263 replies to this topic

[CliffhangerJaeger]

The Boz, on 07 November 2014 - 06:25 AM, said:

To anyone saying stuff like "this was caused by DireStormWolf", no, the other Clams aren't significantly weaker than IS mechs.
1. Not being best in class does not mean you automatically suck.
2. Players don't choose mechs randomly, so they can choose a mech that "doesn't suck".
3. There are IS all-star mechs, just like Clan all-star mechs. And non-star mechs on both sides exist too.


If Russ was truly interested, it'd be easy to compensate for this. Accounting for playtime in a certain weight class, divide Elos into "has Clan mech" and "has no Clan mech". Compare average Elos of players with the same playtime. Observe discrepancy between the has and has-nots. Compensate.
The 100 Elo advantage didn't happen accidentally every single time.

I am concerned that PGI may not have thought of this factor though.
There are multiple instances of PGI glossing over the details like this and adjusting the wrong game element in the name of balance.

[Triban]

Food For Thought: Clan ELO is higher not due to skill, but due to the inherent advantages of their chassis. Therefore, one can assume equal skill pilots on unequal footing. People who pilot clan mechs will, over time, experience an ELO inflation greater than their skill level simply due to the advantages their Chassis brings.


Higher ELO=Higher Win Rate
Higher ELO =\= higher skillset (in regards to piloting better mechs)

[The Boz]

Higher Elo isn't higher win rate; you can have a <50% wins and above average Elo.

[Lily from animove]

Triban, on 07 November 2014 - 06:57 AM, said:

Food For Thought: Clan ELO is higher not due to skill, but due to the inherent advantages of their chassis. Therefore, one can assume equal skill pilots on unequal footing. People who pilot clan mechs will, over time, experience an ELO inflation greater than their skill level simply due to the advantages their Chassis brings.


Higher ELO=Higher Win Rate
Higher ELO =\= higher skillset (in regards to piloting better mechs)

the question is, is this true for clan chassis, or is this only true for the Holy trio of doom of the clan chassis?
The entire balance discussion is nonsense when we create them on C vs IS. when 70% o the clanchassis aren't even used to a degree where they can deliver valid data. Blaance has to be created on chassis level, not on faction level.

Edited by Lily from animove, 07 November 2014 - 07:03 AM.

[RG Notch]

Moenrg, on 06 November 2014 - 08:57 PM, said:

What amazes me is that you simply cannot comprehend that for a given elo difference, the chance of winning changes. The "clan' team had a higher ELO in this test - therefore they had a higher chance to win.

Per Russ's own post the clans started with a higher ELO by 90 points, and this advantage was growing. Do you seriously believe that if statistics show that a team with a higher ELO by 90 points was suppose to only win 50% of the time and the clans won 64% of the time, that anyone with an IQ above double digits would think that's balanced? Do you believe you would be the only person to figure this out?

Maybe it's not obvious to you, but I suspect that 64% is pretty close to what the win chance is for a team with a 90 point ELO advantage, as I remember PGI stating 50 was a pretty big difference. If this simple math is beyond you, then please don't drop on my team...ever.

I would think that people with double digit IQ would understand that Elo is designed for Chess.PGI keeps stating these things about Elo because PGI wants people to believe it matters in this game. I don't think I'm the only one who thinks it's not nearly as useful for balance as some people believe. Anyways, we'll see once people can pick sides and particularly once people can swap sides how it looks.

[Lily from animove]

RG Notch, on 07 November 2014 - 07:12 AM, said:

I would think that people with double digit IQ would understand that Elo is designed for Chess.PGI keeps stating these things about Elo because PGI wants people to believe it matters in this game. I don't think I'm the only one who thinks it's not nearly as useful for balance as some people believe. Anyways, we'll see once people can pick sides and particularly once people can swap sides how it looks.

Elo ist used in many games that are not chess and in many of these games it works very well to create balanced matches, so why do only in the mwo forum some people contsantly say elo is not working because it comes form chess. Thats like trying to say we had wheels one mule carts and thats why bicycles should not have tires.

Edited by Lily from animove, 07 November 2014 - 07:16 AM.

[RG Notch]

Lily from animove, on 07 November 2014 - 07:15 AM, said:

Elo ist used in many games that are not chess and in many of these games it works very well to create balanced matches, so why do only in the mwo forum some people contsantly say elo is not working because it comes form chess. Thats like trying to say we had wheels one mule carts and thats why bicycles should not have tires.

Good analogy, mule carts to bicycles is definitely close to Chess to MWO.

[Hillslam]

They should use the much more useful metric of neckbeard length measurement, larping costume contest, and roleplay act-off competition...

[Evil Ed]

No one seems to consider the fact that players using mostly clan-mechs probably has an inflated Elo due to the fact that they have played "meta mechs". So, with the quirks only a few days old the clan pilots haven't gotten there faces stomped enough to get them down to correct Elo bracket yet. Same with us who drives IS most of the time, we need a few weeks to increase our Elo.
Taken this in consideration: To produce a match today with 50/50 probabilty of winning for both sides with the new parameters the quirks are bringing the matchmaker has to pick a clan-team with higher Elo compaired with the IS counter part.

In short: The in general higher Elo of the clan teams in this test can't be used to excuse the 64% winrate. Maybe in a few weeks when your Elo is down to where it sould be - but not now.

Edited by Evil Ed, 07 November 2014 - 07:36 AM.

[WarHippy]

Russ Bullock, on 06 November 2014 - 02:49 PM, said:

I ran a quick Clan vs IS test to gather some data without the risk of changing player behavior ( solo queue only )

Small sample size with Clan victories at 64%

I am relatively happy with this seeing as the test increased the usual 50 ELO difference between teams to ~90 and growing the longer I ran the test - most matches I observed were around 100-120 ELO advantage to the Clan team.

Again with this in mind I think 64% is something I am fairly pleased with.

The new quirks still need more time to settle in and for people to use them in the best way. I may very well do some more quick Clan vs IS tests to gather more data sets.

Underlined above is a major problem when looking at this as it tells you nothing of real value. If you are going to test IS vs Clan it needs to be done uninterrupted for at least 48 hours. A small window with a small subset of players is mostly meaningless. A 100 point Elo advantage interestingly enough does mean a roughly 64% chance to win, however as you said the Elo difference was changing the longer you performed this test so we really need to see a munch bigger sample size.

[Triban]

Lily from animove, on 07 November 2014 - 07:03 AM, said:

the question is, is this true for clan chassis, or is this only true for the Holy trio of doom of the clan chassis?
The entire balance discussion is nonsense when we create them on C vs IS. when 70% o the clanchassis aren't even used to a degree where they can deliver valid data. Blaance has to be created on chassis level, not on faction level.

You can't keep quoting this "trinity" when IS also have like 6 good chassis (out of ALL THEIR MECHS, many more than clans ATM) and more worse ones.

[Lily from animove]

Triban, on 07 November 2014 - 08:31 AM, said:

You can't keep quoting this "trinity" when IS also have like 6 good chassis (out of ALL THEIR MECHS, many more than clans ATM) and more worse ones.

yes I can because unlike the IS emchs in this test who were spreaded a lot between their chassis, the majority of clanemchs used is reduced to these 3.


Screen your own matches, collect the data for yourself and see what varietly the IS fields and what variety the clans.

I doubt russ would give us any information on the matches the clanners have won, but I would like to see from the matches the clanners have won, how many % were belonging to which clanchassis.
And same for the Is matches.

Edited by Lily from animove, 07 November 2014 - 08:34 AM.

[SaltBeef]

Basically what I am hearing is the everyone gets a trophy argument, Clan Players get sharty mechs and no quirks because they are supposed to be better more experienced players... Even with crappier performing mechs are supposed to just suck it up for the new guy....Well we are so kind.

[LordKnightFandragon]

WarHippy, on 07 November 2014 - 08:25 AM, said:

Underlined above is a major problem when looking at this as it tells you nothing of real value. If you are going to test IS vs Clan it needs to be done uninterrupted for at least 48 hours. A small window with a small subset of players is mostly meaningless. A 100 point Elo advantage interestingly enough does mean a roughly 64% chance to win, however as you said the Elo difference was changing the longer you performed this test so we really need to see a munch bigger sample size.

Or uninterrupted for like a week....THAT would give you a real solid number to work with haha.

[The Boz]

Lily from animove, on 07 November 2014 - 07:15 AM, said:

Elo ist used in many games that are not chess and in many of these games it works very well to create balanced matches, so why do only in the mwo forum some people contsantly say elo is not working because it comes form chess. Thats like trying to say we had wheels one mule carts and thats why bicycles should not have tires.

Elo is stretched beyond limits in LoL, a 10-player game. They had to compensate for it, and *heavily*. Elo loses precision if you increase the number of variables, and does so at an exponential rate. Chess has 3 variables, LoL has 11, MWO has TWENTY FIVE.

[WarHippy]

LordKnightFandragon, on 07 November 2014 - 08:36 AM, said:

Or uninterrupted for like a week....THAT would give you a real solid number to work with haha.

Indeed it would, but that might be a bit too chaotic for the community. Like I said at least 48 hours needs to happen because anything less the data is pretty much meaningless.

[Kain Demos]

WarHippy, on 07 November 2014 - 09:01 AM, said:

Indeed it would, but that might be a bit too chaotic for the community. Like I said at least 48 hours needs to happen because anything less the data is pretty much meaningless.

I thought it would cool to PERMANENTLY separate the "solo queue" by faction (Clan vs Clan, IS vs IS) since CW is getting closer.

Basically, this would simulate all the infighting that takes place among the clans and IS alike. I think for new players coming in after the change it would add a huge amount of immersion potentially facing the opposite faction for the first time. THen, whenever they want to run tests like this they can just single out a similar ELO/tonnage team from each queue and pit them against each other.

Of course something this radical would never happen because I sincerely doubt many other people would think this is a good idea.

Edited by Kain Thul, 07 November 2014 - 09:05 AM.

[Mcgral18]

So Russ, could we get a breakdown of the Chassis' used in these tests?

[WarHippy]

Kain Thul, on 07 November 2014 - 09:04 AM, said:

I thought it would cool to PERMANENTLY separate the "solo queue" by faction (Clan vs Clan, IS vs IS) since CW is getting closer.

Basically, this would simulate all the infighting that takes place among the clans and IS alike. I think for new players coming in after the change it would add a huge amount of immersion potentially facing the opposite faction for the first time. THen, whenever they want to run tests like this they can just single out a similar ELO/tonnage team from each queue and pit them against each other.

Of course something this radical would never happen because I sincerely doubt many other people would think this is a good idea.

Well what you suggest would be fun, but Clan vs IS needs to happen as well. That is the major event that takes place 3050-3052.

[SaltBeef]

STMCRW TMBR TMBR STMCRW DW DW KF KF KF TMBR DW STMCRW