Jump to content

The Number Is In, And It's 90%


692 replies to this topic

#621 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 09 August 2014 - 12:16 PM

Elo Clarification -

Since this seems to confuse people a bit -

If your team has an Elo average of 1250 and my team has an Elo average of 1200, the prediction would run about a 5% advantage for you to win. That's it. Your team has a 1 in 20 better odds than my team. That's about as closed to balanced as you can get and it's actually well within a reasonable margin of error; we may in fact be 50/50 but a few points variance of Elo in a couple people on each team and it might, in reality, be the other way.

Next, what happens on the win/loss.

If you win, because it predicted you would win?

NOTHING. Everyone keeps the same score. So when you get stomped by a flat out better team you don't lose Elo, they don't gain Elo. That match was shits and giggles, CBills and XP.

If I win, because it predicted I would lose?

Everyone on my team would gain about 30 points in Elo. Everyone on your team would lose about 30 pts in Elo. So we'd end up with your team average at 1220, my team average at 1230.

Does that make more sense? 50/50 matchmaking is effectively impossible in Elo. There's always going to be a variance, and there should be. That's what allows people to change their scores up and down. Carrying lower Elo players is literally what allows advancement in Elo to be possible for higher scored players. Otherwise higher scored players would constantly be playing in matches where their Elo is always higher than their opponents - that means it's impossible for them to actually gain Elo.

If the team that is predicted to win (has the higher Elo score) wins, nobody gains or loses anything.

To lose Elo you need to lose to a team with a lower average than your own.

To gain Elo you need to win against a team with a higher average than your own.

You are probably gaining or losing 20-30 points each match that fits the above criteria.

This means that while you may swing significantly in the short run your over-all average after 100 matches or so is reasonably accurate, in the same way that if I shave a bit off 1 side of a 6 sided dice I wouldn't notice at first but after 100 rolls I would see a small but measurable difference.

Everyone, at every Elo scale, is playing the same matches that you are. Same odds of being the high Elo team, same odds of being the low Elo team. Same odds of having 2 disconnects, same odds of having the better mechs on their team. All of these variables, everything about your teammates, averages out with every single other player. Your CONSISTENT performance, how much of a shave off your number on that 6 sided dice you are, that is what your Elo score represents.

Not to derail here but that seems to be coming up a lot.

So, to clarify, the 40-90 pt Elo variance on average is minimal. Literally it's a statistical difference of 5-10% advantage in terms of accurately predicting a win/loss. A 90/10 split is indicative of significant balancing issues between chassis.

#622 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 12:21 PM

View PostMischiefSC, on 09 August 2014 - 12:16 PM, said:

Elo Clarification -

Since this seems to confuse people a bit -

If your team has an Elo average of 1250 and my team has an Elo average of 1200, the prediction would run about a 5% advantage for you to win. That's it. Your team has a 1 in 20 better odds than my team. That's about as closed to balanced as you can get and it's actually well within a reasonable margin of error; we may in fact be 50/50 but a few points variance of Elo in a couple people on each team and it might, in reality, be the other way.

Next, what happens on the win/loss.

If you win, because it predicted you would win?

NOTHING. Everyone keeps the same score. So when you get stomped by a flat out better team you don't lose Elo, they don't gain Elo. That match was shits and giggles, CBills and XP.

If I win, because it predicted I would lose?

Everyone on my team would gain about 30 points in Elo. Everyone on your team would lose about 30 pts in Elo. So we'd end up with your team average at 1220, my team average at 1230.

Does that make more sense? 50/50 matchmaking is effectively impossible in Elo. There's always going to be a variance, and there should be. That's what allows people to change their scores up and down. Carrying lower Elo players is literally what allows advancement in Elo to be possible for higher scored players. Otherwise higher scored players would constantly be playing in matches where their Elo is always higher than their opponents - that means it's impossible for them to actually gain Elo.

If the team that is predicted to win (has the higher Elo score) wins, nobody gains or loses anything.

To lose Elo you need to lose to a team with a lower average than your own.

To gain Elo you need to win against a team with a higher average than your own.

You are probably gaining or losing 20-30 points each match that fits the above criteria.

This means that while you may swing significantly in the short run your over-all average after 100 matches or so is reasonably accurate, in the same way that if I shave a bit off 1 side of a 6 sided dice I wouldn't notice at first but after 100 rolls I would see a small but measurable difference.

Everyone, at every Elo scale, is playing the same matches that you are. Same odds of being the high Elo team, same odds of being the low Elo team. Same odds of having 2 disconnects, same odds of having the better mechs on their team. All of these variables, everything about your teammates, averages out with every single other player. Your CONSISTENT performance, how much of a shave off your number on that 6 sided dice you are, that is what your Elo score represents.

Not to derail here but that seems to be coming up a lot.

So, to clarify, the 40-90 pt Elo variance on average is minimal. Literally it's a statistical difference of 5-10% advantage in terms of accurately predicting a win/loss. A 90/10 split is indicative of significant balancing issues between chassis.


Thank you!

#623 IraqiWalker

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 9,682 posts
  • LocationCalifornia

Posted 09 August 2014 - 12:27 PM

View PostPhaeric Cyrh, on 09 August 2014 - 12:08 PM, said:


Can you direct me to a source that confirms that this is how it works?

Why do you think the MM is intentionally assembling a bunch of terrible IS players offset by a couple elite players while the clans have a relatively homogenous assembly of talent? It doesn't make sense really, except to try and explain the results in a way that more easily fits in with your opinion.

Also, this is a variable that is so terribly easy to see and understand that it took the forum all of 10 minutes to begin harping on yet the guys at PGI are completely ignorant of it? It's possible, but it just comes off as arrogance and a bunch of wannabe game designers thinking they are smarter than everyone else.

The MM has actually been a problem since closed beta. As for the case I showed, that was just an example of why variance isn't as useful as showing what the elo range within each team was.

We've all had matches where new players were pitted against veterans, and I've even seen some new players who had just finished their cadet bonus be slapped against top line players out of the blue.

The game's MM needs massive amounts of work, and PGI keeps trying to improve it. However, we've always had mismatched teams, not in clan vs. IS only, I'm talking even back when it was just IS mechs.

View PostHikyuu, on 09 August 2014 - 12:11 PM, said:

right but following timeline we would only have pre-invasion technology, it's not the technology that's wrong, it's the formula.

I hate to refference WoT, but this is why they have a tier system to their tanks. people don't like to admit that an Atlas is better than a Victor, but by it's stats it's a far more intense assault mech, this is what's happening with Clan assaults. they're far more intense than IS assault mechs and they're not in the same class, treating them in the same class is going to make them overpowered, nerfing isn't the solution, changing how matchmaking works is.

most of this wont matter unless it's CW play, so my suggestion is that CW grade clan mechs differently than IS mechs in terms of battle value, making clans focus more on smaller lighter team setups. while the IS can take heavier mechs on the field.

for pugging it'll probably stay Free For All so that helps balance it out (clans and IS on both sides.)

I would agree with you if it wasn't for the fact that WoT uses a different armor system than ours.

In WoT a tier 1 tank can't even hope to dent a T10 tank after 17 minutes of non-stop fire. While here, my COM-2Ds ML is just as effective as the Atals's own MLs.

So there's no need for the tier system in MW:O, espcially considering how piloting skill is a bigger factor here.

#624 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 12:31 PM

View PostIraqiWalker, on 09 August 2014 - 12:27 PM, said:

The MM has actually been a problem since closed beta. As for the case I showed, that was just an example of why variance isn't as useful as showing what the elo range within each team was.

We've all had matches where new players were pitted against veterans, and I've even seen some new players who had just finished their cadet bonus be slapped against top line players out of the blue.

The game's MM needs massive amounts of work, and PGI keeps trying to improve it. However, we've always had mismatched teams, not in clan vs. IS only, I'm talking even back when it was just IS mechs.


Ok, so even if this is true, why was the MM only assembling teams that favored the clans? You would think there would be a roughly even split of knobbed up matches, or at least something that in some way resembled organic results instead of a statistical probability of 90% in favor of one side or the other.

I have a hard time reconciling your post with MischiefSC's explanation of the Elo system.

#625 IraqiWalker

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 9,682 posts
  • LocationCalifornia

Posted 09 August 2014 - 12:43 PM

View PostPhaeric Cyrh, on 09 August 2014 - 12:31 PM, said:


Ok, so even if this is true, why was the MM only assembling teams that favored the clans? You would think there would be a roughly even split of knobbed up matches, or at least something that in some way resembled organic results instead of a statistical probability of 90% in favor of one side or the other.

I have a hard time reconciling your post with MischiefSC's explanation of the Elo system.

I wasn't saying the MM stacked the odds always in favor of clans.

It's not one big thing that leads to this kind of a stat, it's actually a lot of small things.

Clans have bigger impact in solo drops, MM is borked, and you end up with imbalanced teams on both sides,.

elo variance being almost meaningless, since as shown above, two teams can have equal average elo ratings, but end up with very disparate distribution of elo within them.

For a more proper balance decision (if this test was actually used for balance), the IS v Clans should have been on the 12 v 12 queue, or at most 12v12 and 4+ group queues.

#626 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 09 August 2014 - 12:46 PM

View PostMischiefSC, on 09 August 2014 - 11:57 AM, said:

PPFLD is a significant advantage for probably 5% or less of total players who can consistently, accurately and intentionally hit the same location with each shot. For the other 95% or more that just means that their total DPS is lower because those shots that miss miss completely while a Clan weapon can be walked into target and on to location more easily.


Which begs the following question again: Should balance be dictated by the performance of the top players, or by Joe Regular?

#627 Hobgoblin I

    Member

  • PipPipPipPipPip
  • 131 posts
  • LocationPeoria, IL

Posted 09 August 2014 - 12:49 PM

View PostIraqiWalker, on 09 August 2014 - 12:43 PM, said:

I wasn't saying the MM stacked the odds always in favor of clans.

It's not one big thing that leads to this kind of a stat, it's actually a lot of small things.

Clans have bigger impact in solo drops, MM is borked, and you end up with imbalanced teams on both sides,.

elo variance being almost meaningless, since as shown above, two teams can have equal average elo ratings, but end up with very disparate distribution of elo within them.

For a more proper balance decision (if this test was actually used for balance), the IS v Clans should have been on the 12 v 12 queue, or at most 12v12 and 4+ group queues.


The law of large numbers says that the borkness of the matchups will even out. Sure there were imbalanced sides, just distributed evenly between both sides.

The problem with testing in the 12 man queue is that you would have a much smaller sample size of players, thus increasing the odds of borkness being a factor and not averaging out. You need to have the widest sample of everything you can.

#628 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 12:52 PM

View PostIraqiWalker, on 09 August 2014 - 12:43 PM, said:

I wasn't saying the MM stacked the odds always in favor of clans.

It's not one big thing that leads to this kind of a stat, it's actually a lot of small things.

Clans have bigger impact in solo drops, MM is borked, and you end up with imbalanced teams on both sides,.

elo variance being almost meaningless, since as shown above, two teams can have equal average elo ratings, but end up with very disparate distribution of elo within them.

For a more proper balance decision (if this test was actually used for balance), the IS v Clans should have been on the 12 v 12 queue, or at most 12v12 and 4+ group queues.


That doesn't explain why the MM borked the matches only in favor of the clans. I think you have to take the MM failings out of the equation since given a large enough sample size they should end up relatively even between the two sides.

That leaves us with "Clans have a greater impact in solo drops". So why is this? If one side has to use advanced tactics to win while the other has no such handicap, that in itself suggests an imbalance.

Also I have heard about the test where a competitive team went against clans and won, albeit barely.. This is supposed to support your assertion that clans are more balanced in team environments.. But then I hear that the IS team is one of the best on the scene, using their best mechs, while the clan opposition used un-leveled mechs and were relatively rag-tag.. This doesn't seem to support the original assertion.


View PostHobgoblin I, on 09 August 2014 - 12:49 PM, said:

The problem with testing in the 12 man queue is that you would have a much smaller sample size of players, thus increasing the odds of borkness being a factor and not averaging out. You need to have the widest sample of everything you can.


And this...

Edited by Phaeric Cyrh, 09 August 2014 - 12:53 PM.


#629 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 09 August 2014 - 12:55 PM

View PostPhaeric Cyrh, on 09 August 2014 - 12:31 PM, said:


Ok, so even if this is true, why was the MM only assembling teams that favored the clans? You would think there would be a roughly even split of knobbed up matches, or at least something that in some way resembled organic results instead of a statistical probability of 90% in favor of one side or the other.

I have a hard time reconciling your post with MischiefSC's explanation of the Elo system.


Anecdotal experiences are the only real first hand face people have to put on things. Also the way the human brain works is we remember atypical experiences and tend to forget typical ones. As such we might play 50 games and have 5 lopsided matches packed with newbies on one side plus 2 vets while the other side is a homogeneous mix of experienced players. In hindsight we think that 'that happens all the time' when in reality we had 2 matches where we were the elite player with a team full of nubs and 3 matches where we were in the mix of elites stomping on nubs and we had 45 matches that were a pretty even mix on both sides.

45 seconds after we got stomped because our team was stacked with newbies though we are pissed and rant on the forums, where we get 10 posts from other people who remember that happening to them once. It's called 'confirmation bias'. In our memory this actually accumulates with our perception of personal experience; because we shared our 5% variance experience and got affirmation of similar experience from others we start to remember it as more like 20 or 30% of matches.

It's how our brain works, it's how we recognize potentially dangerous or advantageous patterns and learn new habits. It's an effective survival trait and has helped humanity get where it is today.

It's also why you need to pretty much ignore anecdotal evidence, even when it's put forward very convincingly and passionately. Especially because the people putting it forward absolutely believe and remember that 'this happens all the time', even when in reality it doesn't. You're arguing against their perception of their own memories though, so it's a doomed argument.

Trust the stats. While it's tempting to ask for core data most of us are not actually educated or trained in analytics and so tend to misinterpret the data anyway. Generally this makes more drama and confusion, not less.

Elo works and works well. We probably have close to a 10% margin for error here but given population levels that's good enough. Drilling down would be accomplished more by tailoring it to specific chassis and loadouts but that's a lot of work and would see only a few percentage point improvements for accuracy. It wouldn't make any real difference in your matches; the matchmaker still has to make matches from the people available in that given 2 or 3 minute window. 10% or even 15% variance on accuracy of your Elo won't make a big difference in that. If the game had 2 million active players? Sure. Tens of thousands? No, not really.

The 90/10 data is pretty accurate in representing the difference in chassis and performance. Not everyone is happy with that, but it is what it is.

View PostMystere, on 09 August 2014 - 12:46 PM, said:


Which begs the following question again: Should balance be dictated by the performance of the top players, or by Joe Regular?


Excellent question.

IMO it needs to be both. PPFLD needs balanced for top tier players among each other but it shouldn't be viewed as an advantage overall.

Make sense? You still need to limited big PPFLD alphas because they create their own issues. I'm all for the PPC nerf, I think it was a good choice and long overdue.

I do not think PPFLD needs viewed as a big advantage for the IS though, because overall it isn't. Weapon to weapon it needs balanced but in context of IS vs Clan? It's a tiny factor.

Total DPS vs heat vs range, plus the Clan XL advantage is a way, way WAY bigger set of issues and this showed that.

#630 Hobgoblin I

    Member

  • PipPipPipPipPip
  • 131 posts
  • LocationPeoria, IL

Posted 09 August 2014 - 12:59 PM

View PostIraqiWalker, on 09 August 2014 - 12:43 PM, said:

Clans have bigger impact in solo drops,



Wouldn't that be because they are more powerful than IS mechs...maybe to the tune of, say, a 90/10 advantage?

#631 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 12:59 PM

View PostMischiefSC, on 09 August 2014 - 12:55 PM, said:


Anecdotal experiences are the only real first hand face people have to put on things. Also the way the human brain works is we remember atypical experiences and tend to forget typical ones. As such we might play 50 games and have 5 lopsided matches packed with newbies on one side plus 2 vets while the other side is a homogeneous mix of experienced players. In hindsight we think that 'that happens all the time' when in reality we had 2 matches where we were the elite player with a team full of nubs and 3 matches where we were in the mix of elites stomping on nubs and we had 45 matches that were a pretty even mix on both sides.

45 seconds after we got stomped because our team was stacked with newbies though we are pissed and rant on the forums, where we get 10 posts from other people who remember that happening to them once. It's called 'confirmation bias'. In our memory this actually accumulates with our perception of personal experience; because we shared our 5% variance experience and got affirmation of similar experience from others we start to remember it as more like 20 or 30% of matches.

It's how our brain works, it's how we recognize potentially dangerous or advantageous patterns and learn new habits. It's an effective survival trait and has helped humanity get where it is today.

It's also why you need to pretty much ignore anecdotal evidence, even when it's put forward very convincingly and passionately. Especially because the people putting it forward absolutely believe and remember that 'this happens all the time', even when in reality it doesn't. You're arguing against their perception of their own memories though, so it's a doomed argument.

Trust the stats. While it's tempting to ask for core data most of us are not actually educated or trained in analytics and so tend to misinterpret the data anyway. Generally this makes more drama and confusion, not less.

Elo works and works well. We probably have close to a 10% margin for error here but given population levels that's good enough. Drilling down would be accomplished more by tailoring it to specific chassis and loadouts but that's a lot of work and would see only a few percentage point improvements for accuracy. It wouldn't make any real difference in your matches; the matchmaker still has to make matches from the people available in that given 2 or 3 minute window. 10% or even 15% variance on accuracy of your Elo won't make a big difference in that. If the game had 2 million active players? Sure. Tens of thousands? No, not really.

The 90/10 data is pretty accurate in representing the difference in chassis and performance. Not everyone is happy with that, but it is what it is.



Excellent question.

IMO it needs to be both. PPFLD needs balanced for top tier players among each other but it shouldn't be viewed as an advantage overall.

Make sense? You still need to limited big PPFLD alphas because they create their own issues. I'm all for the PPC nerf, I think it was a good choice and long overdue.

I do not think PPFLD needs viewed as a big advantage for the IS though, because overall it isn't. Weapon to weapon it needs balanced but in context of IS vs Clan? It's a tiny factor.

Total DPS vs heat vs range, plus the Clan XL advantage is a way, way WAY bigger set of issues and this showed that.


Excellent post! Don't have to explain confirmation bias to me, It drives me crazy.

#632 bobF

    Member

  • PipPipPipPipPipPipPip
  • 531 posts

Posted 09 August 2014 - 01:17 PM

Not sure why this thread is 32 pages. Oh, I know why, it's because tryhards who spent cash are trying to defend their p2w mechs. It's just like every other thread in every other game you've played where the developers reveal some metric, and every mouthbreather with agenda logs in to refute or support it. Anyone who's played TT or any pvp game with persistence and customization could've just taken one look at clan mechs and drawn the logical conclusion: "these mechs are way better, no wonder they cost money."

We know who the tryhards are in this thread, because they are obfuscating real issues with some kind of lawyerly uncertainty. "This test can't be right, the conditions were wrong/not enough sampling/elo is broken/not enough mathing" blah blah blah. I downloaded this game and in about 5 mins of browsing the purchasable stuff it was OBVIOUS the paid for mechs were superior, with all nearly all the advantages preserved from the TT game. Clan mechs simply have more flexibility in loadout, enabling you to build a better machine. It has nothing to do with your mad skills, just like when you thought you were so awesome clubbing nubs in your e133t gear playing game XYZ.

Don't be mad about it, just stop claiming you're so good at the game with your purchased skill. No one rational is expecting you to take a leave of absence from your genuinely substantial job, practice 8-12 hours a day, earning cbills and experimenting with loadouts, so you can compete with some student or unemployed bum with nothing better to do. Your time is valuable. If you find yourself still getting indignant, just stop. Understand that your real life skills enabled you to burn disposable income on a video game, and skills to pay the bills are the most important of all.

All that being said, clans shouldn't get a nerf. This is a f2p game, when you show up with real money, you're SUPPOSED to get an advantage. That is the main driver behind the offering, and the central principle in which the model operates successfully. If people are upset about that, that's just too bad. That's the reality. So let's BUFF IS a little instead of devalue mechs people bought for cash. Just stop it with the lawyer arguments, you people sound like washington politicians i.e. disingenuous, pretentious, and obviously specious.

#633 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 01:30 PM

View PostbobF, on 09 August 2014 - 01:17 PM, said:

Not sure why this thread is 32 pages. Oh, I know why, it's because tryhards who spent cash are trying to defend their p2w mechs. It's just like every other thread in every other game you've played where the developers reveal some metric, and every mouthbreather with agenda logs in to refute or support it. Anyone who's played TT or any pvp game with persistence and customization could've just taken one look at clan mechs and drawn the logical conclusion: "these mechs are way better, no wonder they cost money."

We know who the tryhards are in this thread, because they are obfuscating real issues with some kind of lawyerly uncertainty. "This test can't be right, the conditions were wrong/not enough sampling/elo is broken/not enough mathing" blah blah blah. I downloaded this game and in about 5 mins of browsing the purchasable stuff it was OBVIOUS the paid for mechs were superior, with all nearly all the advantages preserved from the TT game. Clan mechs simply have more flexibility in loadout, enabling you to build a better machine. It has nothing to do with your mad skills, just like when you thought you were so awesome clubbing nubs in your e133t gear playing game XYZ.

Don't be mad about it, just stop claiming you're so good at the game with your purchased skill. No one rational is expecting you to take a leave of absence from your genuinely substantial job, practice 8-12 hours a day, earning cbills and experimenting with loadouts, so you can compete with some student or unemployed bum with nothing better to do. Your time is valuable. If you find yourself still getting indignant, just stop. Understand that your real life skills enabled you to burn disposable income on a video game, and skills to pay the bills are the most important of all.

All that being said, clans shouldn't get a nerf. This is a f2p game, when you show up with real money, you're SUPPOSED to get an advantage. That is the main driver behind the offering, and the central principle in which the model operates successfully. If people are upset about that, that's just too bad. That's the reality. So let's BUFF IS a little instead of devalue mechs people bought for cash. Just stop it with the lawyer arguments, you people sound like washington politicians i.e. disingenuous, pretentious, and obviously specious.


The failure of logic in your post is assuming that us IS pilots don't spend money on the game. Some of us have spent much more on this one than any other much more feature-full game in our lives. Constantly being forced to pony up more money because the stuff we already paid for has now been made obsolete feels like extortion and is no way to keep a player base. I have drawn my line in the sand as to how much I will spend on a video game which in all honesty is as shallow as this one. I have spent about $130 in the 4 months I have been playing this game.. Compare that to the last "Total War" game I purchased over a year ago for $50 that I still load up and still play fairly frequently and you see the discrepancy in value.

If this were a subscription service with real content and an evolving storyline like WOW it might make more sense to keep ponying up cash.. But MWO in it's current state is simply an arena team deathmatch game with no single player content.

I don't mind spending cash on a product I like.. I don't like being forced to keep paying cash for the same experience that offers little in the way of new content, or play style beyond "Rock 'em Sock 'em Robots!".

Edited by Phaeric Cyrh, 09 August 2014 - 01:48 PM.


#634 IraqiWalker

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 9,682 posts
  • LocationCalifornia

Posted 09 August 2014 - 01:37 PM

View PostHobgoblin I, on 09 August 2014 - 12:49 PM, said:


The law of large numbers says that the borkness of the matchups will even out. Sure there were imbalanced sides, just distributed evenly between both sides.

The problem with testing in the 12 man queue is that you would have a much smaller sample size of players, thus increasing the odds of borkness being a factor and not averaging out. You need to have the widest sample of everything you can.

View PostPhaeric Cyrh, on 09 August 2014 - 12:52 PM, said:


That doesn't explain why the MM borked the matches only in favor of the clans. I think you have to take the MM failings out of the equation since given a large enough sample size they should end up relatively even between the two sides.

That leaves us with "Clans have a greater impact in solo drops". So why is this? If one side has to use advanced tactics to win while the other has no such handicap, that in itself suggests an imbalance.

Also I have heard about the test where a competitive team went against clans and won, albeit barely.. This is supposed to support your assertion that clans are more balanced in team environments.. But then I hear that the IS team is one of the best on the scene, using their best mechs, while the clan opposition used un-leveled mechs and were relatively rag-tag.. This doesn't seem to support the original assertion.




I would agree with you guys if everyone had access to clan mechs, but since it isolates the pool into 2 pools, you end up with whacked results as far as the MM is concerned.

View PostHobgoblin I, on 09 August 2014 - 12:59 PM, said:


Wouldn't that be because they are more powerful than IS mechs...maybe to the tune of, say, a 90/10 advantage?

Not really, in this case it's because the typical behavior in PuG play enforces what the clans are good at.

Clan mechs excel at long range engagements and pretty much 99% of all PuG drops are 10 minutes of long range pokes, and little use of tactics. Which just exacerbates the IS mechs' weakness at long range. leading to a big imbalance.

Once you start a poke fest against a clan mech, unless you're strapping Gauss, you're going to be under-ranged, and you will lose through attrition. Most people still have a problem with poptart meta, which was solved by plenty of us (players) by simply hiding behind a rock, and using cover to close in on our opponents.

In Pug play it's more the players, than the mechs that are the real problem.

#635 bobF

    Member

  • PipPipPipPipPipPipPip
  • 531 posts

Posted 09 August 2014 - 01:48 PM

View PostPhaeric Cyrh, on 09 August 2014 - 01:30 PM, said:


The failure of logic in your post is assuming that us IS pilots don't spend money on the game. Some of us have spent much more on this one than any other much more feature-full games in our lives. Constantly being forced to pony up more money because the stuff we already paid for has now been made obsolete feels like extortion and is no way to keep a player base. I have drawn my line in the sand as to how much I will spend on a video game which in all honesty is as shallow as this one. I have spent about $130 in the 4 months I have been playing this game.. Compare that to the last "Total War" game I purchased over a year ago for $50 that I still load up and still play fairly frequently and you see the discrepancy in value.

If this were a subscription service with real content and an evolving storyline like WOW it might make more sense to keep ponying up cash.. But MWO in it's current state is simply an arena team deathmatch game with no single player content.

I don't mind spending cash on a product I like.. I don't like being forced to keep paying cash for the same experience that offers little in the way of new content, or play style beyond "Rock 'em Sock 'em Robots!".


Now you understand that despite my incendiary screed, my position is to NOT NERF things, but BUFF underpowered things. In a few more months, clan pilots will feel the same way if they get nerfed. The money they spent becomes a bad "gaming investment" if you will. This is why i'm extremely wary of spending money on this game, despite having original boxed sets of bt, ct, at and succession wars in storage someplace.

My big complaint is that people just aren't honest and admit they prefer to pay money for advantages in pvp. If you paid money AND have skill? You're a walking death machine on the battlefield, gg,no re. All that is fine. Just don't make a post like you're trying to debunk theoretical physics, arguing global warming, or the implications of building 7. People who pay hard-earned cash DESERVE the advantage. That's what fuels a f2p model. So lets not nerf them and make them pissed, but buff UP'd things instead so everyone has a good time. Seems pretty straightforward, no?

#636 Hobgoblin I

    Member

  • PipPipPipPipPip
  • 131 posts
  • LocationPeoria, IL

Posted 09 August 2014 - 01:48 PM

View PostIraqiWalker, on 09 August 2014 - 01:37 PM, said:



I would agree with you guys if everyone had access to clan mechs, but since it isolates the pool into 2 pools, you end up with whacked results as far as the MM is concerned.


Well, players of any skill level can buy clan mechs. The experience factor was taken into account with the prediction of a 60/40 split...it was 90/10.

#637 Mcgral18

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • CS 2019 Top 8 Qualifier
  • CS 2019 Top 8 Qualifier
  • 17,987 posts
  • LocationSnow

Posted 09 August 2014 - 01:50 PM

View PostHobgoblin I, on 09 August 2014 - 01:48 PM, said:


Well, players of any skill level can buy clan mechs. The experience factor was taken into account with the prediction of a 60/40 split...it was 90/10.


What do you predict the private match testing will yield?

Though, I guess we still don't have the specifics for those tests yet. 15th at any rate. Will any of you try making it?

#638 Hobgoblin I

    Member

  • PipPipPipPipPip
  • 131 posts
  • LocationPeoria, IL

Posted 09 August 2014 - 01:56 PM

View PostMcgral18, on 09 August 2014 - 01:50 PM, said:


What do you predict the private match testing will yield?

Though, I guess we still don't have the specifics for those tests yet. 15th at any rate. Will any of you try making it?


I would guess the private match testing would be greatly skewed towards ELO ratings over mech tech as the matching of players wouldn't be random or balanced out at all...but I guess that would depend on how you intend to control that issue.

#639 Phaeric Cyrh

    Member

  • PipPipPipPipPip
  • Mercenary
  • Mercenary
  • 123 posts

Posted 09 August 2014 - 01:57 PM

View PostbobF, on 09 August 2014 - 01:48 PM, said:


Now you understand that despite my incendiary screed, my position is to NOT NERF things, but BUFF underpowered things. In a few more months, clan pilots will feel the same way if they get nerfed. The money they spent becomes a bad "gaming investment" if you will. This is why i'm extremely wary of spending money on this game, despite having original boxed sets of bt, ct, at and succession wars in storage someplace.

My big complaint is that people just aren't honest and admit they prefer to pay money for advantages in pvp. If you paid money AND have skill? You're a walking death machine on the battlefield, gg,no re. All that is fine. Just don't make a post like you're trying to debunk theoretical physics, arguing global warming, or the implications of building 7. People who pay hard-earned cash DESERVE the advantage. That's what fuels a f2p model. So lets not nerf them and make them pissed, but buff UP'd things instead so everyone has a good time. Seems pretty straightforward, no?


I disagree with your entire premise. I didn't pay money so I would have an equipment advantage, and I don't want one. In fact my best mechs for grinding C-Bills and the ones I enjoy piloting the most are ones' I paid for with C-Bills. My DS is nice, but I have more fun and make more space cash in my Raven 3L, and do far more damage in my Atlas AS7-D-DC. My overall favorite mech is my BJ-1 due to it's balance of speed, agility and firepower.

I have spent MC on premium time, mech bays, Hero mechs, paint and once even converted XP.. I plan on purchasing more swag for my cockpits and more paints and camo patterns.. None of these was done so I would gain an advantage on the battle field.

You don't need to make a F2P P2W to have it be viable.. Make it fun and provide for cash content that doesn't make you a better player, but adds to your flair and sense of individuality.. Look at TF2.. Valve is BANKING on the economics they put into that game. None of that stuff is P2W.. Hats, beards, ******* birds on your shoulder.. But people love it, and will pay through the nose if they can afford it.

Edited by Phaeric Cyrh, 09 August 2014 - 01:58 PM.


#640 Mcgral18

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • CS 2019 Top 8 Qualifier
  • CS 2019 Top 8 Qualifier
  • 17,987 posts
  • LocationSnow

Posted 09 August 2014 - 02:00 PM

View PostHobgoblin I, on 09 August 2014 - 01:56 PM, said:


I would guess the private match testing would be greatly skewed towards ELO ratings over mech tech as the matching of players wouldn't be random or balanced out at all...but I guess that would depend on how you intend to control that issue.


Ideally, both parties should have Clan mechs (or swap accounts...risky for some) and switch sides every now and then. Mix the teams up at whatever rate as well.

This should mix the pilots up, to prevent a better team from winning every time, no matter the mechs. It should also help to see if the Clan side has a significant advantage. Switch from 12 VS 10 to 12 VS 12 if it becomes too lopsided.



That should get some data, with limited bias.





18 user(s) are reading this topic

0 members, 18 guests, 0 anonymous users