Jump to content
You need to play a total of 20 battles to post in this section.
Zim_Xero

Silent Secret MM trick

33 comments in this topic

Recommended Posts

Members
765 posts
2,849 battles

WG is currently using a secret (un-announced) MM trick.  

 

It is categorizing ships into top and bottom performers within a tier, and trying to match them.  Does anyone have any comments on this?

 

BTW... you can tinfoil hat me for this one... but I am sure I am correct.

Share this post


Link to post
Share on other sites
1,359
[LEGIO]
Members
3,736 posts
10,874 battles

Somebody with that in-game stats program and a whole lot of time on their hands could probably put together some sort of pattern if one exists.

Share this post


Link to post
Share on other sites
9,860
[NMKJT]
Beta Testers
24,800 posts
3,947 battles

How would you even measure player skill? There is no way to do it that isn't unfair to a large portion of the playerbase.

Share this post


Link to post
Share on other sites
Members
1,427 posts
9,373 battles

Look up some of the stats of the top clans and you'll see that it's entirely possible to consistently win and do well while winning. 

  • Cool 2

Share this post


Link to post
Share on other sites
Members
765 posts
2,849 battles

its not measuring or using player stats.  Its separating ships into "Top performers" and "the rest"  within each tier.  The evidence is clear if you run Belfasts, Fijis, Kamikazes, etc...  Over 50% of the time you will be matched against another similar top performing ship.  Ishizuchies and Arkansas Betas for instance, tend to get head to head more often than would randomly occur otherwise.   I'm not sure if they are doing it with Carriers, but it is evident with BBs, CAs, and DDs.

Share this post


Link to post
Share on other sites
52,071
[MAUS]
Members
13,697 posts

Yeah, you can't throw percentages around without actual data to back it up, I'm afraid.  I'd be happy to help sift through replays and collect examples that you've witnessed, but we'd need several hundred games of you queueing with the same ship (yes, the exact same ship) over and over again to properly measure this.

  • Cool 4

Share this post


Link to post
Share on other sites
9,860
[NMKJT]
Beta Testers
24,800 posts
3,947 battles
3 minutes ago, Zim_Xero said:

its not measuring or using player stats.  Its separating ships into "Top performers" and "the rest"  within each tier.  The evidence is clear if you run Belfasts, Fijis, Kamikazes, etc...  Over 50% of the time you will be matched against another similar top performing ship.  Ishizuchies and Arkansas Betas for instance, tend to get head to head more often than would randomly occur otherwise.   I'm not sure if they are doing it with Carriers, but it is evident with BBs, CAs, and DDs.

But in order to know someone is a "top performer" it needs to measure player stats.

Share this post


Link to post
Share on other sites
261
[EGI]
[EGI]
Beta Testers
1,243 posts
32,219 battles

But wouldn't the top performing ships by tier and category tend to be the most popular ? Thereby putting them more often in the pool available ships waiting to enter a game because people choose them more often to press Battle with ? It would be more surprising if players chose "over 50% of the time" to play underperforming ships imho.  

 

edit: I am pretty sure the OP is referring to ships in the mix and not the players driving them.  

Edited by Swine_007

Share this post


Link to post
Share on other sites
Members
765 posts
2,849 battles

I'm not gonna collect data on this.... cause it would take an enormous amount to prove.  I just ask you to pay attention to which ships line up against other ships.... taking into account which ships are over 51% Win Rate and which are not. 

Share this post


Link to post
Share on other sites
163
Members
1,022 posts
13,341 battles

he is not talking about top players or player stats.  

just the ships themselves.  example a Belfast gets put on each team  to even it out.    instead of 2 belfasts vs 2 ARPs. each team would get a "top ship"  1 belfast 1 arp   each.

 

  i really dont think WG cares this much :) and you just happen to see lots of good ships on both teams do to people playing more of them

 

Share this post


Link to post
Share on other sites
Members
1,427 posts
9,373 battles

It wouldn't be hard to collect data.

 

Screenshot and wait a month. Then do it again. Compare.

Share this post


Link to post
Share on other sites
Members
1,507 posts
1,138 battles
3 minutes ago, Zim_Xero said:

I'm not gonna collect data on this.... cause it would take an enormous amount to prove.  I just ask you to pay attention to which ships line up against other ships.... taking into account which ships are over 51% Win Rate and which are not. 

It could easily just be down to raw popularity of said ships. I will say I've noticed a significant amount of them tonight

Share this post


Link to post
Share on other sites
7,039
Members
34,409 posts
10,768 battles
21 minutes ago, Zim_Xero said:

 Ishizuchies and Arkansas Betas for instance, tend to get head to head more often than would randomly occur otherwise.   

 

So then you should be able to tell us what the chance of that happening should be, and what it actually is.

 

Otherwise, to quote Blue Swede, you're just "hooked on a feeling".

Share this post


Link to post
Share on other sites
9,421
[GWG]
[GWG]
Alpha Tester
29,183 posts
15,765 battles
25 minutes ago, Zim_Xero said:

its not measuring or using player stats.  Its separating ships into "Top performers" and "the rest"  within each tier.  The evidence is clear if you run Belfasts, Fijis, Kamikazes, etc...  Over 50% of the time you will be matched against another similar top performing ship.  Ishizuchies and Arkansas Betas for instance, tend to get head to head more often than would randomly occur otherwise.   I'm not sure if they are doing it with Carriers, but it is evident with BBs, CAs, and DDs.

Of course they are likely to be on opposite sides, the MM has a tier 4 BB and when the other ques up it has another tier 4 to put against it.

21 minutes ago, Destroyer_Kiyoshimo said:

But in order to know someone is a "top performer" it needs to measure player stats.

He is talking about ship performance not player but he needs to get some tin foil.

Share this post


Link to post
Share on other sites
2,848
Members
5,597 posts
7,121 battles

Anything is possible.

 

I see no reason to ridicule someone that's noticed something and asking  others to keep an eye out themselves.

 

 

 

 

  • Cool 1

Share this post


Link to post
Share on other sites
10,399
[BLNCE]
Members
13,459 posts
44,054 battles

Lets say you are correct. To prove it, youd need a large sample, like 200 games. Then youd need a weighting system to work out the relative chance of a given ship appearing based on the number of players owning that ship and number of battles played. You might need to factor in peak vs off peak tendencies as well. 

 

There is no way you could know MMs habits without this data. Your hunch is meaningless.

Share this post


Link to post
Share on other sites
Members
1,243 posts
5,495 battles

I've seen 2 Gearing go up against 2 Shima before at tier X.  I've also seen 2 Gearing and a Khab against 3 Shima recently.

 

I think they simply try to balance ships by class and tier for random battles - nothing more.  Combos like the above should be allocated differently since if the MM tried to balance teams, then it wouldn't distribute the tier X destroyers they way I have observed on more than one occasion - after all it has the pieces to split the two ships equally.

Share this post


Link to post
Share on other sites
Members
499 posts
7,185 battles

MM isn't even aware of RADAR equipped ships, and loves putting them all on one side.  It's seem incredibly implausible that it's taking ship performance stats into account now.

As far as I can tell MM is heavily weighted to fast queue time, so just throws anything together that can even vaguely pass as balance.   If it was doing anything remotely clever you'd expect far more balanced games, and the deviation of WR would be far less pronounced than it actually is.

Well it's either that or MM is actually a self aware strong AI, and has decided it's sole meaning in existence is to troll the player base. 

It's the only way to explain all the "great" players running 45% win rates that only ever get crap teams...if only MM gave them decent teams, then they could win (Get Carried) way more often.

 

Edited by Veasel

Share this post


Link to post
Share on other sites
180
[TSG4R]
[TSG4R]
Members
972 posts
5,062 battles

I've never seen any objective evidence that the MM takes the relative Win Rate of a specific ship (global or otherwise) into account when creating matchups, and I'd want to see hard statistical evidence or a confirmation from a WG before I'd be willing  to go down that rabbit hole.

However, IIRC WG has stated that the MM it does take into account ship nationality, and tries to distribute ships of the same type and nationality between the two teams.

So if,  for example, if MM has 4 RN cruisers to work with,  it will prefer to put two on each team.

A probable side effect of this is that popular ships like Cleveland which are often present numbers of two or more in any given game will tend to be distributed more or less evenly between the teams, but relative Win Rate of the ship type, as such, is not what is causing the MM to do this.

Share this post


Link to post
Share on other sites
Members
1,494 posts
12,756 battles

So what I think the OP is getting as is if you are in queue in a Fujin and there are other Fujin and/or Kamikaze R in the queue along with Minekazes in queue the MM will put the Fujin/Kamikaze R head to head and the Minekazes head to head.  But the only way to prove this out would be to get a massive sample of Fujin/Kamikaze R/Minekaze games and see what the team composition is (granted this could be done for the Belfast vs Fiji or Shinonome vs Fubuki/Mutsuki etc).  The larger question though...does anyone really care if this is indeed happening?  

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×