Jump to content

Statistical Analysis Of The 12-0


187 replies to this topic

#121 Lostdragon

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 2,711 posts
  • LocationAlabama

Posted 01 February 2017 - 09:13 AM

View PostCarl Vickers, on 31 January 2017 - 04:59 PM, said:

Depends on the match, if everyone on one team is of equal skill then everyone will be doing between 1100 to 1400 damage. Over 1500 and you are starting to get into carry land damage wise.

Then there are some people who are just beasts and will do high damage no matter who else is on the team.


If everyone on the team is of equal skill and has equal opportunity they will all do 250-300 damage. It is not mathematically possible for everyone to do 1100 to 1400 damage.

#122 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 01 February 2017 - 09:58 AM

View PostLily from animove, on 01 February 2017 - 03:24 AM, said:

I have some massive issues with your new idea of how pilots should be distributed, because your example actually takes an ideal skill distribution as a base for the queue. Sure PGi could EASILY do that but then waiting times may be horrible. The queue of people wating for a game usually does not look even close to the one you took as a base. An idealised example like this is why a implemented idea fails, because the conditions for the idea aren't met ingame very often, especially not when people don't wanna wait serveral minutes without starting to complain.

what you proposed is probably what PGI implemented initially, but simply fails by the reality of available players and weight class distribution.


I think you're misinterpreting my suggestion.

Finding 24 players for a match wouldn't change. Queue times wouldn't change. The conditions for finding those 24 players wouldn't change at all.

The only thing that would change, is after those 24 players are found, matchmaker than looks at what it has thrown together, and investigates a little deeper - it looks at cumulative matchscore or WLR or some other variable, and just swaps a couple players from one team to the other to further balance what it already has.

#123 Khereg

    Member

  • PipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 919 posts
  • LocationDenver, CO

Posted 01 February 2017 - 03:05 PM

View PostLostdragon, on 01 February 2017 - 09:13 AM, said:

If everyone on the team is of equal skill and has equal opportunity they will all do 250-300 damage. It is not mathematically possible for everyone to do 1100 to 1400 damage.


He's talking about FP, not QP. So multiply your numbers by roughly 4 and you'll see they're in agreement.

#124 RAM

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Resolute
  • The Resolute
  • 2,019 posts
  • Google+: Link
  • LocationVancouver, BC

Posted 01 February 2017 - 06:38 PM

First, this is most impressive! Very good start.

View PostTarogato, on 30 January 2017 - 01:43 PM, said:

- I included only matches that ended in 12-0, 12-1, or even 12-2 if it felt particularly stomp-y

Total matches collected: 116

Now it looks much more balanced! PGI, why can't your matchmaker do something like this?

However, your small sample size and variable selection criteria decrease the accuracy of your findings:

12-0
12-1
12-2
etc, should only be compared to themselves.

Perhaps you could look for community submissions to increase the number of samples analyzed.

Finally, despite ELO doing what you propose at the end, it was near universally reviled so I am skeptical that a change towards better balance would be welcomed.


RAM
ELH

#125 NighthawK1337

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 373 posts
  • LocationInner Sphere, Terra, Asia, Philippines

Posted 01 February 2017 - 07:59 PM

View PostTarogato, on 01 February 2017 - 09:58 AM, said:

I think you're misinterpreting my suggestion.

Finding 24 players for a match wouldn't change. Queue times wouldn't change. The conditions for finding those 24 players wouldn't change at all.

The only thing that would change, is after those 24 players are found, matchmaker than looks at what it has thrown together, and investigates a little deeper - it looks at cumulative matchscore or WLR or some other variable, and just swaps a couple players from one team to the other to further balance what it already has.


Best MM suggestion ever. If PGI can code your algorithm then it'll make pub better matched.

#126 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 02 February 2017 - 05:26 AM

View PostRAM, on 01 February 2017 - 06:38 PM, said:

First, this is most impressive! Very good start.


However, your small sample size and variable selection criteria decrease the accuracy of your findings:

12-0
12-1
12-2
etc, should only be compared to themselves.

Perhaps you could look for community submissions to increase the number of samples analyzed.

Finally, despite ELO doing what you propose at the end, it was near universally reviled so I am skeptical that a change towards better balance would be welcomed.


RAM
ELH


For what it's worth, I noticed the trends when I was about 40 matches deep, and I was considering making a post about it then. But then I decided... hrmmm, maybe I should double the sample size... and then some unit mates who knew I was doing it started pitching in, and it quickly tripled. And the trends only got stronger as I went on, so I was like, "it's about time."

#127 Mudhutwarrior

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 4,183 posts
  • LocationThe perimieter, out here there are no stars.

Posted 02 February 2017 - 05:32 AM

This is going no where. The reason I say this is the variables that will effect it cannot be discussed openly here.

Your statistics are junk just as any statistics here are. We are only fed the information PGI wants us to have. The rest is conjecture.

#128 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 02 February 2017 - 06:07 AM

View PostMudhutwarrior, on 02 February 2017 - 05:32 AM, said:

This is going no where. The reason I say this is the variables that will effect it cannot be discussed openly here.

Your statistics are junk just as any statistics here are. We are only fed the information PGI wants us to have. The rest is conjecture.


This is a curiously pessimistic comment. Care to expound on your thoughts? Particularly what "cannot be discussed openly"?

#129 Bud Crue

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Rage
  • Rage
  • 9,953 posts
  • LocationOn the farm in central Minnesota

Posted 02 February 2017 - 06:12 AM

View PostTarogato, on 02 February 2017 - 06:07 AM, said:

This is a curiously pessimistic comment. Care to expound on your thoughts? Particularly what "cannot be discussed openly"?


Oh boy.

Some one post the "and here we go" meme.

#130 Mudhutwarrior

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 4,183 posts
  • LocationThe perimieter, out here there are no stars.

Posted 02 February 2017 - 06:34 AM

View PostTarogato, on 02 February 2017 - 06:07 AM, said:

This is a curiously pessimistic comment. Care to expound on your thoughts? Particularly what "cannot be discussed openly"?


I can't. If I do it will be moderated. What I can say is the information this is based on ignores the hidden factors that pretty much every long time player knows about. You have a core group that performs actions that skew any result you wish to get. It cannot be discussed or inferred on this forum. Been a while but before reddit had multiple threads dedicated to it.

Just like anything power seeks advantage then protects that advantage at all costs. PGI is complicit as these are the whales they need to keep this game funded. As and example there was group banned, Yet they were allowed to return. Everyone knew and many complained so PGI changed the posting rules to cover it up.

#131 Ori Disciple

    Member

  • PipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 66 posts

Posted 02 February 2017 - 07:10 AM

Did I just seriously read a conspiracy theory about a freaking video game?

#132 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 02 February 2017 - 08:00 AM

View PostOri Disciple, on 02 February 2017 - 07:10 AM, said:

Did I just seriously read a conspiracy theory about a freaking video game?


*debates whether to PM him about it, or just let it go because this sounds dumb.*

#133 Dimento Graven

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Guillotine
  • Guillotine
  • 6,208 posts

Posted 02 February 2017 - 08:05 AM

View PostTarogato, on 02 February 2017 - 08:00 AM, said:

*debates whether to PM him about it, or just let it go because this sounds dumb.*
The only thing I can come up with that fits the parameters is hacking/cheating.

That would skew results, and is typically NOT allowed to be discussed in detail on the forums.

I don't think hacking is quite common enough to have a significant effect on the statistics gathered, however, it is interesting to note that PGI is looking to hire people who have experience with online gaming cheating/hacking (on a secondary list of desirable experience)... So... Who really knows, right?

#134 The6thMessenger

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Nova Captain
  • Nova Captain
  • 8,104 posts
  • LocationFrom a distance in an Urbie with a HAG, delivering righteous fury to heretics.

Posted 02 February 2017 - 08:13 AM

View PostDimento Graven, on 02 February 2017 - 08:05 AM, said:

The only thing I can come up with that fits the parameters is hacking/cheating.

That would skew results, and is typically NOT allowed to be discussed in detail on the forums.

I don't think hacking is quite common enough to have a significant effect on the statistics gathered, however, it is interesting to note that PGI is looking to hire people who have experience with online gaming cheating/hacking (on a secondary list of desirable experience)... So... Who really knows, right?


As with any outliers, perhaps we can just discard it.

#135 Dimento Graven

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Guillotine
  • Guillotine
  • 6,208 posts

Posted 02 February 2017 - 08:18 AM

View PostThe6thMessenger, on 02 February 2017 - 08:13 AM, said:

As with any outliers, perhaps we can just discard it.
If you could identify the specific matches cheating/hacking actually took place in, you could.

In this case, you'd have to make some sort of calculation of the percentage of matches in the sample set had actual hacking/cheating take place and add that as a +/- percentile in the accuracy of the interpretation.

As we're not really allowed to know the specifics on the amount of positively identified hacking/cheating taking place, but if we take Russ at his word (which is very dangerous to do, but nonetheless...) it's probably less than 1%.

So the data would have a +/-1% accuracy.

Or did I malf up in this somewhere? I always hated statistical analysis.

Putting the "A N A L" in analysis is really never that 'fun' for me...

#136 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 02 February 2017 - 08:28 AM

If he's talking about cheaters, then the reality is... they have an average performance just like every other player. Even if exactly 50% of the players in my study were cheaters, it wouldn't change anything. Because what we're looking at is a correlation between how individuals perform on average regardless of who they are, and their likelyhood of being on the winning or losing side of a stomp.

#137 Dimento Graven

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Guillotine
  • Guillotine
  • 6,208 posts

Posted 02 February 2017 - 08:39 AM

View PostTarogato, on 02 February 2017 - 08:28 AM, said:

If he's talking about cheaters, then the reality is... they have an average performance just like every other player. Even if exactly 50% of the players in my study were cheaters, it wouldn't change anything. Because what we're looking at is a correlation between how individuals perform on average regardless of who they are, and their likelyhood of being on the winning or losing side of a stomp.
Very true.

Good clarity of thought there...

#138 Ultimax

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 6,979 posts

Posted 02 February 2017 - 09:45 AM

View PostTarogato, on 02 February 2017 - 06:07 AM, said:

This is a curiously pessimistic comment. Care to expound on your thoughts? Particularly what "cannot be discussed openly"?


He peddles in a bunch of conspiracy theories, don't bother wasting your time.

I've seen him in matches a few times in the past, there is no conspiracy against him - his decision making is though.

Edited by Ultimax, 02 February 2017 - 09:45 AM.


#139 GotShotALot

    Member

  • PipPipPipPipPip
  • 125 posts

Posted 02 February 2017 - 10:19 AM

The thing that strikes me about this (very interesting) analysis is that the difference between the stompers and the stompees is in fact fairly small.

In World of Tanks, where very similar MM issues occur, there was a great post a few years back showing how the first few losses of the match often led to the stomp-roll. Basically, when 1 mech/tank happens to encounter 2-4 foes and goes down quickly doing little damage, the scales start to tip. When the 2nd vehicle does the same, it's now 10 guns against 12. This leads to a vicious cycle where mech after mech gets seriously outgunned and taken down before it can deal proportional damage.

Obviously, this does not always happen, and good gameplay and strategy can come back from some losses. But it is the way things generally trend and thus the frequent roflstomps.

In which case, stomps are more of a 'chaos of the battlefield' result and less of a 'MM team selection' issue. Yes, on average, slightly better teams will cause those first couple early deaths to happen somewhat more often. But 'better' MM won't significantly cut down on them.

#140 Dimento Graven

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Guillotine
  • Guillotine
  • 6,208 posts

Posted 02 February 2017 - 10:28 AM

View PostGotShotALot, on 02 February 2017 - 10:19 AM, said:

The thing that strikes me about this (very interesting) analysis is that the difference between the stompers and the stompees is in fact fairly small.

...
That's the thing though, that 'fairly small' percentage adds up to a few hundred points of damage, minimum, in a given match.

'A few hundred points' of damage is effectively one 'mech's worth of damage.

So those 'fairly small' percentages add up in a big way.

Like someone said earlier in the thread, the billion dollar gambling industry in Vegas makes it with an even smaller percentage, so... yeah, small things have a big impact.





10 user(s) are reading this topic

0 members, 10 guests, 0 anonymous users