of Steiner
  • 667 posts
  • LocationYou gotta lose your mind in Detroit Rock City!

Posted 08 June 2020 - 05:54 PM

View PostPaul Inouye, on 08 June 2020 - 02:27 PM, said:


In summary, the formulas need to be agreed upon by the community



How in the heck do you propose to measure that?

#42 Termin8rSmurf

    Member

  • PipPip
  • Wrath
  • Wrath
  • 40 posts
  • LocationNew Zealand

Posted 08 June 2020 - 05:54 PM

I have always maintained that a good light pilot, getting targeting information, target locks, critically valuable UAV deployments, and generally being a good sort, should be rewarded more. They may not do a lot of damage, but their assistance on obtaining the win is often vital, and even critical. Some wins have only been made possible by the light putting his shiny metal butt in the sling to get the information for his team. This new options for change will make the vital light pilots rewards better. Sure, a LRM-[Redacted] might get 1500 skill free damage, spraying his LRMs all over the place like an alley cat marking his territory, but they tiny metal sneaky boy who makes it possible for the tom cat to spray everywhere never got any recognition. Not, let the lights be the heroes, while the LRM-[Redacted] sprayed damage becomes less Match Score worthy than it's always been in the past.

In a recent match, I racked up two headshot kills, and three other kills besides and 5 assists, scoring 500 damage and survived the match intact. My friend, in an identical mech scored 1 kill with 10 assists, and died to a sneaky machine gun light parasite, scored 644 damage. His match score was higher than mine, despite my shots being more on target, my damage being more effective per point, and I survived while he did not. Surely there's a problem with the Match score only really taking damage into account. Whereas, accuracy, effectiveness, survival, etc should also count, and much more than it currently does!

#43 C337Skymaster

    Member

  • PipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 363 posts
  • LocationNew England

Posted 08 June 2020 - 05:56 PM

View PostHawker Siddeley, on 08 June 2020 - 05:41 PM, said:


Sorry I should have exampled my point better. How do you differentiate someone who ignores the call, or YOLOs in, wasting their mech but still taking as much damage as someone who actually tanks the push? How do you differentiate between someone using their armour effectively and someone who isn't?



As others have said, and perhaps put a little more simply: if their team goes with them, they live longer to deal more damage, spot more targets, and twist more damage. If they're alone, they take anywhere from 4 to 12 alphas to the CT and go down before their ping can catch up. It's a self-balancing issue.

#44 Kageru Ikazuchi

    Member

  • PipPipPipPipPipPipPipPip
  • The Determined
  • The Determined
  • 1,188 posts

Posted 08 June 2020 - 06:02 PM

Thanks. That should work just fine.

As a follow-up tweak, you could adjust the Tier brackets (behind the scenes, of course ... we don't need to know the boundaries) occasionally to ensure more or less even distribution among current active players.

#45 Anomalocaris

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • 454 posts

Posted 08 June 2020 - 06:02 PM

View PostMisterSomaru, on 08 June 2020 - 05:32 PM, said:

can we keep this as simple as possible? Like, seriously. My idea is that Top 8 players go up, middle 8 get nuthin', bottom 8 players go down, like, +4, +4, +3, +3, +2, +2, +1 +1, 0, 0, 0, 0, 0, 0, 0, 0, -1, -1, -2, -2, -3, -3, -4, -4 for example.
I just pulled those numbers out of my ***, but it equals zero sum, if my math isn't that rusty. Specifics could be determined later, but the point stands. People who do very well go up, people who are middling are where they belong. people consistently doing poorly go down to the tier they belong in. It's as simple as that.
KISS Keep It Simple Stupid
This is a pvp game, lets keep the metrics by which we change our own PSR simple, so nothing needs to be broken even harder.

Edit: this system is based around match score, and rewards players with uncarriable teams that work their asses off, as well as punishing people that win and do **** diddly.


I'm pleasantly surprised and I believe a plus/neutral/negative setup as Somaru proposed is the way as well. Whether its 10-4-10, 9-6-9 or 8-8-8 doesn't really matter so much to me. And staggering it at least a little would be better I think, but I'm not married to it. This system is simple, and works whether you change the matchscoring system or not (I don't think MS needs much tweaking, it will just change the meta, not who is good at it).

That said, the issue of groups is still unaddressed. Others have mentioned some solutions, but you must balance groups against solos to win back the people that are tired of stomps and unbalanced matches. Even fixing the ranking system like this will do no good if group balance is not solved.

#46 Navid A1

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2019 Silver Champ
  • CS 2019 Silver Champ
  • 4,561 posts
  • LocationVancouver, BC

Posted 08 June 2020 - 06:07 PM

Thanks for opening up the discussion Paul.

Would it be possible to provide a bit more info on matchscore parameters?... Most importantly the tick rate? as most of the parameters can be a continuous score tick.

#47 Tarl Cabot

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Tai-sho
  • Tai-sho
  • 6,985 posts
  • LocationImperial City, Luthien - Draconis Combine

Posted 08 June 2020 - 06:10 PM

Adding my spill into this.

Adding another PSR level and making the 201-300 the break even. The current PSR setting had the break even in the 101-250 MS range, thus we had players with an overall avg 171 MS 0.83 W/L ratio making it to Tier 2 and Tier 1 simply via brute force of playing 25K games.

Posted Image

The zero-sum Paul has now proposed seems interesting, and I will be taking note of my matches but I am, or my Alt, is closer to the 300 MS, but I have had several drops where if we used Paul's proposal, imagine what Tier 4 games would look like.. 200 MS top mark players moving up the same amount as a 401+ MS player, yes?

Edited by Tarl Cabot, 09 June 2020 - 02:26 PM.


#48 PhatRaptor

    Rookie

  • The Benefactor
  • The Benefactor
  • 1 posts

Posted 08 June 2020 - 06:14 PM

If match score is used won't everyone try to run higher damage mechs? Even if you get killed early you potentially get a high match score. Assaults will have an advantage in general. Just my thoughts.
'

#49 yrrot

    Member

  • PipPipPipPipPip
  • The Mercenary
  • The Mercenary
  • 174 posts

Posted 08 June 2020 - 06:14 PM

With zero sum, people that play in groups most often should end up rising to top-tier, which would help reduce their impact on match quality (since their opponents will have a be from the upper part of available players). If there isn't another group or top-tier solo players to balance them out, no matchmaker changes will fix it anyway.

Paul/Russ, to clarify, are these just match score values we can change, or can we propose a formula? In other words, can we only change the ratio between these values, or can we change it from a sum of the values to something more complicated? For example, making match score include functions like the minimum between damage score and assist score?

#50 Socal Bronco

    Member

  • PipPip
  • @ui_title_%s_Rank_4
  • @ui_title_%s_Rank_4
  • 23 posts

Posted 08 June 2020 - 06:15 PM

Thanks for including is in the brainstorming.

First, the point of the MM and tiers is to try and balance player skills on each team. For each highly skilled player on Team A, we want to put a highly skilled player on Team B. Its job is not to award people for playing the way we want them to play ( i.e. team players). With that said, assuming that we have 5 tiers, then I propose T1 and T5 each contain 5% of the player base, T2 and T4 would each contain 15% of the player base. The remaining 60% are T3.

Ratings would be a rolling average based on the 500 or so matches . This keeps the value close to real time without too much penalty or bonus for a single game. For an active player this is about a month, for most everyone else its probably 3 to 5 months of QP.

If you do this then you don't really have to redo the points system, do that sort once a day or once a week or what ever makes sense for MM. By default its 0 sum since the groups are all fixed % of the total players.

If you go this way you may want to consider some achievements or awards for players who have played a lot. Historically T1 was a badge for completing the grind. It would be nice if there was a replacement i.e. 5000 matches won,or something.

#51 thievingmagpi

    Member

  • PipPipPipPipPipPipPipPip
  • 1,376 posts

Posted 08 June 2020 - 06:19 PM

copypasted from another thread



I don't really feel like a zero-sum game is necessary. Reward good play, don't reward bad play. I don't think it should be relevant to the team etc. I don't think it feels very fair for a good (lucky) team that drops where all 12 players put in relatively equal effort (let's say avg MS of winning team is 300 with a 50 pt spread between #1 and #12) for there to be such a disparate result between top and bottom.

I'd definitely prefer some sort of MS based bracketed MS as in the OP, one with volatility and one where good players on bad teams don't get punished and bad players on good teams don't get carried.

If in fact that does lead to an ultimately upward sorting XP-bar style progression, then so what? We've had a brokeass PSR/MM for 8 years. Just reset it every 4 months/6months.




,

That said, in regards to MS triggers, I'm pretty much fine with how they are. They can be improved, yeah, but they can certainly be made worse. This game is about giant robots fighting each other. Focus on that. Deciding to bias the game towards cap rewards (MUH OBJECTIVE) will absolutely be a step in the wrong direction.

Increase KMDD/Solo etc MS rewards.

Decrease or reduce to zero the incidental MS rewards (flanking, light/med/heavy/assault protected, hit and run).

Perhaps slightly increase the cbill payout for said incidentals so the lore types can feel good.

Edited by thievingmagpi, 08 June 2020 - 06:19 PM.


#52 Kamikaze Viking

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 368 posts
  • LocationStay on Topic... STAY ON TOPIC!!!

Posted 08 June 2020 - 06:28 PM

To everyone talking about Match Score changes. YES it needs to be addressed next. But NOT AT THE SAME TIME.

Its a separate system. Treat it as such.
Yes it needs tweaking but if you do it at the same time as the PSR system, you would mess up the data that we need to confirm that the PSR system changes works.
ISOLATE the changes, else you cannot verify their success/failure as your measurements are also being affected by the 2nd (3rd, 4th etc) change.

STAY ON TOPIC.

#53 thievingmagpi

    Member

  • PipPipPipPipPipPipPipPip
  • 1,376 posts

Posted 08 June 2020 - 06:30 PM

View PostKamikaze Viking, on 08 June 2020 - 06:28 PM, said:

To everyone talking about Match Score changes. YES it needs to be addressed next. But NOT AT THE SAME TIME.

Its a separate system. Treat it as such.
Yes it needs tweaking but if you do it at the same time as the PSR system, you would mess up the data that we need to confirm that the PSR system changes works.
ISOLATE the changes, else you cannot verify their success/failure as your measurements are also being affected by the 2nd (3rd, 4th etc) change.

STAY ON TOPIC.


Agree

is this not the PSR thread also?? lol idk

#54 MisterSomaru

    Member

  • PipPipPipPipPip
  • The Death Wish
  • The Death Wish
  • 198 posts

Posted 08 June 2020 - 06:31 PM

View PostAnomalocaris, on 08 June 2020 - 06:02 PM, said:


I'm pleasantly surprised and I believe a plus/neutral/negative setup as Somaru proposed is the way as well. Whether its 10-4-10, 9-6-9 or 8-8-8 doesn't really matter so much to me. And staggering it at least a little would be better I think, but I'm not married to it. This system is simple, and works whether you change the matchscoring system or not (I don't think MS needs much tweaking, it will just change the meta, not who is good at it).

That said, the issue of groups is still unaddressed. Others have mentioned some solutions, but you must balance groups against solos to win back the people that are tired of stomps and unbalanced matches. Even fixing the ranking system like this will do no good if group balance is not solved.

eh, like another said, those that group up will more likely rise to the top, and be a non issue for those in the lower end.

#55 NumberFive

    Member

  • Pip
  • 11 posts

Posted 08 June 2020 - 06:32 PM

The most important change to ensure the long term player satisfaction is the use of a rolling average PSR rating rather than any absolute accumulated score.

This is the only feasible way to:
* prevent the game from becoming boring due to an over-reliance on 'the meta'
* allow new mechs to be levelled
* remove the impact of those games where your team just lets you down and you get crushed (e.g. 'being the assault left behind in a nascar')
* smooth out the outlier games (like when 'Assault' is won by capping before anyone really gets real fighting done) without needing to add complex changes to calculations

#56 Anomalocaris

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • 454 posts

Posted 08 June 2020 - 06:35 PM

View PostMisterSomaru, on 08 June 2020 - 06:31 PM, said:

eh, like another said, those that group up will more likely rise to the top, and be a non issue for those in the lower end.


The PSR of the group must be compensated for. If four 99%ers drop in a group on one side, then if the matchmaker can't find an equivalent group to drop against them, it needs to put the first 5 top players in the queue opposite that group.

Right now I don't think the MM considers anything of the sort when constructing a match, which is why a top group can end up with two high level players on their team in addition to themselves but the other team has a bottom feeder group plus potatoes.

Edited by Anomalocaris, 08 June 2020 - 06:38 PM.


#57 nocomfort

    Member

  • PipPipPipPipPipPip
  • 228 posts

Posted 08 June 2020 - 06:37 PM

This is a bait and switch. We asked for change, they're giving us a spreadsheet and telling us that everyone has to agree on the numbers before they'll do anything. This is a smokescreen for them to do nothing while saying they tried their best. I am not keen on that happening for like a fifth time.

How about this: Reset PSR immediately using the last 25/50/100 games as a player's seed, and let us play in the new tiers that result with the current system while we're waiting for the twelfth of never to freeze over and everyone to agree where the new deck chairs are going to go. Give us something right now so we can at least be satisified that maybe something we propose will be taken seriously.

Also the current match-making is not fun at all. Like really not fun. Like, I am to the point where I would almost rather not play at all than play some of the games I've been in lately. I don't think changing the match score calculation is going to do anything about that.

#58 MisterSomaru

    Member

  • PipPipPipPipPip
  • The Death Wish
  • The Death Wish
  • 198 posts

Posted 08 June 2020 - 06:37 PM

View PostAnomalocaris, on 08 June 2020 - 06:35 PM, said:


The PSR of the group must be compensated for. If four 99%ers drop in a group on one side, then if the matchmaker can't find an equivalent group to drop against them, it needs to put the first 5 top players in the queue opposite that group.

Right now I don't think the MM considers anything of the sort when constructing a match, which is why a top group can end up with two high level players on their team and potatoes on the other side.

Alright then, groups should be sorted in MM based on the highest tier in that group.
Most groups of friends in the game are of similar skill level anyway.

Edited by MisterSomaru, 08 June 2020 - 06:38 PM.


#59 Anomalocaris

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • 454 posts

Posted 08 June 2020 - 06:43 PM

View PostMisterSomaru, on 08 June 2020 - 06:37 PM, said:

Alright then, groups should be sorted in MM based on the highest tier in that group.
Most groups of friends in the game are of similar skill level anyway.


Agreed, but you have to look at the interaction of groups and solos in the match. A lot of times there aren't 2 high level groups available (or they're not in the queue at the same time), in which case you need to seed the opposing team with enough high level solos to level the playing field against the group.

I'm not worried about mixed groups bringing in a newbie and being ranked lower for the drop so much as I am 4 top level players dropping in complementary mechs and devastating the opposing team. They'll still have an advantage against 4 high level solos (better communication, more likely to have sync'd up builds) dropping opposite them if there isn't an equal level group available, but at least the odds will be better for constructing a competitive match.

I like the suggestion of schoolyard selection process (A-B-A-B-A), but with groups you have to break that up front to equalize against the group before dividing the remaining players.

Edited by Anomalocaris, 08 June 2020 - 06:44 PM.


#60 yrrot

    Member

  • PipPipPipPipPip
  • The Mercenary
  • The Mercenary
  • 174 posts

Posted 08 June 2020 - 06:43 PM

View PostMisterSomaru, on 08 June 2020 - 06:37 PM, said:

Alright then, groups should be sorted in MM based on the highest tier in that group.
Most groups of friends in the game are of similar skill level anyway.


His point was that since MM is going to always take a team vs another team, you can get the best team in the game against a group of friends in goofy builds. This works out in the end, because the top team should end up in a higher tier in the long run anyway, which reduces the chances that they'll get matched with low-tier teams. It can still happen, but it will trend more towards "rare" with enough player population.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users