88 replies to this topic

[Ultimax]

pyrocomp, on 19 June 2017 - 09:53 PM, said:

To be able to do so the stats should show mech performance across different tiers (yes, no selfproclaimed elitism) on different maps and a general metrics for the loadout diversity (however you can do that). Right now even NTG has less than 100 per map. And an unknown number of real pilots in those drops. Really, some are good with lights, some with heavies.

Those stats are taken from Div A MRBC, and on those teams (and competitive teams in general) it's usually the case that the pilots play a certain class of mechs most of the time.

So the numbers you see there are from pilots who are used to, and excel at, piloting those classes of mechs.

[I_AM_ZUUL]

El Bandito, on 20 June 2017 - 04:15 AM, said:

Comp players are the fastest people to find out and use "the latest effective mech/build/tactic", cause they tryhard the most. Common puggers cant even compare. Therefore the most popular mech/build etc they use are naturally the best.

Vindicator was used in "comp"... OP CONFIRMED!!!!!!!!!!!!!!!!

[El Bandito]

I_AM_ZUUL, on 20 June 2017 - 04:50 AM, said:

Vindicator was used in "comp"... OP CONFIRMED!!!!!!!!!!!!!!!!

Frequency matters the most. Experimental mechs dont usually last long.

[VanillaG]

Bud Crue, on 20 June 2017 - 03:08 AM, said:

Consider the Victor for the three years it sat untouched. Yet PGI stuck to their guns that it was fine. How could they have possibly concluded that? For THREE YEARS. My guess is that they didn't look at frequency of play in their analysis. They looked only at its average damage, average this, average that. Its stats probably were within the mean of their target values based on the few hundred matches a month that they were looking at. That did not mean it was a good mech, but rather a few dedicated (probably good) players were playing it. As compared to the Dire, or Kodiak were EVERYBODY was playing it (see leaderboards throughout 2016) It wasn't until last December (when they finally hired new balance folks?) that suddenly changes were made to encourge Vics to actually get played again. I think that is because they finally started looking at frequency of play.

But even now, this is still a factor. If a mechs is determined to be crap by the players, its data set gets skewed due to relative infrequency of play compared to other mechs that are being played. All it takes is a few good players to play that crap mech exceptionally and now that mechs data set is also exceptional, because it is so limited. There is nothing PGI can do to change that except to quirk it in some manner and hope that the change gets it more play so that they can get a better data set thereby. If those quirks are not enough, they keep dumping em on until the lack of player interest finally breaks and the mech gets played. See for example the current state of the Dragon 1C.

Frequency is only one of the metrics you can use but with over 200+ mechs variants in the game it should not be the most important. PGI stated that they are looking for over performers and ways to reign them in not at under performers and ways to raise them up. They tried that with the massive quirkening and decided to reign the quirks back because of the power creep that it introduced. So once you find the over performers you then start looking at the see what makes them over perform. In the most recent case PGI determined that mobility was the cause and implemented engine decoupling. Looking at the data, coming up with a solution, and implementing a solution took months. That change was ready to go in March but was pushed back because of the skill tree.

I would argue that the main reasons for the mechs being rarely seen has to do with hitboxes and hardpoint types and locations. PGI can give give defensive quirks for hitbox issues, see the Firestarter and King Crab this patch, but there is little they can do for hardpoints. Sure they can offensively quirk them for specific weapons but by doing that you turn them into one trick ponies which is what PGI is moving away from. The reality is that some mechs just do not translate well from the lore to MWO and no amount of quirking are going to fix them.

[Ultimax]

El Bandito, on 20 June 2017 - 07:09 AM, said:

Frequency matters the most. Experimental mechs dont usually last long.

The first thing players in this thread (and you might already know, but most don't) need to understand is that this is for MRBC so the mech classes are dictated per drop - with 5 drops for a single "week/match" (1 team vs. 1 team) - and in this season MRBC had 10 weeks (10 matches) for the season (still not complete for some of the teams). 4/5 drops have 2 heavies limit/required. (so 8 heavies per week per team is a given, which is not the case for assaults).

Second is the "one duplicate" rule.

This means if the drop calls for 2/2/2/2 - you may only duplicate ONE chassis.

A legal deck:

ACH & FS9
HBK-IIC & HMN
NTG x2
KDK & MAD-IIC

So that's 8 mechs, 2/2/2/2 and only duplicates of the NTG


Teams work to select a duplicate that adds the best overall improvement to the drop deck.

One season, that was adding 2x Oxides.
Last season (before recent nerfs & before MAD-IIC) it was 2x KDKs becasuse nothing else was even remotely close in raw power and it was able to obliterate mechs with 2x Gauss + 2x CERPPCs or the full Dakka Build

This season, adding a second KDK wasn't necessary as the MAD-IIC could do basically the same thing in a similar package (and after the nerfs overtook the KDK in some ways). So 2x Assaults with near identical capability (or "acceptable level")

The Huntsman was able to replicate what the HBK-IIC CERPPC build did at an acceptable level. 2x Mediums with near identical capability

ACH was often run with a second nearly as good light such as Wolfhound or Firestarter (depending on pilot preference and if you needed JJs). 2x Lights with near identical capability

So that left the Heavies where the single biggest improvement could be made because no other heavy can run Dual Gauss and a PPC - it's the best raw firepower available to the heavy class. Not possible to replicate the capability with a currently available different heavy mech.


(NOTE: There are a lot of other choices possible here, this is just a very simplified example)


When you see other heavy mechs in there, usually something else was going on - the map or strat might have needed a faster or more agile heavy, or perhaps it made more sense to have an ERLLAS sniper such as GRH-5P or perhaps the team was focused on duplicating something else because it simply made more sense so they didn't bring 2x NTGs.


So as I said the previously, the frequency is because the NTG is (was?) the best choice in a variety of situations, maps and strats - but it wasn't always the best for every single strat which is why you see other mechs in there.

Edited by Ultimax, 20 June 2017 - 07:52 AM.

[Deathlike]

pyrocomp, on 20 June 2017 - 03:57 AM, said:

See the point about elitism in the post you quoted. I've seen enough games chocked down by elitist. That group, however good at their abilities to read games mechanics and metrics is too small to sustain any game at minimally acceptable income (just enough to keep those servers running). So, while I'll listen to those who name themselves `comp.players', I'll take that with a significant doubt in regards to what is good for the game and it's future (and from the experience of encountering those in the GQ I'll doubt their undestanding of what this game really is). For the comp.scene the Solaris is a must and the only target. For the rest the maps, modes, PvM and coop.campaing. Completely different things to the point of mutual exclusion.
The only (yes, only) common attribute of each and every game is that it should be `easy to understand, hard to master'. But that is beyound the point of discussion in this thread.

And turning back to the original idea... You see, the comp.players are prone to others opinion and the some common understanding of what a meta is or how the game should be played. At some point I bought a DWF. Struggled with it a bit and came up with a build that I was comfortable with. That was 2xcGR + 6xcMLs. And it was almost half a year before that became a meta. And I checked, no changes for those weapon systems were made. Meta is a.. let's say this way, a todays fashion. Really, it changes with no more than a whim or belief. And it takes almost two month to settle.

So NTGs are heaviliy used. They are more suitable for the playstyle that is percieved as meta (right now that is a long range trade from staionary positions). The meta will get more spread, the players will get acustomed to it and the meta will shift as you will be more effective in other builds that exploit the habbits of playing with previous meta in mind (like a switch to the midrage fast skirmish and a NASCAR as it's simpliest form). The meta has it's seasons and cycles. So are the Mechs usages. As an example, the KDK-3 were hit with a mobility-engine decoupling. The percieved changes were that the KDK-3 was not nerfed but obliterated. And KDKs disppeared from the battlefield completely. Recently, in past two or so weeks the KDKs are back. You see them in matches and so on. And not in a random or joke builds. In reasonable builds. And the patch was not even here. So no, from the mech usage alone by a small players sample in a very narrow time window no reliable conclusion can be made. If we had graphed stats from 4000 players for a year - then there was basis for some conlusions. This is a main point. Any reasonable usage statistics cannot span less than a few months.

The problem you think is that comp players strictly just copy the meta laid before them unquestioningly... when it's really the fact that most of them experience amongst themselves and eventually come to a consensus on what is what.

The average casual player doesn't go that far... at best they notice "hey, I see EmP running groups of X mech, that must mean it's the new meta" and that's the end of that line of logic. No further reasoning/understanding/comprehension done further as to WHY that is... just that WHAT it is.

What also happens with the average casual is that over time, other players (non-casual, but not exactly comp) pick up on this and just simply copy the builds that seem to make them do better. At times, it takes months (for whatever reason) to propagate this "level of understanding" to the masses and then we get threads about metas "that were run for months after the major patch".

We can figure out what is what from the data. The context of that data throughout the season is a whole different ballgame as balance changed (Kodiak-3s became the less obvious first pick when the Skill Tree dropped) and whether you recognize it or not... you'd have to actually experience what a comp team goes through in its decision making process before putting out the majority of builds that are fielded. It's not just conjecture... it's a reality and not something that gets commonly discussed until it's months "too late".

[MischiefSC]

I work in analytics. Mostly I just generate reports, big research programs are uncommon. Sometimes I get handed something new to create reports for and I build the databases and reporting tools.

One of the reasons I post on the forums so much is a big part of my TIME is spent in meetings and conference calls. I don't do anything, not presenting anything, just sitting and listening and people want me in the room. It does not feel like the most productive use of my time as I usually have reports to run or tweak, you know. Stuff to DO. I've had 80+ hour weeks where I probably spent 50 of those hours in meetings where I spoke, like, 2 times.

Then I see threads like this and it reminds me why they want me there. There's some people who just can't wrap their heads around math or statistics. They have this agenda and it's the only lens through which they can see the world - they look at things like the statistics of that reddit post and all they see is the stuff relevant to their agenda. They can't see the whole picture. Literally can not.

Sometimes those people speak up at work and someone like me is in the room to slap them down and drop reality on them and they get hushed by the social pressure of getting proven totally wrong in front of 20-100 people, peers and managers all. In a work environment that really only has to happen to people one time and they seem to get their **** together and learn to either look at the whole picture or at least shut up about their agenda when in public.

So I want to thank Gyrok for the reminder of why I get pulled into meetings where I don't have any real purpose save to prevent someone from trying to post something as stupid as this in a situation where real money (millions and billions) is involved. Sometimes I guess it's just about prevention as much as resolution.

Edited by MischiefSC, 20 June 2017 - 08:21 AM.

[Ultimax]

MischiefSC, on 20 June 2017 - 08:19 AM, said:

I work in analytics. Mostly I just generate reports, big research programs are uncommon. Sometimes I get handed something new to create reports for and I build the databases and reporting tools.

One of the reasons I post on the forums so much is a big part of my TIME is spent in meetings and conference calls. I don't do anything, not presenting anything, just sitting and listening and people want me in the room. It does not feel like the most productive use of my time as I usually have reports to run or tweak, you know. Stuff to DO. I've had 80+ hour weeks where I probably spent 50 of those hours in meetings where I spoke, like, 2 times.

Then I see threads like this and it reminds me why they want me there. There's some people who just can't wrap their heads around math or statistics. They have this agenda and it's the only lens through which they can see the world - they look at things like the statistics of that reddit post and all they see is the stuff relevant to their agenda. They can't see the whole picture. Literally can not.

Sometimes those people speak up at work and someone like me is in the room to slap them down and drop reality on them and they get hushed by the social pressure of getting proven totally wrong in front of 20-100 people, peers and managers all. In a work environment that really only has to happen to people one time and they seem to get their **** together and learn to either look at the whole picture or at least shut up about their agenda when in public.

So I want to thank Gyrok for the reminder of why I get pulled into meetings where I don't have any real purpose save to prevent someone from trying to post something as stupid as this in a situation where real money (millions and billions) is involved. Sometimes I guess it's just about prevention as much as resolution.

This is easily one of the most interesting posts I've read on these forums in months.

[Deathlike]

Ultimax, on 20 June 2017 - 08:57 AM, said:

This is easily one of the most interesting posts I've read on these forums in months.

When MRBC was good to us... most of the time spent was actually planning and not so much "doing".

It's funny when you look back at things and wonder "how much effort it took".

[Bud Crue]

VanillaG, on 20 June 2017 - 07:13 AM, said:

Frequency is only one of the metrics you can use but with over 200+ mechs variants in the game it should not be the most important. PGI stated that they are looking for over performers and ways to reign them in not at under performers and ways to raise them up. They tried that with the massive quirkening and decided to reign the quirks back because of the power creep that it introduced. So once you find the over performers you then start looking at the see what makes them over perform. In the most recent case PGI determined that mobility was the cause and implemented engine decoupling. Looking at the data, coming up with a solution, and implementing a solution took months. That change was ready to go in March but was pushed back because of the skill tree.

I would argue that the main reasons for the mechs being rarely seen has to do with hitboxes and hardpoint types and locations. PGI can give give defensive quirks for hitbox issues, see the Firestarter and King Crab this patch, but there is little they can do for hardpoints. Sure they can offensively quirk them for specific weapons but by doing that you turn them into one trick ponies which is what PGI is moving away from. The reality is that some mechs just do not translate well from the lore to MWO and no amount of quirking are going to fix them.

I hear ya, and even agree to an extent. Frequency of play is but one stat that they may or may not look at. I was just pointing out, that as absurd as Gyrok's assertions maybe in his OP, from a mathematical point of view PGI MUST do something similar with mechs that don't get played very much. If they don't have a decent data set they either have to make educated (we hope) guesses or they rely on very limited and thus potentially very skewed data sets. If they do the latter, then it should be no surprise that some mechs, that we as players recognize as garbage, go long periods without getting any help and why on occasion some mechs get "help" that is way out of line. A limited data set gives them limited information upon which to base their decisions and eventual in game changes. Meh.

[Gyrok]

MischiefSC, on 20 June 2017 - 08:19 AM, said:

I work in analytics. Mostly I just generate reports, big research programs are uncommon. Sometimes I get handed something new to create reports for and I build the databases and reporting tools.

One of the reasons I post on the forums so much is a big part of my TIME is spent in meetings and conference calls. I don't do anything, not presenting anything, just sitting and listening and people want me in the room. It does not feel like the most productive use of my time as I usually have reports to run or tweak, you know. Stuff to DO. I've had 80+ hour weeks where I probably spent 50 of those hours in meetings where I spoke, like, 2 times.

Then I see threads like this and it reminds me why they want me there. There's some people who just can't wrap their heads around math or statistics. They have this agenda and it's the only lens through which they can see the world - they look at things like the statistics of that reddit post and all they see is the stuff relevant to their agenda. They can't see the whole picture. Literally can not.

Sometimes those people speak up at work and someone like me is in the room to slap them down and drop reality on them and they get hushed by the social pressure of getting proven totally wrong in front of 20-100 people, peers and managers all. In a work environment that really only has to happen to people one time and they seem to get their **** together and learn to either look at the whole picture or at least shut up about their agenda when in public.

So I want to thank Gyrok for the reminder of why I get pulled into meetings where I don't have any real purpose save to prevent someone from trying to post something as stupid as this in a situation where real money (millions and billions) is involved. Sometimes I guess it's just about prevention as much as resolution.

Actually, I can see lots of relevant data.

Lots of it met expectations, which goes without requiring any necessary commentary, because that is what was expected.

Commentary, in my mind, is required when data does *NOT* align with expectations.

I was honestly surprised to see the TW so far down on that list, with lots of drops. That high a number of drops, but with extremely poor efficiency comparatively to other mechs in the same weight class.

I would think someone with a background in analytics could appreciate the difference between results that align with what is likely, and results that go against the expectations.

So, I would like to thank you for coming here with an agenda, and not looking at the data analytically...you have quite literally proven your own point, but not about me.

[pyrocomp]

Bud Crue, on 20 June 2017 - 11:23 AM, said:

I hear ya, and even agree to an extent. Frequency of play is but one stat that they may or may not look at. I was just pointing out, that as absurd as Gyrok's assertions maybe in his OP, from a mathematical point of view PGI MUST do something similar with mechs that don't get played very much. If they don't have a decent data set they either have to make educated (we hope) guesses or they rely on very limited and thus potentially very skewed data sets. If they do the latter, then it should be no surprise that some mechs, that we as players recognize as garbage, go long periods without getting any help and why on occasion some mechs get "help" that is way out of line. A limited data set gives them limited information upon which to base their decisions and eventual in game changes. Meh.

A side note, PGI mostly does not have to rely on single digit drops statistics as they have a way large sample in which you get less skewed data. Thus no that specifically limited data sample.

[Gas Guzzler]

Bulletsponge0, on 19 June 2017 - 01:32 PM, said:

statistics are a wonderful thing aren't they

I heard you can manipulate statistics to support your own agenda.

[Bud Crue]

pyrocomp, on 20 June 2017 - 01:41 PM, said:

A side note, PGI mostly does not have to rely on single digit drops statistics as they have a way large sample in which you get less skewed data. Thus no that specifically limited data sample.

Not sure what that means. You can't get a larger data sample if no one is providing data. Have you ever seen a Griffin 2N in scouting? How about a 3M? Yes to both I presume; and quite frequently too. Thus, PGI probably has a nice data pool for which to judge those two mechs in the confines of the scouting mode. Now, when is the last time you ever saw a Griffin 1S? Pretty rare? Perhaps never? So if you have a zillion matches with some mechs and derive your averages from that data pool, the dramatically lesser played mechs averages are easily skewed by a few exceptionally good or bad performances. There is no way to account for that other than by guessing. Educated guessing based on comparisions of similar data points sure, but it is still guessing. There is no math that lets you fill in an utter lack of all relevant data.

[Bulletsponge0]

Gas Guzzler, on 20 June 2017 - 01:49 PM, said:

I heard you can manipulate statistics to support your own agenda.

only 50% of the time. the other 50% they are manipulated to support the other side's agenda. This is true 100% of the time.

[Y E O N N E]

Gyrok, on 20 June 2017 - 11:44 AM, said:

I was honestly surprised to see the TW so far down on that list, with lots of drops. That high a number of drops, but with extremely poor efficiency comparatively to other mechs in the same weight class.

I'm not surprised.

The TBR has always been a mediocre pop-tart. Its ability to pop-tart and ability to steamroll at brawl-range are the two things that it had going for it. With a vastly superior pop-tart like the NTG around, I expect lesser players to choose the TBR for that and be out-classed accordingly. At brawl ranges, targets are usually dead in at most two shots, and it's also far more likely for the TBR to get focused down quickly. The lesser number of drops and lower damage compared to the NTG, to me, suggests that this is the case. Less players naive enough to go TBR for pop-tarting, less maps and drops where brawling is viable.

[Scout Derek]

Bulletsponge0, on 20 June 2017 - 01:54 PM, said:

only 50% of the time. the other 50% they are manipulated to support the other side's agenda. This is true 100% of the time.

but 0% as far as for PGI's agenda that is.

[Scout Derek]

If you really want to see a hole in all of this.... aside from the others here:

It's that not all the mechs were used the same amount of times. If you can give me statistics where mechs all have the same amount of matches used, maybe then, people would actually take you seriously.

See Gyrok, this is why people always call you a troll, you don't provide solid statistics. They're always, how do I put it, half-baked? And then you try to defend yourself amongst in all. No wonder why your other wolf clansman isn't able to post here anymore.


You may be a troll, or you may be a someone who isn't consistent, hell if I know, but I'm just going to do what my gut says and consider this troll material. All of it. Every word said, is just an attempt to rile up people here in the forums by the mockery you put out usually in this sea that's turned quite brown, at least in my eyes.

[pyrocomp]

Bud Crue, on 20 June 2017 - 01:51 PM, said:

Not sure what that means. You can't get a larger data sample if no one is providing data. Have you ever seen a Griffin 2N in scouting? How about a 3M? Yes to both I presume; and quite frequently too. Thus, PGI probably has a nice data pool for which to judge those two mechs in the confines of the scouting mode. Now, when is the last time you ever saw a Griffin 1S? Pretty rare? Perhaps never? So if you have a zillion matches with some mechs and derive your averages from that data pool, the dramatically lesser played mechs averages are easily skewed by a few exceptionally good or bad performances. There is no way to account for that other than by guessing. Educated guessing based on comparisions of similar data points sure, but it is still guessing. There is no math that lets you fill in an utter lack of all relevant data.

That's the thing. I sometimes see Orions and all sorts of Griffins. Even if I don't see it doesn't mean that it is not played at all in T4-5. So the it is a bold assumption that some mechs are not used at all. They might not see usage in T1 or in some competitive legues. But that does not mean that PGI has no sats whatsoever. My guess is that mechs on heneral in lower tiers see quite a number of drops. So no skewing.

[Quicksilver Aberration]

Yeonne Greene, on 20 June 2017 - 01:58 PM, said:

I'm not surprised.

The TBR has always been a mediocre pop-tart. Its ability to pop-tart and ability to steamroll at brawl-range are the two things that it had going for it. With a vastly superior pop-tart like the NTG around, I expect lesser players to choose the TBR for that and be out-classed accordingly. At brawl ranges, targets are usually dead in at most two shots, and it's also far more likely for the TBR to get focused down quickly. The lesser number of drops and lower damage compared to the NTG, to me, suggests that this is the case. Less players naive enough to go TBR for pop-tarting, less maps and drops where brawling is viable.

I'll admit I definitely underestimated the impact of Dual Gauss DPS with the Night Gyr and how the Nova would suddenly become a dominant medium in many matches and how both would basically kill the Timby. Not that I ever thought the Timby was a god mech because it does have some things against it (mostly mount locations and the fact the arms don't really take enough damage) but I didn't expect it to fall as hard as it ended up doing.