54 replies to this topic

[Jay Leon Hart]

Stinger554, on 25 October 2017 - 11:56 AM, said:

Yep, however the difference is that not releasing it is easier than the other options.

[The Lighthouse]

PGI can release stats for each mechs and those stats will be utterly worthless, just like one for the recent event.

They obviously won't release stats that are divided by each tier of players. People say "Brown Sea Tier 1" but the fact is that the quality of games is completely different between Tier 4~5 and Tier 1 (anyone who has smurf accounts for lower Tier understand this.)

Look at Catapult C4. Really only viable build for higher tier game is putting bunch of MRMs or 4 x ASRM6s, which is rather underwhelming brawler (I mean my 45 ton Blackjack 2 can do this, with far better spread for missiles.) I can safely bet this mech's performance in higher level games will be really bad.

However, in lower tier games, even stock 2 x LRM20 is devastating for new/bad players. 4 x LRM10 is even more powerful.


So, in the end if you look at the mech performance stat, you will notice C4 will be placed rather high on the ranking. You will be confused with the stat that disagrees with reality. After a while you will eventually realize that mechs at highest ranks are pretty much mechs that are mostly used for LRM boat.

Then you will also realize why PGI is keep nerfing LRMs and bad IS mechs. This is what happens when you misuse statistics which gives you very wrong conclusion.

Someone who knows how to play the game with some very basic statistics is absolutely needed for PGI to actually balance this game properly, otherwise this game will be never balanced.

Edited by The Lighthouse, 25 October 2017 - 12:40 PM.

[Gas Guzzler]

I'm really not looking to make judgements on mech performance and which ones are good. I'm more interested to see how I perform in mechs compared to the rest of the playerbase.

The data may be garbage, but maybe it will offer insight into changes PGI makes so we aren't completely dumbfounded when changes come through.

Edited by Gas Guzzler, 25 October 2017 - 12:42 PM.

[The Lighthouse]

Gas Guzzler, on 25 October 2017 - 12:41 PM, said:

I'm really not looking to make judgements on mech performance and which ones are good. I'm more interested to see how I perform in mechs compared to the rest of the playerbase.

The data may be garbage, but maybe it will offer insight into changes PGI makes so we aren't completely dumbfounded when changes come through.

Top 10 will be all dominated by traditional LRM boats, and competition fav such as Summoners would be placed rather low.

It is extremely obvious and predictable if you think about how match score is being calculated, and what playstyle requires some skills.

Edited by The Lighthouse, 25 October 2017 - 12:46 PM.

[Sjorpha]

Information is always useful, and this would a great addition.

People keep trying to claim different kinds of data is useless because factor x or y makes it not perfect, or becaue it's not broken down in a million ways that they want, but that simply isn't true.

Especially if the "contamination" of the data is random/chaotic things like bad builds, skill difference etc. Those kinds of disparities tend to average out more often than not, all chassis have bad builds and are driven by differently skilled people, perform differently in different game modes etc, so you will still see meaningful statistical trends in the average performance.

[Gas Guzzler]

The Lighthouse, on 25 October 2017 - 12:46 PM, said:

Top 10 will be all dominated by traditional LRM boats, and competition fav such as Summoners would be placed rather low.

It is extremely obvious and predictable if you think about how match score is being calculated, and what playstyle requires some skills.

Well they can include WLR and kdr and stuff too which offers a little more context. I also said they could separate by PSR tier too.

Sjorpha, on 25 October 2017 - 12:54 PM, said:

Information is ammunition

FTFY

[The Lighthouse]

Gas Guzzler, on 25 October 2017 - 12:54 PM, said:

Well they can include WLR and kdr and stuff too which offers a little more context. I also said they could separate by PSR tier too.

Yeah, they won't separate by PSR, for very obvious reason. WLR may be a bit more useful, but kdr and other stuffs are closely related to damages being done.

Seriously, I know I sound pessimistic.... But if this happens, it will be the biggest waste of the time, only just validating what we already know about the flaws of PGI balancing.

Sjorpha, on 25 October 2017 - 12:54 PM, said:

Information is always useful, and this would a great addition.

People keep trying to claim different kinds of data is useless because factor x or y makes it not perfect, or becaue it's not broken down in a million ways that they want, but that simply isn't true.

Especially if the "contamination" of the data is random/chaotic things like bad builds, skill difference etc. Those kinds of disparities tend to average out more often than not, all chassis have bad builds and are driven by differently skilled people, perform differently in different game modes etc, so you will still see meaningful statistical trends in the average performance.

It would be actually beneficial if the contamination of data is random/chaotic, because it indeed does average out.

The problem is the contamination of data is not random, thus we have insane results such as showing IS is winning the faction warfare, for instance.

[C E Dwyer]

Gas Guzzler, on 25 October 2017 - 09:55 AM, said:

I mean, I think it would be interesting to see what the playerbase as a whole is doing with each mech, and maybe even break it up by PSR tier as well, and keep a real time stat tracker of all the mech performance figures.

Thoughts?

My thoughts are it will never happen.

P.G.I won't even show the population numbers, so they certainly won't show the data, and give the community, the proof they make bad balance passes.

It would be good to see what P.G.I see though

Edited by Cathy, 25 October 2017 - 01:10 PM.

[East Indy]

No reason at all not to. It could more accurately guide quirks, provide the basis for weighted rewards, and even put hearsay to rest.

It's like pre-release testing of a 'Mech. Pull in world champ-level players for a weekend, paying 'em in-game rewards. Test hill-clearing. Test movement. Roll damage -- which locations take fire even when there's a talented guy twisting? And so on.

[Aggravated Assault Mech]

It would be interesting, however it would probably just be misused by people crying for nerfs/buffs.

Without being able to cross-reference multiple performance metrics to see how a certain thing behaves at different skill levels you can't really pull useful information- just seeing raw averages is too inaccurate.

Ultimately averages also tend to be influenced by metaslaves. A thing is determined as being good or bad by the cognoscenti, so players invested enough to participate in discussion about it will shift towards it or away from it, influencing performance up or down. Whoever made the spreadsheet about player tier vs. perception of performance different chassis earlier this year revealed some of this.

[SeventhSL]

As much as I'd love more stats, I don't think a fair chunk of the community is responsible enough to handle having them. E.g The stats from the last faction warfare event. A lot of tears, misconception and blame just because a lot of players lack the ability to put things in context.

Edited by SeventhSL, 25 October 2017 - 02:58 PM.

[chucklesMuch]

I wants more stats give me more data!!!!

*I create, analyst and explain various data/stats/reports in my day job*

[Sjorpha]

The Lighthouse, on 25 October 2017 - 01:02 PM, said:

It would be actually beneficial if the contamination of data is random/chaotic, because it indeed does average out.

The problem is the contamination of data is not random, thus we have insane results such as showing IS is winning the faction warfare, for instance.

Well the information we have on faction warfare shows that Clans are winning 12v12 modes and IS are winning scouting, it is better to have this information that to not have it and it is informative in a number of ways. That PGI presents it in a silly way and makes silly win conditions for events is another matter, it doesn't mean the information itself isn't worth having.

As you say random/chaotic contamination is better, and with statistics on average mech performance it would be.

Another point here about the skewing of results from low skill play is that the lowest tiers in MWO are not the ones with the most players. There are many more long time players than new players in MWO, and there is a strong upward bias to PSR. So tier 1 and 2 are likely the largest in number of players. This means lower tiers might not affect the average result as much as some people expect.

[Tarogato]

Bud Crue, on 25 October 2017 - 11:49 AM, said:

Now excuse me, I am going to go prove this average equivelance by taking on every Deathstrike and Mistlynx G I can find with my St. Ive’s Blues because I believe in PGI!

Pardon my good sir, but it appears you have have misphrased. Allow me to improve,

"I'm sorry, but I routinely last-hit the most frequently encountered mechs on the battlefield with my trusty St. Ive's Blues, the meta-sheep are only making the game more boring for themselves."

[MischiefSC]

Tarogato, on 25 October 2017 - 05:31 PM, said:

Pardon my good sir, but it appears you have have misphrased. Allow me to improve,

"I'm sorry, but I routinely last-hit the most frequently encountered mechs on the battlefield with my trusty St. Ive's Blues, the meta-sheep are only making the game more boring for themselves."

Pfft. Who's the meta sheep?

This is my best build for QP and the only IS heavy I have consistently been able to beat Orion IIC with in 1 v 1 duels.

http://mwo.smurfy-ne...493adb0901bb72b

[Quicksilver Aberration]

Tarogato, on 25 October 2017 - 05:31 PM, said:

Pardon my good sir, but it appears you have have misphrased. Allow me to improve,

"I'm sorry, but I routinely last-hit the most frequently encountered mechs on the battlefield with my trusty St. Ive's Blues, the meta-sheep are only making the game more boring for themselves."

What sorry mistake for a St. Ive's Blues is that? St. Ive's Blues don't use PPCs.
A REAL ST. IVES' BLUES

Edited by Quicksilver Kalasa, 25 October 2017 - 08:24 PM.

[JediPanther]

You only need to take note of the types of mechs you don't see at all in solo or fw to know if the mechs is good. chances are if you don't see a mech it isn't going to have good stats. Spider 5vs any one? Ice ferret? Ansi hero?

[Snazzy Dragon]

Quicksilver Kalasa, on 25 October 2017 - 08:23 PM, said:

What sorry mistake for a St. Ive's Blues is that? St. Ive's Blues don't use PPCs.
A REAL ST. IVES' BLUES

No no no, you've got it all wrong!

http://mwo.smurfy-ne...b#i=205&l=stock

[Lupis Volk]

I'm sorry but as some said people would still challenge the validity of the data and we only need one Alex Jones in this life time, this forum wouldn't be able to handle any sort of Alex Jones wannabe.

[El Bandito]

More data is better, yes.