312 replies to this topic

[ArchSight]

HBizzle, on 27 June 2014 - 01:57 PM, said:

And I will repost my response so folks here can see it.

I don't think you got what I was saying, or I wasn't clear enough, probably the later. Two key stats are below.

Total Damage per Faction
Faction Total %
IS 306883 39.91%
Clans 462113 60.09%

Tonnage per Faction
Faction Total %
IS 76695 44.28%
Clans 96520 55.72%

Right now the total Tonnage of all IS vs Clans is a 44.28% to 55.72%. That means if everything was equal, as in damage, the IS and Clans break down of total damage should be 44.28% to 55.72% because that is the split in weight, which is the only metric that is equal between the two. Instead the percentage of damage is 39.91% to 60.09%, which means the difference between the weight and expected damage is -4.37% against the IS.

That is the key stat as far as balance goes, because if things were perfectly balanced, which they aren't, then the split would be 0% between the difference between damage share and tonnage share. So all we really know right now is that Clan mechs are outperforming their weight share compared to expected damage by only 4.37%, which with such a small sample size isn't statistically significant.

Damage per ton is assuming each ton is the same, which we know it isn't. Damage compared to weight share, or %, is a better estimate since that should be an indicator of expected damage share if both sides were equal.

You've also provided a sample of which mechs are more popular than other's with the top 10's lists. The low # fielded sample sizes show the mechs that seem under played and the high # fielded sizes show the mechs that are being the most played but it's at your ELO bracket as player's pass through to their own ELO brackets.

If a mech has a low participation in matches it could have something wrong with it but that might only matter at the highest and lowest ELO brackets because of performance of how well it may do in game.

[Roland]

HBizzle, on 27 June 2014 - 01:57 PM, said:

Right now the total Tonnage of all IS vs Clans is a 44.28% to 55.72%. That means if everything was equal, as in damage, the IS and Clans break down of total damage should be 44.28% to 55.72% because that is the split in weight, which is the only metric that is equal between the two. Instead the percentage of damage is 39.91% to 60.09%, which means the difference between the weight and expected damage is -4.37% against the IS.

Your analysis here is incorrect.

Aside from the fact that this is a somewhat more convoluted method of calculating efficiency (as damage per ton is much more straight forward, and is just as valid a comparator... but regardless, even working with your method here, you have an error in your methodology.

You are calculating the relative percentages for IS and clan for both tonnage and damage, but then you are only using HALF of that difference in your analysis of how far off the damage numbers are.

To explain further, in your methodology here the IS number is off by 4.37%... but the Clan damage is ALSO off by 4.37% in the other direction. This combines to create an overall damage disparity of 8.84%. You forgot to take into account that the result of your calculation has each party off of the ideal in different directions, and the total deviation from the expected value includes both of them.

[Gigastrike]

Here's a statistic for you: your post took up half the page. That means that you posted more stuff than every other person on page 1, and fluff from the website bordering the posts combined.

[HBizzle]

Roland, on 27 June 2014 - 02:51 PM, said:

Your analysis here is incorrect.

Aside from the fact that this is a somewhat more convoluted method of calculating efficiency (as damage per ton is much more straight forward, and is just as valid a comparator... but regardless, even working with your method here, you have an error in your methodology.

You are calculating the relative percentages for IS and clan for both tonnage and damage, but then you are only using HALF of that difference in your analysis of how far off the damage numbers are.

To explain further, in your methodology here the IS number is off by 4.37%... but the Clan damage is ALSO off by 4.37% in the other direction. This combines to create an overall damage disparity of 8.84%. You forgot to take into account that the result of your calculation has each party off of the ideal in different directions, and the total deviation from the expected value includes both of them.

Good point. I think damage per ton is a bad indicator considering the amount of tons dedicated to dealing damage decreases as a % of a mech's build as it becomes lighter. 8.84% still isn't extraordinarily out of balance considering the Clan mechs are supposed to be better.

[Pariah Devalis]

I feel like damage itself as a basis for comparison is flawed, anyways. Given equal piloting and equal aim, damage to kill a given mech is likely going to be higher for Clan mechs firing on IS mechs just due to the mechanics. Everything you do splashes, and that splash is, as a result, going to be doing non-lethal damage. It fluffs the number, but has little to do with the kill. Meanwhile, something like a Gauss Jager, for example, could be dropping people with an average of 150 damage per kill to a Clan mech's (pulling a number out of my butt here) 200 damage per kill. Same target, same hits, but wasted splash damage.

OTOH, kills are terrible metrics, too, since all it takes is a single laser strike to a wounded mech to claim it as your own.

Edited by Pariah Devalis, 27 June 2014 - 04:16 PM.

[Dago Red]

Not really relevant to the intent of your study but I'd be interested to see it broken down by individual house and clan as well. But that's pure fluff.

[Kin3ticX]

Pariah Devalis, on 27 June 2014 - 04:12 PM, said:

I feel like damage itself as a basis for comparison is flawed, anyways. Given equal piloting and equal aim, damage to kill a given mech is likely going to be higher for Clan mechs firing on IS mechs just due to the mechanics. Everything you do splashes, and that splash is, as a result, going to be doing non-lethal damage. It fluffs the number, but has little to do with the kill. Meanwhile, something like a Gauss Jager, for example, could be dropping people with an average of 150 damage per kill to a Clan mech's (pulling a number out of my butt here) 200 damage per kill. Same target, same hits, but wasted splash damage.

OTOH, kills are terrible metrics, too, since all it takes is a single laser strike to a wounded mech to claim it as your own.

yeah but if you combine damage farm and kills with a rank system it clears out the noise of both stats

for example: hypothetically



you get double confidence something is up.

Edited by Kin3ticX, 27 June 2014 - 04:58 PM.

[Pariah Devalis]

Well, yes. As I said, damage itself. However, we have people looking at the numbers and thinking bigger numbers is necessarily proof of things being 'better' when, when it comes to effectiveness, less is often more.

[Kin3ticX]

you could also throw in damage per kill as a third ranking variable to look at kill efficiency

[InspectorG]

TexAss, on 27 June 2014 - 09:38 AM, said:

I still wish we would fight 10 clans vs 12 IS and not mixed....

Stock Mech Mondays.
We did 5 Clan vs 8 IS. Worked well and was fairly even. Had one brawl down to the very last.

Even though im just a underhive pugger, take my advice and check out SMM. Sign up here: http://mechforceclassic.enjin.com/

You will need teamspeak

[InspectorG]

Shouldnt kill efficiency be something like: Damage done divided by kills? Lower number the better. And some way to factor in kill assists and component destruction.
I would bet in a good team, the kills get 'shared' if everyone pitches in and strips a target down. Overall an efficient team will have lower damage but higher kills.
Prolly high kill assists as well, assuming good players get more 'touches' on the enemy team.

High damage and one kill would indicate spreading the damage around on the other team(could be good if it led to many kill assists), Or a solo kill, which wouldnt the efficiency be the tonnage/firepower difference between mechs?

Atlas kills a Direwhale, 100 vs 100 tons(i think)
Kitfox kills a Direwhale, 30 vs 100 tons...'less' was used to kill 'more' enemy.

Likely the only solid trending would be comp play. Where players are already top efficiency and there is less variance in skill/team composition/elo/loadouts/etc.

Pardon me i i ramble, im no math person and i am a bit drunk now.

[RussianWolf]

Kin3ticX, on 27 June 2014 - 04:53 PM, said:

yeah but if you combine damage farm and kills with a rank system it clears out the noise of both stats

for example: hypothetically



you get double confidence something is up.

unless you are using the same player in similar builds (an atlas is not an atlas is not an atlas), you still have too many variables unaccounted for and your data may or may not be skewed because of it. That's the thing. You just don't know from a screen shot.

To get good data, you really need to run this in a controlled environment where you keep the same players on the teams and run specific builds from both camps.

[chrispalasz]

Kiiyor, on 24 June 2014 - 04:04 PM, said:

More data! Does anyone else collect EOM screenshots? Have any factions been using private matches to host their own IS vs CLAN grudge matches? DOES ANYONE CARE?

Anyone who wants to send me EOM screenshots (and if you do... please, please be objective - no selective collections of one sided stomps please, or no edits of the data to make all the scores spell 80085 (boobs - hilarious!) you can email them to:

sendmemwojunk@hotmail.com

THANK YOU

I'm now collecting untainted data for your charts. You have excellent chart-making skills and these are really useful. I might suggest some more, too, such as percentage of IS mechs in a game compared to Clan mechs... and percentage of wins for teams with more clan mechs...

you could do all sorts of cool things with this data.

This thread is by far the most interesting I've read. Amazing!

Kiiyor, on 24 June 2014 - 04:04 PM, said:

Overall:

Overall, I think the clans are remarkably well balanced when compared to IS mechs, and that some of the disparate figures above are more indicative of the IS needing to develop a new meta to combat the way the Clans are built to wage war. At the moment though, the clans do seem to have an advantage, but it isn't as great as many people seem to think it is. I took the bull by the horns for a few matches and tried to force any IS heavy groups into defence, and we seemed to do very, very well, the usual pug-related one man charges and the like notwithstanding. Many Clan mechs tend to have low slung weapons, and have to expose lots of their profile to unleash their furious barrages of doom.

Also, these are new machines and new weapons; ongoing balance will undoubtedly help bridge any gaps in faction viability.

THOUGHTS?

Here are my thoughts. I do love these charts, but I disagree with your conclusion that the advantage of clan mechs isn't as big as you might think. It is exactly as big as I suspect. It's a major imbalance - and every time I play, I think there's more evidence to support my opinions about it.

Playing with all these clan mechs is super fun for people who own clan mechs. It WILL be fun in the future when I can buy clan mechs, but it's not right now.

Can IS mechs be useful? Yes. Can they win? Yes. But that doesn't mean all is dandy in MWO.

The strategies that IS mechs are forced to use to BE efficient are 2-dimensional and predictable. If two teams enter a battle field and one is clan-mech heavy and the other isn't... the one that isn't is fighting a gigantic up-hill battle. With a great strategy and great team work AND with the enemy playing slightly stupid, they can win. But with JUST a great strategy and careful play... if the enemy doesn't make enough mistakes, they simply cannot win. Cannot. The power difference in weapons is THAT great.

Lower heat? Fires faster? AND does MORE damage???? That's asinine. Fine, give clan tech an advantage... but an advantage on all fronts is insane.

Adding these clan mechs basically changes the game by wiping the slate clean regarding IS mechs. I have 20+ IS mechs and 0 clan mechs. In order to compete and have fun, I feel I MUST have several clan mechs, and I feel that the greater majority of my IS mechs are a team liability on the battle field.

I'm tired of these huge adjustments. I've spent lots of money on this game and lots of time playing. I don't want to feel like all that has been wiped clean and I have to start over and spend more money and lots more time. I just want to play with what I have and enjoy it.

The OP Clan tech takes away the enjoyment I had. I'll be searching for other online multi-player games and check back later, here. But in all honesty - having been around so long with this game --- my expectations that things will change enough to make this game worth playing again are VERY low. So many poor development decisions are made in the frequent patches...

Edited by chrispalasz, 27 June 2014 - 09:07 PM.

[Kiiyor]

KGB GRU, on 27 June 2014 - 05:31 AM, said:

While I applaud the effort that went into this, there is an unusual absense of victors and cataphracts which are the backbone of the IS. Thats like removing Madcats from the clans. So what I gathered from this is less innersphere meta more clan win.

Roland, on 27 June 2014 - 05:32 AM, said:

That's largely because the Madcat has replaced the Cataphracts. It's basically an improved Cataphract 3D.

In my mind, those running meta builds would have been very curious to see how the MadCat matched up against the venerable DS and Phract, and would likely have shelved their IS mechs in the name of science.

This sample was taken over about 5 days, which in my mind would have been enough for those IS Meta pilots to roadtest and master the Cat. Maybe that explains the lack of IS Meta.

It could also be that my Medium ELO was low - and that I wasn't being matched up with higher calibre IS machines.

I'm up to about 50 matches now for the IS, and i'm seeing a few more Meta machines, but it's worth noting that many of the big names I recognize are in clan mechs.

[Kiiyor]

Pariah Devalis, on 27 June 2014 - 04:12 PM, said:

I feel like damage itself as a basis for comparison is flawed, anyways. Given equal piloting and equal aim, damage to kill a given mech is likely going to be higher for Clan mechs firing on IS mechs just due to the mechanics. Everything you do splashes, and that splash is, as a result, going to be doing non-lethal damage. It fluffs the number, but has little to do with the kill. Meanwhile, something like a Gauss Jager, for example, could be dropping people with an average of 150 damage per kill to a Clan mech's (pulling a number out of my butt here) 200 damage per kill. Same target, same hits, but wasted splash damage.

OTOH, kills are terrible metrics, too, since all it takes is a single laser strike to a wounded mech to claim it as your own.

Good points. In my mind, damage / kill (when balanced with the # of kills) is one of the more valid stats. To me, the end state of an exchange is the telling factor - a Clan mech can deal out obscene damage, but if it's the one cored and lying in a pool of it's own coolant under the unyielding metal boot of the DragonSlayer who killed it with a third less damage, then the DragonSlayer is the better killer. On the other hand, a DragonSlayer can kill as efficiently as it wants, but is still as dead as any other mech if the superior yet less focused firepower of a clan mech can kill it first.

To me, that's the tipping point in all of this - the balance between IS efficient killing, and Clan weight-of-firepower killing. That's where the balance needs to occur.

I'd hate to see that homogenized with normalized damage.

[N0MAD]

Who wants to bet that if the OP does exact same test in 2 weeks the Clan Efficiency % goes up?
And personally i think by a substantial %.
Anyone?

[Merit Lef]

Great work! keep it up and I would love to see what a larger sample will show.

[Appogee]

Incoming statistical anomaly ...

N0MAD, on 27 June 2014 - 10:30 PM, said:

Who wants to bet that if the OP does exact same test in 2 weeks the Clan Efficiency % goes up?

People are still leveling their Clan Mechs, let alone determining the best loadouts and piloting styles. So yes, their effectiveness will go up over time.

[Kiiyor]

N0MAD, on 27 June 2014 - 10:30 PM, said:

Who wants to bet that if the OP does exact same test in 2 weeks the Clan Efficiency % goes up?
And personally i think by a substantial %.
Anyone?

It already was increasing in the data sample towards the end of the 110 matches. I think it will increase further as time goes on, but not as drastically. It only took me a couple of days to master my StormCrows, but that may be due to me realizing I had over 100 days or premium time laying fallow just before I started recording match data. I imagine that there were players with far more match time than me mastering theirs earlier, especially IS meta players testing meta MadCats.

One thing that may have affected it would be the type of mech people ran in first; which makes sense to be the mechs people were most excited about, or had the most initial success in. Once people master their most effective Clan killing machines, they may move onto others they are less confident in.... maybe. Also, I think (and hope) that the initial glut of Clan players will stabilize into something closer to a 1:1 ratio with IS mechs, which will lead to statistics that tell a far more viable story.

It's hard to predict how the data will swing - I used a progressive average trend line at one stage (it's fat too convoluted to display anything really meaningful though) and while clan overall efficiency was increasing, IS damage/kill was decreasing, and IS mechs were also becoming more survivable.

It'll be interesting to see how the data pans out. I'm already getting data from other sources, which should help tremendously.

[Kiiyor]

chrispalasz, on 27 June 2014 - 08:02 PM, said:

THANK YOU

I'm now collecting untainted data for your charts. You have excellent chart-making skills and these are really useful. I might suggest some more, too, such as percentage of IS mechs in a game compared to Clan mechs... and percentage of wins for teams with more clan mechs...

you could do all sorts of cool things with this data.

This thread is by far the most interesting I've read. Amazing!




Here are my thoughts. I do love these charts, but I disagree with your conclusion that the advantage of clan mechs isn't as big as you might think. It is exactly as big as I suspect. It's a major imbalance - and every time I play, I think there's more evidence to support my opinions about it.

Playing with all these clan mechs is super fun for people who own clan mechs. It WILL be fun in the future when I can buy clan mechs, but it's not right now.

Can IS mechs be useful? Yes. Can they win? Yes. But that doesn't mean all is dandy in MWO.

The strategies that IS mechs are forced to use to BE efficient are 2-dimensional and predictable. If two teams enter a battle field and one is clan-mech heavy and the other isn't... the one that isn't is fighting a gigantic up-hill battle. With a great strategy and great team work AND with the enemy playing slightly stupid, they can win. But with JUST a great strategy and careful play... if the enemy doesn't make enough mistakes, they simply cannot win. Cannot. The power difference in weapons is THAT great.

Lower heat? Fires faster? AND does MORE damage???? That's asinine. Fine, give clan tech an advantage... but an advantage on all fronts is insane.

Adding these clan mechs basically changes the game by wiping the slate clean regarding IS mechs. I have 20+ IS mechs and 0 clan mechs. In order to compete and have fun, I feel I MUST have several clan mechs, and I feel that the greater majority of my IS mechs are a team liability on the battle field.

I'm tired of these huge adjustments. I've spent lots of money on this game and lots of time playing. I don't want to feel like all that has been wiped clean and I have to start over and spend more money and lots more time. I just want to play with what I have and enjoy it.

The OP Clan tech takes away the enjoyment I had. I'll be searching for other online multi-player games and check back later, here. But in all honesty - having been around so long with this game --- my expectations that things will change enough to make this game worth playing again are VERY low. So many poor development decisions are made in the frequent patches...

Thanks in advance for the data! If you can see your way to sending me some more IS stuff as well as Clan, i'd be very grateful.

I do agree somewhat on your points about the overall Clan superiority, but i'm leery of making that great a conclusion until I see some more balanced data. There are just too many variables to consider when you weigh in the increased number of clan mechs - it means that the IS samples are diluted somewhat, even if there is 1 more clan mech than IS mechs in a battle. There is also no way to know who is shooting who - It could be that more clan mechs in fights were shooting other clan mechs, or only IS mechs... etc.

Ideally, we'd use full Clan sides vs full IS sides. I imagine that would be possible with private matches, but would be hard to organize, and probably not viable as a truly unique representation of overall player skill/effectiveness as you'd likely be using the same players over and over.