51 replies to this topic

[Yellonet]

No, not a good idea. PSR should change though.

[NoiseCrypt]

Yellonet, on 23 July 2016 - 04:31 AM, said:

No, not a good idea. PSR should change though.

Care to elaborate ?

[crashlogic]

leaderboards are PGI's dumbest idea yet.

[NoiseCrypt]

crashlogic, on 23 July 2016 - 06:43 AM, said:

leaderboards are PGI's dumbest idea yet.

Care to elaborate ?

[Champion of Khorne Lord of Blood]

OP fails to realize just how horribly flawed the leaderboards are.

Most of the top rated people only have even had a single match. Other ratings are based entirely on just farming matches rather than how well you did in them. In general the leaderboards don't tell you anything except that they suck as a metric for anything meaningful.

At the moment to get something similar to what you want in effect but with a system that works at least somewhat, you could take the PSR number rather than the tier and have people match up against the closest possible players to their PSR number each match. You could also then make bad performance in matches make you go down further and good performances a higher requirement, essentially making PSR on a win similar to what it would do on a loss, but slightly better. Say 250 score to go up, under 150 you go down, over 400 you go up greatly, 150-250 stay the same. This makes people have to work and be good to move up and have a chance of falling if they do bad.

Simply I'd say we just fix PSR by redoing the rating system in a way like this, setting players to start in T3, resetting tiers, then they move up or down from there through the tiers. Also make PSR tiers more volatile so that if you are doing really bad you'll go down decently fast and if you do really good you go up pretty quick. People may smurf in a situation like this, but I see it as better to have a system that can properly setup matches based on player skill than to have one that allows chronicbads into tier 1.

[KHETTI]

So we all know those currently at the top of the leaderboards are legitimately the best players, way better than the top comp players, and no-one is exploiting the sh!te out of a flawed system in a lame attempt to look l33t.

I say go for it, just another reason to encourage good players to go play something else.

[EvilCow]

(W/L + K/D) * AvgScore (last 60 days of stats only)

Then map:

Top 10% in T1
2nd 15% in T2
3rd 20% in T3
4th 25% in T4
Bottom 30% in T5

Games would be formed from +-1 tiers like now.

Edited by EvilCow, 23 July 2016 - 11:17 AM.

[Mystere]

NoiseCrypt, on 23 July 2016 - 08:21 AM, said:

Care to elaborate ?

Mixed solo and group queue stats?

[NoiseCrypt]

Dakota1000, on 23 July 2016 - 10:18 AM, said:

OP fails to realize just how horribly flawed the leaderboards are.

Most of the top rated people only have even had a single match. Other ratings are based entirely on just farming matches rather than how well you did in them. In general the leaderboards don't tell you anything except that they suck as a metric for anything meaningful.

At the moment to get something similar to what you want in effect but with a system that works at least somewhat, you could take the PSR number rather than the tier and have people match up against the closest possible players to their PSR number each match. You could also then make bad performance in matches make you go down further and good performances a higher requirement, essentially making PSR on a win similar to what it would do on a loss, but slightly better. Say 250 score to go up, under 150 you go down, over 400 you go up greatly, 150-250 stay the same. This makes people have to work and be good to move up and have a chance of falling if they do bad.

Simply I'd say we just fix PSR by redoing the rating system in a way like this, setting players to start in T3, resetting tiers, then they move up or down from there through the tiers. Also make PSR tiers more volatile so that if you are doing really bad you'll go down decently fast and if you do really good you go up pretty quick. People may smurf in a situation like this, but I see it as better to have a system that can properly setup matches based on player skill than to have one that allows chronicbads into tier 1.

Why would you post that wall of text witouth actually reading the entire thread first ?

And there already is another thread regarding fixing the current PSR system, if thats your thing. I totally agree with zero-sum PSR.

[NoiseCrypt]

Mystere, on 23 July 2016 - 01:39 PM, said:

Mixed solo and group queue stats?

I cant se how one of the queues would be favorable over the other, wih regards to individual player standing.

[NoiseCrypt]

KHETTI, on 23 July 2016 - 10:32 AM, said:

So we all know those currently at the top of the leaderboards are legitimately the best players, way better than the top comp players, and no-one is exploiting the sh!te out of a flawed system in a lame attempt to look l33t.

I say go for it, just another reason to encourage good players to go play something else.

(Double post)
The main reason that the leaderboard is useless right now is that teams and matchups arent based on skill. Is it that hard to read a thread before posting ?

[Champion of Khorne Lord of Blood]

NoiseCrypt, on 23 July 2016 - 04:22 PM, said:

Why would you post that wall of text witouth actually reading the entire thread first ?

And there already is another thread regarding fixing the current PSR system, if thats your thing. I totally agree with zero-sum PSR.

I did read through the whole thread, infact I posted a bit of a wall of text because in all the posts before me you ask them to please elaborate, so I complied.

Otherwise I'd have just outright said "you're idea is bad and you should feel bad" and be done with it. Instead I posted why your idea was bad and gave a constructive counter idea.

A lot of your posts give me the feeling you have no idea what you are talking about when it comes to the leaderboard or knowledge of queues.

[Lukoi Banacek]

Dakota1000, on 23 July 2016 - 10:18 AM, said:

Most of the top rated people only have even had a single match. Other ratings are based entirely on just farming matches rather than how well you did in them. In general the leaderboards don't tell you anything except that they suck as a metric for anything meaningful.

As others have noted in another thread about this, introducing a minimum # of matches before consideration (100, 150? who knows) would get rid of the outliers you're referring to here.

There is not a single metric immune to folks gaming/farming if they put their mind to it.

But the more metrics available, the more granularity, the better the match making can be.

Right now it's purely Wins coupled with matchscore. Groups can "farm" wins against smaller less focused groups. Matchscore due to it's reliance on damage can be farmed to a degree as well.

But more metrics can be considered in MM and the leaderboard is a decent first step in that direction. I'm not saying using them "as is" is the best idea, but it's certainly no worse than the current PSR system, and frankly, likely to be much better.

Perhaps players play their first 25 matches in a pure "newbie" category, followed by a weighted MM system for another # of matches, before their granular stats are compared in some MM formula that puts them into general population broken up by these five tiers PGI doesn't seem to want to get away from.

Or perhaps once you get into gen-pop, you get a score that goes up and down and the matchmaker uses this score or Pilot value, coupled with some Battle value based on Mech demand/usage to further refine the data before MM'ing folks.

Ultimately, all I'm saying is more information, more metrics are better. And the more important ones exist, the harder it is for folks to "farm" them to appear better than their real skill level.

I think all of the following are potential valuable metrics the Board could display and use to influence some next-gen MM:

KDR
WLR
KMDD per match
Solo kills per match
Damage per match
Kills per match
Deaths per match
Average Match score
Team damage
Damage taken
Infotech points/targeting points
Components destroyed

And the list I'm sure goes on.

The leaderboard isn't "flawed" at all. It merely displays factual data. If you honestly think anyone considers someone the best player in the game because they're #1 on the board with one match, then that's an issue of interpreting the data, which one could argue the interpretation of is flawed. The data itself, simply is.

If you average a matchscore of 300 over the course of 3 matches....that might just be a good run eh? More impressive would be someone who is averaging that matchscore over hundreds of matches and frankly is better indicator of skill overall. Things tend to smooth out over time with a high enough # of reps.

Yes, we all play **** mechs sometimes to level them or experiment or whatever, and the tryhards tend to play the hardcore stuff they wanna play all of the time, but the reality is, once you hit a relatively good sized sampling of matches, your "skill" is pretty well displayed by the leaderboard. Just ignore the 10 match outliers and look at those with a decent sample and then ask yourself who you've seen ingame. I bet most of the better known, better players just so happen to have higher KDR, WLR and AMS than most do. You simply cannot "farm" the system for hundreds of matches unless you have some modicum of skill to go with it.

[PitchBlackYeti]

Just dump the whole pointless grind-based tier system and reintroduce a fully random MM, at least for solo queue. Not much will change, sometimes you will face a pro and get stomped and sometimes you will be the one that does the stomping. Statistically it will all average itself out. Games like War Thunder, World of Tanks never had a skill-based mm and it works just fine.

[EvilCow]

PitchBlackYeti, on 24 July 2016 - 11:58 PM, said:

Just dump the whole pointless grind-based tier system and reintroduce a fully random MM, at least for solo queue. Not much will change, sometimes you will face a pro and get stomped and sometimes you will be the one that does the stomping. Statistically it will all average itself out. Games like War Thunder, World of Tanks never had a skill-based mm and it works just fine.

I have problems with both being stomped and do the stomping. I don't want good games in average, I want good games each single time.

Edited by EvilCow, 25 July 2016 - 12:22 AM.

[Satan n stuff]

adamts01, on 22 July 2016 - 06:10 AM, said:

It just depends on what stats you use. Use W/L and things will sort themselves out nicely, eventually.

It would if we didn't have the tier system of before that PSR. You can have an amazing W/L in tier 5, that doesn't mean you're actually good.

[NoiseCrypt]

NoiseCrypt, on 23 July 2016 - 03:37 AM, said:

• Right now the leader-board doesn't actually make sense, since it's based people being matched up based on tier rather than skill.
  
• With ELO, the leader-board could actually show players real skill level/standing. If it was based on the right stats, and the MM could ensure that earlier teammates would end up as opponents later matches.

Dakota1000, on 23 July 2016 - 10:18 AM, said:

OP fails to realize just how horribly flawed the leaderboards are.

Dakota1000, on 24 July 2016 - 06:25 PM, said:

I did read through the whole thread,...A lot of your posts give me the feeling you have no idea what you are talking about when it comes to the leaderboard or knowledge of queues.

My entire point with this thread is that the leader-boards might actually add some value to this game if the current "rating system" was updated/changed.
But of course its much easier to just use personal attacks, when disagreeing with something.

[Angel of Annihilation]

There is a fatal flaw in using the leaderboards for MM purposes. The leaderboard scoring doesn't mean your the best player in the game, rather it means that your the best player in your Tier.

Keep in mind that all the players are already broken up into tiers for MM purposes so all I have to do at Tier 3, is to outperform all the other players how are also at Tier 3 to get a monster leaderboard score. Hell you could theoretically have a bottom Tier player win a leaderboard challenge just because is just good enough to dominate the lowest tier but not good enough to advance a tier.

Also keep in mind that the higher you go in tiers, the more and more skilled players become and as players become more skilled, the ability to stand out gets harder and harder. I wouldn't be surprised if very few of the top leaderboard players are actually playing at Tier 1.

[Clownwarlord]

To be honest I have not even looked at the leader-board. As for me I have just over a 1 for KD and Win and that kinda all that matters to me. I look at my match end results. Did I make 400 damage? Did I get a kill or 2? Did I get 5 or more Kill Assists? Did I destroy 4 or more components? Those are the stats I look at together more so than Kill Most Damage Done or Kills or Damage separately. The reason why is because it is tough to get 5 KMDD in a light (not impossible but tough). When it comes to getting 1k damage but not getting many kills just points out an inability to focusing your own fire. If you just look at kills well a light mech can just run through a scrum and zap 3 mechs in a row that they hadn't previously done any damage to. If you want Kill Assist put on a ERLL and swipe it like a credit card all over the match and if they get killed boom kill assists done by everyone else doing the work.

Yeah kinda crappy to just look at one but KMDD is at least a combo of kills and damage done. I just wish there was more to it like a mathematical equation to show who is best that brings in more stats than just those.

[kf envy]

NoiseCrypt, on 22 July 2016 - 04:42 AM, said:

(drops mic and walks out...)

just