Jump to content

Experiment To Test The Clan/is Skill Gap Hypothesis


58 replies to this topic

#41 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 22 December 2016 - 10:06 AM

View PostMawai, on 21 December 2016 - 07:46 PM, said:


A couple of questions ...

Are you dropping on the IS or clan side? (from your comments it sounds like clans)

Are you dropping solo or in a group? (I couldn't tell from your comment but I am guessing solo)

If you are dropping clan solo then you may be biasing your results since ...
1) The other side could have larger teams
2) You will never see the effect of clan 12 mans

It is quite possible that random players on both clans and IS sides are more or less equal in skill ... however, the coordination of a 12 man group is far more significant in determining match balance since unless 12 mans are generally playing other 12 mans or at least other large groups (which I don't believe is the case at the moment) then these large units will tend to win relatively easily. If one side in FW has substantially more 12 man teams or organized groups dropping then it should have a distinct advantage whether or not the skill of individual random players is the same as the IS side or not.


If you look at the screenshots, you can see that im dropping on the clan side. I am dropping solo by the way.

Dropping solo will actually have less bias than dropping in a group. Lets say i drop in a 12 man full of top players. We win 99% of our matches. What does that show? That clans have better players on average? No, it just shows that our 12 man had better players than our opposition. Dropping solo means you are randomizing your team and is the only way to see if your side really has better players on average. Sure, you wont end up on a 12 man, but you can still end up on teams with partial groups, and the same applies to the enemy team. IS 12 mans wont exclusively be paired vs our team either...if anything, you would expect IS 12 mans to get paired vs those "clan teams full of top players" more often than not if there really is a huge skill/population gap.

Ive already explained this several times, but I will try again :

-Steam release skill gap between clans and IS was very obvious. Everyone could see it. Dropping solo, most clan vs IS matches showed a very obvious skill gap.

-People on the forums have been claiming there is a huge skill gap and that IS pug teams are full of terribly bad players.

-If the above is true, then you should be able to see this easily by dropping solo and seeing if players on the IS side are really as bad as people say, just like you could see it during the steam release.

-So far at least, I have seen pretty much the opposite. It is entirely possible that this may change with more games played.

Why dont we try something for a change : Instead of nitpicking everything, why not IS players drop solo and take screenshots of all their FW matches so we can do a comparison?

#42 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 22 December 2016 - 10:11 AM

View PostJehofi, on 22 December 2016 - 03:35 AM, said:

And to answer all of OPS burning questions:

With 11 results and a estimated chance IS/Clans being 43/57 (this is what i took to be the real value which it is probably not) you get an confidence interval of 38.86%

Meaning with an confidence level of 99% your results can be 38.86% off. Im sorry to have to quote you but "yea there is a problem here".


At no point whatsoever did i say the sample size would remain fixed at 11 games. Why are you doing calculations now and saying the confidence interval is too low, knowing fully well that the sample size will increase?

#43 L3mming2

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,304 posts

Posted 22 December 2016 - 10:16 AM

View PostWintersdark, on 21 December 2016 - 06:06 PM, said:



Well, if you're just addressing random vapid forum posts and "what people say", there's not really much point in the discussion.

Fundamentally, there's no difference between a Clan player and an IS player; hell, most - even the newest - are both Clan AND IS players, after all. After all, I've got a Clan tag, but I garauntee I've more IS mechs than most IS players do =)

In terms of actual results, as Mr. Winter says above, you'd need a MUCH larger sample size to say anything even remotely useful. A dozen matches or so is so small as to be utterly useless. I'm not saying your right or wrong with this, mind, just that your... 11? matches above are just random.

After all, we've all had those days where you lose a dozen QP matches, or win a dozen, when nothing really changed. Just the way the cookie crumbled.




All the above said, proof is in the pudding. Seems to me that Clans are winning more matches overall. Or am I wrong? Someone posted stats about this elsewhere, but I'm not sure where they got it from.


there is a tab "war logs" in FW, it shows all mach results during the current engagement window. you just count the number of is and clan wins..

#44 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 22 December 2016 - 10:23 AM

View PostL3mming2, on 22 December 2016 - 10:16 AM, said:


there is a tab "war logs" in FW, it shows all mach results during the current engagement window. you just count the number of is and clan wins..


As of the time of this post we are more than 4 hours into the NA cycle and IS have a lead in invasion mode, although clans have a huge lead in scouting.

The war log results arent very helpful for proving if one side is OP or whatever though, since it doesnt show most of the variables involved.

#45 Angel of Annihilation

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Infernal
  • The Infernal
  • 8,881 posts

Posted 22 December 2016 - 11:56 AM

I think we can make things easy on you and save you some work.

PGI, who has access to all the data about everyone, every game and well everything, found that the Clan Faction had a larger distribution of highly skilled players than the IS faction.

Experiment can end now because PGI did the work for you and will alot more accuracy I am sure.

#46 Alex Morgaine

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 2,049 posts

Posted 22 December 2016 - 05:23 PM

I suggest putting on the front how many matches you need for minimum data like 100 or 1000 or some set number. When that number is reached, more accurate comparisons may be made.
I know it's been done in the past for pretty much the exact same question: is faction A more powerful/training more, and what does the data show. Commonality of chassis/varients used was a metric iirc.

View PostViktor Drake, on 22 December 2016 - 11:56 AM, said:

I think we can make things easy on you and save you some work.

PGI, who has access to all the data about everyone, every game and well everything, found that the Clan Faction had a larger distribution of highly skilled players than the IS faction.

Experiment can end now because PGI did the work for you and will alot more accuracy I am sure.


Or this xD

#47 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 22 December 2016 - 05:41 PM

View PostViktor Drake, on 22 December 2016 - 11:56 AM, said:

I think we can make things easy on you and save you some work.

PGI, who has access to all the data about everyone, every game and well everything, found that the Clan Faction had a larger distribution of highly skilled players than the IS faction.

Experiment can end now because PGI did the work for you and will alot more accuracy I am sure.


Is this the same PGI who claims that many underperforming mechs like victors and warhawks are fine?

If the clan faction really does a higher distribution of better players, then it stands to reason that on average, there will be a clear skill disparity in matches which anyone can easily see when dropping solo.

This experiment is just to see whether that holds true in reality.

#48 BootyBlaster17

    Rookie

  • 8 posts

Posted 22 December 2016 - 07:18 PM

View PostEl Bandito, on 22 December 2016 - 07:33 AM, said:


Attack on Crete was ultimately wasteful effort though. Malta should have been captured instead, cause Allied naval base there had caused so many shipping losses for the German effort in North Africa.




It's both. Without the Blitzkrieg doctrine, Germany never could have overturned the French flanks with ease. The concept of massed armor breakthrough is what enabled Germany beat France.



There was no damned doctrine, it was just propaganda to explain to superstitious dumb dumbs who would not accept that they lost because they weren't paying attention. Literally reactionary to being cored out from behind by SPL crow or a Jenner IIc.

Edited by BootyBlaster17, 22 December 2016 - 07:19 PM.


#49 Jehofi

    Member

  • PipPipPip
  • Bad Company
  • Bad Company
  • 98 posts

Posted 23 December 2016 - 02:28 AM

View PostJun Watarase, on 22 December 2016 - 09:27 AM, said:


I have already addressed the small sample size thing. Please read the thread.

The sample size will increase as i play more games. At no point whatsoever have i said the sample size would remain fixed at 11 games.

Jolly good, and how do you propose to fix the bias YOU introduce into your data?

View PostJun Watarase, on 22 December 2016 - 10:11 AM, said:

At no point whatsoever did i say the sample size would remain fixed at 11 games. Why are you doing calculations now and saying the confidence interval is too low, knowing fully well that the sample size will increase?
Because for you to get to a confidence interval of 10% you need around 160 games.
5% 590 Games and so on. Hell if i get a 55% winchance just by picking the right side i will take that!

GL getting to 160 games without PDI changing stuff, then you still need to eliminate the bias YOU introduced. BTW why should competitive players not organize in likeminded groups?

Edited by Jehofi, 23 December 2016 - 02:34 AM.


#50 QuantumButler

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 4,534 posts
  • LocationTaiwan, One True China

Posted 23 December 2016 - 04:16 AM

View PostJun Watarase, on 21 December 2016 - 12:30 PM, said:

Obviously if you drop in a 12 man full of top players, 99% of the time the game is won before it even began. Thats not an accurate judgement of whether there is a big skill gap between the two sides. Its like playing poker where you deal yourself two aces at the start...you cannot accurately judge the local poker scene by doing this. There are plenty of reasons for why clans would be winning most of the time...the most obvious one being that a skilled group of clan players are playing many more matches than their IS counterparts and that shifts the tug of war advantage to the clan side.

Anyone who played during the steam release should remember that 90% of the time you would have the clans face teams comprised of mostly IS newbies, even when dropping as randoms. The skill disparity was very, very obvious at the time to everyone (except perhaps PGI).

At the very least, my experience so far shows that the skill gap isnt as big as during the steam release. Of course, your mileage may differ.

Also those screenshots were taken during NA prime time.


You can actually even just drop in a 12 man of really godawful players vs randoms and have a better than 50% winrate, assuming all the players are on voice comms together and can at least follow orders even if they can't shoot straight.

Being in a coordinated group is just that powerful in this game, you hardly need top tier players, one side in FW just needs more groups on comms together than the other side to, on average, totally dominate.

View PostBesh, on 21 December 2016 - 02:58 PM, said:

Tbh, over the course of this day, I have on occasion made a complete ******* out of my self in inGame Chat, simply because I could not believe and stand the totally, absolutely and extremely frustrating CRAP lvl of play IS Groups displayed .

ALL the usual Stuff :

People do NOT communicate AT ALL, no VOIP, NO TXT, nothing .
They dont respsond, even if I gave clear and precise Information ( "guys, there are only 3 Mechs in f5, if we push em hard fast they are easy kills!" -> People keep standing still behind cover, not moving ) .
People seem to have NO clue of positiong, no Map awareness, many do not even seem to look at their MiniMap .

I am really at a loss as to what has been happening in those Matches. It definitely felt as if people did not have a clue, OR did not even try .

Subsequently got rolled by Clan PUGTeams, who were not very good themselves either, but definitely MORE aware of their surroundings, and MORE working together in a coordinated fashion .

reminder that some of the in game chat options are off by default or can be turned off in options and people actually do this because they just want to be truly miserable I guess?

#51 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 24 December 2016 - 09:17 AM

Updated again, total score is now 6 clan wins vs 12 IS wins.

#52 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 24 December 2016 - 11:19 AM

View PostJun Watarase, on 21 December 2016 - 12:12 PM, said:

Edit : Because people aren't reading, I'm going to highlight two important points.

1. The number of games will keep increasing. The sample size is not fixed at some low number as some people have claimed.

2. No final conclusions have been drawn yet, just initial impressions. The current results do not prove or disprove anything.

End edit.

Many players have been posting about how Clans have most of the top units/players now (whether loyalist or via merc contracts) and that most FW matches are ending up with teams of IS newbies facing top clan teams.

If this is true, then it stands to reason that the majority of clan vs IS matches will show a very obvious skill disparity even if you drop solo and make no effort to stack your side full of top players before the match starts. I mean, if most of the top players are on the clan side...then most of the low and mid tier players will be on the IS side.

So i decided to run an experiment to see if real world results matches the hypothesis.

I will be taking screenshots of the end game scores and posting them as I get them. Screenshots taken during NA prime time.

Score so far : 6 clan wins, 12 IS wins.

http://imgur.com/bVsb2pa : Clan win. By 10 points in conquest. The IS could easily have won this since they wiped our first wave with no losses, but after the first wave they just lost focus, mostly milled about in the center, and we managed to barely outcap them.

http://imgur.com/8zwCZFG : IS win. IS unit proceeds to curb stomp clan randoms.

http://imgur.com/1VyacPn : IS win. IS grabs kill and cap advantage from the beginning and maintains it for the rest of the game.

http://imgur.com/OSEIuDB : IS win. IS unit proceeds to curb stomp clan randoms.

http://imgur.com/3ijMUHM : IS win. IS randoms beat clan randoms with a 19 kill advantage.

http://imgur.com/DEs6fjP : Clan win. Was even till the last 8 minutes or so, then half the IS team tried to push while the other half stayed back and Clans got the advantage.

Initial impressions : Its quite clear that the average IS player is actually better than the average clan player, not taking into account the drops where its a unit vs randoms. Not sure where all these "clan teams full of top players" or "terribly bad IS pub teams" are...the only terribly bad matchups ive seen are the ones where the IS have the advantage.

It will be interesting to see if this changes as I play more games and the sample size increases.

Edit 1 :

http://imgur.com/oSYNX6E : IS win. IS unit proceeds to curb stomp clan randoms.

http://imgur.com/NOaeTEg : IS win. Same IS unit shows up again and curb stomps clan randoms.

http://imgur.com/jzYwBdQ : IS win. IS out trades clans at long range, on alpine peaks. IS had at least one lance with a narc/LRM boat setup.

http://imgur.com/LDf1FTk : IS win. IS basically camped right outside their base and refused to move, and several mechs just sat at the spawn gates the whole game with LRM boats or sniper builds.

http://imgur.com/qMcwF4e : Clan win. From what I could tell, several clan players running laser vomit basically focused fired the IS to oblivion and the IS didn't have anyone on their level.

IS are still winning the majority of matches, especially in matches where its mostly randoms vs randoms. Still have not seen any "clan teams full of top players" or "terribly bad IS pug teams".

Edit 2 :

http://imgur.com/nuHqM4R : IS win. IS unit proceeds to curb stomp clan randoms.

http://imgur.com/zSgrGZR : IS win. Same IS unit shows up again and curb stomps clan randoms.

http://imgur.com/oUtNSZT : Clan win. Clan randoms stomp IS 3 man + randoms.

http://imgur.com/ijTBUsG : Clan win. Clan partial premade stomps IS partial premade.

http://imgur.com/sIu7SHI : Clan win. Clan partial premade stomps IS partial premade.

http://imgur.com/Wjdmixm : IS win. IS partial premades stomp clan randoms.

http://imgur.com/xeBZ6Og : IS win. IS unit stomps clan randoms.

Slightly more even results this time. IS still holds a 2:1 win ratio though (in total).

View PostL3mming2, on 21 December 2016 - 12:23 PM, said:


if u look at the war logs you will see that the clans win most maches, a quiq chec now gave 49/120 IS victory's (this is the higest i have seen IS in the 7 times i checked the logs BTW) so yes as a clan pug u will get stomped by good IS pre made's but thats the case to, and even more so when u play as a IS pug...

in short, pugs always get the short stick when in the same que as groops..


Well, it didn't take long to show your personal experience with a small data pool is an anomaly, did it? You claim IS wins 2:1, but when you look at a larger sample it's the Clans winning nearly 3:1. Once more anecdotal evidence is proven to be fallible.

#53 A Shoddy Rental Mech

    Member

  • PipPipPipPipPipPipPip
  • The Boombox
  • The Boombox
  • 590 posts
  • LocationOn my Island, There are many like it, but this one is mine.

Posted 24 December 2016 - 11:24 AM

View PostJun Watarase, on 22 December 2016 - 05:41 PM, said:

If the clan faction really does a higher distribution of better players, then it stands to reason that on average, there will be a clear skill disparity in matches which anyone can easily see when dropping solo.


Only way you could make such a hypothesis if there was a dedicated solo que. Adding groups into the data skews the data.
Most top players drop in groups. Dropping solo means you're less likely see to top players on your side. It just makes you more likely to see top players on the other side.
Only once in your screenshots did you drop with a top tier 6 man (VRGD)
You've dropped against 6+ man Dude, HHOD, MJ12


Russ has the numbers. He stated there are 25% more top tier players on clan side.

View PostJun Watarase, on 22 December 2016 - 05:41 PM, said:

Is this the same PGI who claims that many underperforming mechs like victors and warhawks are fine?


Under performing warhawk?

Still the 3rd best clan assault.
Posted Image

Edited by The Nerf Bat, 24 December 2016 - 11:25 AM.


#54 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 24 December 2016 - 11:49 AM

View PostThe Nerf Bat, on 24 December 2016 - 11:24 AM, said:




Under performing warhawk?

Still the 3rd best clan assault.
Posted Image


Maybe it's underperforming because there are IS mechs ahead of it? :P

#55 Y E O N N E

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 16,810 posts

Posted 24 December 2016 - 11:51 AM

That moment when the Awesome is now considered tied for 3rd place in the race for "best Assault."

Fascinating.

Edited by Yeonne Greene, 24 December 2016 - 11:51 AM.


#56 Jun Watarase

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 3,504 posts

Posted 24 December 2016 - 02:37 PM

View PostThe Nerf Bat, on 24 December 2016 - 11:24 AM, said:


Only way you could make such a hypothesis if there was a dedicated solo que. Adding groups into the data skews the data.
Most top players drop in groups. Dropping solo means you're less likely see to top players on your side. It just makes you more likely to see top players on the other side.


Dropping in groups skews the data. Dropping solo skews the data. There isnt any way for someone to drop without skewing the data. Dropping solo just skews the data the least.

And if clans really do have much more teams dropping, then it stands to reason that the odds of a IS team facing a clan team should be higher than the odds of a IS team facing a random team.

Either way this experiment was never meant to produce 100% accurate data in the first place, because thats impossible.

I wanted to see if actual game experience was anywhere near what people on the forums keep saying it is, especially whether IS pug teams really are as bad as people claim.

#57 Davegt27

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 7,020 posts
  • LocationCO

Posted 24 December 2016 - 02:59 PM

Posted Image

very next match

Posted Image

Edited by Davegt27, 24 December 2016 - 03:29 PM.


#58 Davegt27

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 7,020 posts
  • LocationCO

Posted 24 December 2016 - 06:27 PM

so I swore off FP a few days ago but since this is for science

Posted Image

#59 draiocht

    Member

  • PipPipPipPipPipPipPip
  • 791 posts

Posted 24 December 2016 - 11:06 PM

[mod]Removed the last few off-topic posts.

Please keep replies relevant to the original topic.
Thank you.[/mod]





8 user(s) are reading this topic

0 members, 8 guests, 0 anonymous users