Elo Ratings And Granularity
#21
Posted 11 December 2013 - 05:02 PM
#22
Posted 11 December 2013 - 05:07 PM
Then I remembered sample size.
I do think being in or out of a team matters a lot, but having one Elo per chassis is probably the limit …
Just four Elos' is, and continues to be, silly.
Sandpit, on 11 December 2013 - 04:52 PM, said:
In the absents of voice-comms, major leg damage, and limiting Target Decay to LRMs? Too damn long.
#23
Posted 11 December 2013 - 06:54 PM
While I think you've given a wonderful example of a mechanic that would improve gameplay I'd guess it's unlikely to be implemented anytime soon if at all.
#24
Posted 11 December 2013 - 07:36 PM
Duncan Aravain, on 11 December 2013 - 03:45 PM, said:
Or even if they cared why would you expect them to implement this when they can't even get the UI right and the effort to try to get the UI right is their top priority now. After that it will possibly CW if that ever is implemented. They can't even do the things they have already planned with any alacrity and you expect them to do something that isn't even on their planning boards. It's boggles the mind that people who appear intelligent don't get that these folks can't even do what they have stated they plan to do in a reasonable time or manner and yet they suggest new things.
#25
Posted 11 December 2013 - 08:27 PM
John MatriX82, on 11 December 2013 - 03:33 PM, said:
Since the chassis is "fresh" your ELO shouldn't be the one you have within the same class.
That's why it's a flexible system. When you have enough matches, it shifts to variant-by-variant; otherwise, it's using your chassis score. And if that doesn't have enough matches, it's using the weight class' average like now.
After thinking about this, you're totally right about the starting value. Even reading my example where the 733P shows up and has to tumble down the ladder in a brutal manner doesn't sound great. The easiest way to fix that is to shave points off whatever Elo you were going to get to compensate. So, if you were going to get a 1300 Elo, you're dropped to 1200; if you were going to get a 2400, you're dropped to 1600; a starting Elo of below 800 would be unaffected.
Arcturious, on 11 December 2013 - 03:59 PM, said:
I'd like to see it, but that would be quite a bit more difficult. I've thought about how to do it, but no matter what it's odd.
Do you class things based on range, giving each 'mech a 1-10 rating for short, medium, and long range capabilities? But what about The HGN-733C vs the HGN-733. They'd both say exceptional long range, when really one is a sniper / direct fire god that's effective at any range, while the 733 is an LRM boat.
Do you make up arbitrary roles and assign based on that? Brawler, sniper, bracket, etc.? But then that leaves out capabilities like the Stalker / Jager's peeking and jumpjets.
It's just messy when it comes to loadout. I'd like to see something, but I'm alright with some random chaos as long as it's not such a wide gap in effectiveness.
Tonnage limits will solve most problems anyways.
Colonel Pada Vinson, on 11 December 2013 - 04:47 PM, said:
Easier just to fix the game balance issues certain mechs are causing.
First of all, what the ****? Did you even read it? It's just Elo with a more granular sorting capability - not BV. I'm all for tonnage limits and ways to encourage people to get the hell out of the assault 'mechs, but counting on perfect balance to solve imbalances between variants and chassis is a stretch of the imagination.
There will probably always be something that is the flavor of the month. At the very least, it'll be that way for a long time, particularly when Clan weapons arrive and cause a whole new set of problems. Things will ideally all be on a competitive field together, but that doesn't mean a more granular way of classing your performance wouldn't help.
Some 'mechs just don't agree with you, while others feel like home. Sometimes you're just dicking around with machine guns and flamers. It's unrealistic to suggest there will ever be anything close to perfect balance, where every 'mech and weapon are on equal footing.
Hell, the idea already works: imagine if you only had a single Elo instead of one for each weight class. Expecting a 733C performance from a Locust? Lol no. All I'm suggesting is a flexible extension of what they're already doing.
Sandpit, on 11 December 2013 - 04:52 PM, said:
Since the Highlander.
Ghogiel, on 11 December 2013 - 05:02 PM, said:
The Stalker is an exceptional chassis in that it doesn't need jumpjets to perform incredibly well. First, the trait that all Stalkers share: super-high mounts. It can win any peek war. It can even often go toe-to-toe with poptarts. Where other 'mechs are shooting the dirt 30m ahead, the Stalker can command a ridge with relatively small risk. Additionally, the Stalker's side torsos almost always take damage before the center torso is cored, and it makes them more durable than almost any other 'mech.
There are two reasons the Misery is such a fantastic variant. First, it has a ballistic slot. Ballistics synthesize well with PPCs, so it can run the standard loadouts of all jumpsnipers. Second, it can mount 40 points of damage on its left side, using the right side as a big shield that sucks up almost 80 points of damage.
#26
Posted 11 December 2013 - 08:31 PM
The solution is NERF JUMP JETS
#27
Posted 11 December 2013 - 08:38 PM
No, seriously, though, we're nowhere near the first to come up with it, and it's how Elo really should've been implemented in the first place. That being said, I support everything outlined here. Except the bit about killing me. That's just not very nice. ^.^
At the same time, we still need an answer for the underlying problem - which is that JJs and other things are still out of whack balance-wise. If we can fix that, then this becomes more of a "quality of life" issue rather than a fundamental deal-breaker.
Edited by Siriothrax, 11 December 2013 - 08:40 PM.
#28
Posted 11 December 2013 - 08:40 PM
Homeless Bill, on 11 December 2013 - 08:27 PM, said:
There are two reasons the Misery is such a fantastic variant. First, it has a ballistic slot. Ballistics synthesize well with PPCs, so it can run the standard loadouts of all jumpsnipers. Second, it can mount 40 points of damage on its left side, using the right side as a big shield that sucks up almost 80 points of damage.
So what you are saying is that JJ are OP?
#29
Posted 11 December 2013 - 09:26 PM
Haree78, on 11 December 2013 - 08:31 PM, said:
The solution is NERF JUMP JETS
Ghogiel, on 11 December 2013 - 08:40 PM, said:
I can't deny that jumpjets are pretty much make-or-break, particularly combined with high-alpha builds. Only a few chassis thrive without them. That said, I don't think fixing balance and improving matchmaking are mutually exclusive.
Siriothrax, on 11 December 2013 - 08:38 PM, said:
No, seriously, though, we're nowhere near the first to come up with it, and it's how Elo really should've been implemented in the first place. That being said, I support everything outlined here. Except the bit about killing me. That's just not very nice. ^.^
At the same time, we still need an answer for the underlying problem - which is that JJs and other things are still out of whack balance-wise. If we can fix that, then this becomes more of a "quality of life" issue rather than a fundamental deal-breaker.
I've seen it proposed many times in the past; it's just always had the problem of being too granular for most players.
I don't disagree that this is a C feature (as I've said in the past). Of all the things I would want them to spend time on right now, this is not it. Until Community Warfare shows up, I really want the ******** side projects kept to a minimum - this one included. I still intend to do an article when the time is appropriate, so I figured I might as well see what holes people can poke ahead of time.
#30
Posted 11 December 2013 - 09:42 PM
All the l33t squads run 12 man meta mechs MOST of the time I see them while most everyone else is leveling stuff or playing no meta mechs. I want to see them lay waste in BLR and TDR lances for once. It isn't like they can't with their skills but day in and day out its 733Cs, STK-M, CTF-3D 12 man meta easymodes. I had been playing for about 3 months before I finally went ahead and got a 733C 3 weeks ago and OMG I couldn't believe how easy it was to wreck shop. It doesn't even take as much skill to be good in that thing as every other mech in the game. That really opens your eyes concerning perceived skill of people who are highly regarded.
When I come across them in their meta drop deck PUGing I just shake my head and put up a top 3 match score on my team in the steamroll.
Hopefully CW will bring the population needed to actually allow ELO to work properly. Jump jets pretty much make any mech without a much less desirable one to play.
Edited by GRiPSViGiL, 11 December 2013 - 09:55 PM.
#31
Posted 11 December 2013 - 09:51 PM
Homeless Bill, on 11 December 2013 - 09:26 PM, said:
They aren't. But STK balance and JJ balance are both balance issues, so pointing to the distinction that they aren't MM issues doesn't answer the question.
#32
Posted 11 December 2013 - 10:33 PM
Listen, I understand you can't fix everything. Bad builds are bad. And I'm not crying: "Fix MY experience for EVERYTHING I want to play!".......no no no, you have to be realistic and understand that there will always be better chassis and better builds for those chassis, and if you take something stupid into the game, you're likely to lose a lot with it.
But what you propose here is a great compromise, I think. It doesn't divide the player base, it isn't all that susceptible to exploits, and it isn't a big tear-up of the framework PGI already put into place. This is a good suggestion, IMO. I hope something like this makes it into the game.
Edited by EJT, 11 December 2013 - 10:35 PM.
#33
Posted 11 December 2013 - 11:22 PM
#34
Posted 11 December 2013 - 11:25 PM
Duncan Aravain, on 11 December 2013 - 03:45 PM, said:
one of hundreds of great ideas ive seen here.
Im over pgi. like an ex.. Happened just a few days ago and its sad. Ill still play but i have zero expectations at this point nor will i devote any brainpower musing whats best for the game.
#35
Posted 12 December 2013 - 12:21 AM
The fact is some mechs / loadouts are far more efficient than others. This is a proof of unbalancing.
With the speed of PGI, this problem will never be solved, you will have hero mechs instead.
There are countless of good ideas in that forum...
I'm keeping an eye on MWO but TBH, i stopped playing it.
Quote
I had been playing for about 3 months before I finally went ahead and got a 733C 3 weeks ago and OMG I couldn't believe how easy it was to wreck shop. It doesn't even take as much skill to be good in that thing as every other mech in the game. That really opens your eyes concerning perceived skill of people who are highly regarded.
I had 40 mechs.
Tried to play different things.
Ended with STK-M & AC40 Jag ... and saw the huge gap of efficiency between meta & non meta builds ...
And then sold most of the useless variant...
It's so easy to core people with that builds... and then just stopped playing. Maybe i'll come back when the game will be balanced.
Edited by loupgaroupoilu, 12 December 2013 - 12:30 AM.
#36
Posted 12 December 2013 - 12:30 AM
Sarsaparilla Kid, on 11 December 2013 - 11:22 PM, said:
I don't know how it be easier to do in terms of a new mech, but a simple average from the first mech's ELO to a newbie (1100) or average player ELO (supposed to be 1300 or so) should suffice. It's easier to do it within the same weight class (Highlander to Atlas) but significantly harder with different ones (Assault to Light). I figure it would be easier to use the average of the first mech's ELO to the second mech of the same weight class's ELO for mechs within the same weight class. If you're starting in a whole new weight class, you probably want to average out with the newbie's starting ELO or maybe even just start with the newbie ELO and go from there.
There's no perfect system, but once you play enough matches, ELO will resolve itself eventually.
Homeless Bill, on 11 December 2013 - 08:27 PM, said:
QFT
Making ELO complicated by range is kinda difficult to make.
Consider this... you have a classic Gaussback (on the 4G or 4H). Should that be classified as a long range build or brawler? You could technically do both, but that makes the system complicated. It's easier to base ELO on overall success on the mech build, whether or not it is long or short range. Of course, better built in stat tools and "private showing of your ELO" would help, but... this is PGI we're talking about.
#37
Posted 12 December 2013 - 12:49 AM
loupgaroupoilu, on 12 December 2013 - 12:21 AM, said:
Troll mechs will still exist even if there was perfect balance. There will always be bad builds and optimal builds. Just look at any stock load out compared to a custom one.
#38
Posted 12 December 2013 - 01:03 AM
if you could come up some idea for pre-made matches that would be great. as i simply stoped playing in groups because the games i having are simply awful. Anything more than a Co-op is unbearable. Ending in a rolling game what is boring in both end but i love to play with my friends.
its just not funn so Im forced to PUG
#39
Posted 12 December 2013 - 01:37 AM
Colonel Pada Vinson, on 11 December 2013 - 04:47 PM, said:
Easier just to fix the game balance issues certain mechs are causing.
Yep, that would be much easier. Just fix the balance. I don't understand why PGI hasn't done that already. Perhaps they wrote it down on a yellow post-it, but then damp air messed with the glue and then the post-it fell down behind their desk, so they left the game in a state of poor gameplay balance for a year. That's happened to me sometimes.
Can someone get on Twitter and remind PGI to fix the game balance issues?
Cheers, Colonel.
#40
Posted 12 December 2013 - 05:59 AM
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users