Jump to content


'Rigged MM', proven, explained, beaten


  • Please log in to reply
941 replies to this topic

SoTrue #1 Posted Jul 24 2016 - 07:32

    Major

  • Players
  • 33313 battles
  • 3,302
  • [ATTOP] ATTOP
  • Member since:
    04-01-2011

 

Hey all, it's time for another of the budha's 'MM is rigged' posts.  (EDIT:  I've updated this first thread over time.  So it's an easy way to get the jist of all my arguments, and my responses to the better challenges.  If you want to just get to the heart of my position, read the BLUE text near the bottom of the thread.)

-

So in this post I'm going to:

Explain what 'Rigged MM' really means (it's not what most of you think).

Provide documentation as proof.

Explain how mm rigging works.

Show you how to beat it.

Here we go...

-

'Rigged MM'  simply means the way the MM program code has been written -  MM dictates the outcome (or at least heavily influences the outcome) of the battles we play.  Rigged does not mean MM hates any one player specifically (although it can really feel like it, and with the possible exception of me specifically - more on that later).  Are we clear then?  No conspiracies, no tinfoil hats, just the common understanding that software has been written that has the effect of dictating the outcome of battles in this game in a way meant to influence average player win ratios.

-

Now for the documentation.  I'll start with a data set that's not too old, from earlier this year (2016).  Here it is:

Lv6Rpqd.jpg

You can see a much more detailed discussion of this data here: http://forum.worldoftanks.com/index.php?/topic/496320-1-player-1-tank-1183-battles-stats-and-the-case-for-skill-based-mm/page__fromsearch__1  I am providing the spreadsheet again here simply to show that I'm not 'going off my gut' or my 'feelings'.  I have been tracking the data over an extended period of time to support my positions.

Between the data above and the data below there is a gap, but it's not important.  The data below is for the last 5 days I've played.  It's in these last 5 days that the MM has revealed itself to me (more on that after the data).  So here is the data:

DAY 1:  Day 1 is very important as it's the 'last day' in a series of days of regular play.  In other words, I was playing 3-4 days a week for several weeks in a row.  This is important because it means I was 'in the bucket system'.  MM had enough data to gauge whether I was winning or losing to much and then put me into the appropriate winner or loser bucket.

65 W

62 L

59 W

24 L

59 W

70 D

53 W

39 L

53 L

51 L

56 W

72 W

37 W

46 W

50 L

47 L

33 L

48 L

75 W

52 W

44 W

35 L

50 L

39 L

55 W

65 W

59 W

38 W

56 W

Results:  55% overall win rate; 51% overall average chance to win.  Note my overall average for the M6 is 54%.  So I'm right in there where I 'should be'.

DAY 2: Day 2 is very significant because it represents the first day back from two weeks off. This means MM had no current data to gauge whether I was winning or losing to much.  Thus it truly put me in random battles/put me in the 'new player/better chance to win' bucket.  It is very apparent because there are only 2 battles below 50% to win, whereas in Day 1 there were 11/29 or 37% of the battles were under 50% to win.  This 'new player/better chance to win' bucket is the same bucket new players are put in when they start the game, or when you get a new tank.  MM asses that newness and gives you break for the first 'x' number of games.  It's this software mechanism that is also used to put you into a winners or losers bucket based on how well you are doing.

64 W

67 W

59 L

56 W

69 W

61 W

57 W

48 L

47 L

57 W

Results: 70% overall win rate; 58% overall average chance to win.  I was actually going to write a whole separate post about this day.  I've never had such a high overall average chance to win.  Miracles do happen!  But you can see MM clearly was giving me an easy day.

DAY 3: Day 3 is the second day back from the two week break.  MM still doesn't have enough data to gauge whether I was winning or losing too much.  So I'm still in the 'new player/better chance to win' bucket.

51 L

55 W

67 W

41 W

Results: 75% overall win rate; 58% overall average chance to win. Again, another easier day.

DAY 4: Day 4 was a quickie day.

54 W

Results: 100% overall win rate; 54% overall average chance to win.

DAY 5: Day 5 it is key to note at this point that I've now played 3 days in close enough proximity that MM now DOES HAVE enough data to gauge whether I'm winning or losing to much.  And it was too much.  Over the past 3 days my overall average win rate was 11/15 or 73%.  Way to high above MM's 50% target. What you will see in the data below is MM's obvious attempt to bring my win rate back to normal.  I'm going to break this day down into 4 'sub-days'.  The first sub-day represents what MM does to me as a solo player who has been winning to much.  You will see I get much lower overall average chance to win for this sub-day.

Day 5 Sub-01

59 L

46 L

42 L

49 L

66 W

50 W

57 L

42 L

63 L

46 W

27 L

56 L

66 W

44 L

66 W

46 L

36 W ACE

46 L SKUNK

49 W

54 W

Results: 40% overall win rate; 50% overall average chance to win.  Now this is my 5th day back, I'm not rusty at this point.  You can see I got an ace tanker in a 36% battle.  Definitely I'm playing well, yet I went from 73% win rate in 15 battle, to 40% win rate in 20 battles.  That's not me.  What happened was, MM saw that I was winning entirely too much and started to give me much worse teams.  10, or 50% of my battles, were below 50%, compared to 3, or 20%, in the previous 4 days.  The number of sub 50 battles is purely controlled by MM.  That it went from 20% to 50% is no accident.  At this point I'm going to DEFINE BUCKETS.  You've probably been scratching your head as you read the word 'buckets' above.  What are buckets?  It's my slang for 'server pool'.  And by that I mean the game server can't have everyone on one server.  There are many sub-servers.  And MM splits the player base up into chunks, and put them onto different sub-servers.  When it does this, it also breaks the player base up into smaller 'buckets', or pools of players.  These buckets are 'winner' buckets, for those that are new, or have been playing poorly.  And 'loser' buckets', for those who have been playing to well.  As proof these buckets exist, I ask you to think back to the last time you saw a unicum platoon.  Ever notice how they are always on your team/the enemy team if you see them multiple times?  Seeing these unicum platoons over and over verifies that sub-severs exist.  Otherwise, when 20,000 players are playing you should never ever see the same players twice.  The only way to see the same players in a 30 person battle is if those players are all in a smaller sub-server bucket.  Seeing these unicum platoons over and over on your team/the enemy team confirms the 'winner' and 'loser' buckets exist.  If the MM was truly random, you'd see them on your team equally as much as on the enemy team.  So when you see the unicum team on your team 4 out of 4 times (in 6-7 battles), you know you are in the same bucket.

Day 5 Sub-02 Sub-02 is an important break in Day 5.  In the last 54%er battle, I saw another M6 and asked if he wanted to platoon.  He did.  He was a very good player, xvm had him as green.

43 W PLATOON 1

47 L

37 W

51 W

57 W

38 W

30 L

42 W

63 W

33 L

37 L

67 L

50 W

57 W

42 L

Results: 60% overall win rate; 46% overall average chance to win.  Having the second M6 didn't seem to effect the MM from keeping me/us into the 'loser' bucket.  If anything MM punished us for platooning, as the overall average chance to win dropped from 50% down to 46%.  It is important to note that both of us were xvm green.  Which means that, in theory, we should actually increase our average chance to win since we are guaranteeing 2 higher win rates be factored into the xvm calculation.

-

Moving on, this is the point where I show you how to BEAT THE MM.  That's right, there is a way to beat the MM.  Even though MM made it even tougher to win by lowering the overall average chance to win, the win rate went up dramatically, from 40% to 60%.  Platooning between good players is enough to swing your results significantly.

Day 5 Sub-03  Sub-03 is an expansion of 'how to beat MM'.  In the last 42%er we noticed another M6 player, and he joined our platoon.

54 W PLATOON 2

62 W

66 W

63 W

51 W

61 W

73 W

55 W

57 L

Results: 88% overall win rate; 54% overall average chance to win.  It is important to note the impact of 3 good players on your team.  You can see the overall average chance to win jumped from 46% to 54%.  With 3 good players, you can actually bump the xvm calculation in your favor.  This is a big part of the secret to being a unicum.  Having 3 skilled players working together is the biggest advantage you can have in the game.  Especially considering the insane amount of tomatoes the current player base seems to be composed of.  (Note, the other part of being a unicum is actually being a good player, which requires practice and skill).  (Oh, to be clear, I'm not a unicum, I'm just OK).

Day 5 Sub-04  Back to the original 2 of us in the platoon

40 W PLATOON 1

45 W

40 W

81 W

67 W

Results:  100% overall win rate; overall average chance to win 54%.  So even with only 2 in the platoon, still much better than I was doing solo.

-

Whew, that was a lot to take in.  Thanks for making it this far.

-

In summary, what the above represents is my explanation to explain the MM phenomenon as I've seen it play out over thousands of battles.  My general position is a 55% average player should average 50-60% win rate.  Some day a little higher, some days a little lower.  What my data has shown over thousands of battles is, a 55% player can have a few days at 50-60%, but then MM steps in and gives a 'reset' day, a day where a 55% player wins at 30-40%.  I've seen this pattern over and over and over again.  It's too consistent to be random.  And it's just not bad days, I've had days where I start the day playing 20 battles at a 30% win rate, then on the same day, I play 20 more battles and get my win rate for the day back up to 50%.  Now I could consider these 'reset' days to be part of the random nature of MM.  But here's the thing.  Over thousands of battles, I've never had a random day where I just can't lose and win 70% of my battles.  Never not once.  This means there is no bell curve of results.  And if there is no bell curve, it can not be random.  Thus MM is putting me into the 'loser' bucket every now and then to keep me from getting to high a win rate.

-

Note:  Yes yes I know, MM and xvm are two separate and independent programs that do not interact.  But xvm has been documentented as being accurate as you move away from 50% chance to win.  (I.e.  40% chance to win prediction will be more accurate than a 48% chance to win,etc).

-

Note:  There is a quick and easy solution for this.  Skill based MM.  Simply select the 30 tanks with the current MM, then shuffle a few players between teams to bring the chance to win into the 40-60% range.

-

Note:  having xvm chance to win turned on does not mean I quit at a low chance to win game.  I bang my head on my desk, but I still play to win.  See the Ace tanker game for the 36% chance to win. Obviously I don't just quit.

-

Earlier I mentioned the MM may actually be out to get me personally.  I'll let you be the judge.  In the video below, I pull and ace tanker game out of a 36% crap basket.  It's important to realize I won the 'supposed to lose battle' on a night where MM was trying to reset me.  As punishment, look at what happened the very next game (second video).  Yes, the MM hates me personally.  In over 14,000 battles this was the first time I'd ever been on the losing end of a skunk.

-

Video One:  Provided to show I'm not a total noob, total noobs don't get ace tankers.  Provided to show I don't quit when xvm chance to win is crap.  Provided to show that it is worth using team chat to try and give the team direction.  Notice I tell them at the very beginning to play defensively.  They did, and it worked.

-

Vidoe Two:  Provided to show just how much MM hates me as a person.  Really, I felt like shooting myself.  14,000+ without a skunk was a great run.

-

Video three.  Provided to show the power of a platoon.  I can't imagine how easy this game is for purple platoons...  Provided to give cudos to my fellow mutant platoonmates.  They were a fun couple of guys.  I had a really good time battling with them.  It really was a night that showed how much fun this game can be, and how nice some of the player base can be.  FYI, these guys really knew how to work the M6.  They make me look very good in this battle.  While I did get the 6 kills, I did less than 1,500 damage.  Alpha got over 4,000 damage.  Alpha and Pai, my hats off to you both!

-

-

EDIT 01:

So I thought I'd post another data set from another tank.  Note the exact same patter of 'several days of similar results, then a crap day'.

T-54 Stats. Note, I had not been playing the t-54 very much (not including 2011 the first year I played the tank, like a dozen battles a year - not much).  Decided I want to better the 46% win rate I had from 2011.  I'm up to 48%, so moving in right direction.  Also note, the first couple of days reflect this, you'll notice the first day are 'losing' days, but eventually the trend is towards consistent 'winning days'.  Nothing magical, just over 50% (as reflected in the change of win rate from 46% to 48%).  But you'll notice that every couple of days, I have a 'huge loss' day.  This is the same pattern I got with my cruiser IV (the reason for this post).

Green day/just over 50% wrLight green day/way over 50% wr, Black exactly 50%, Purple day/just under 50% wr, light purple, way under 50% wr,

W/L

4/1

19/25 - borderline huge loss day

14/13

16/18

15/23 - borderline huge loss day

8/13

12/10

4/2

23/22

9/8

5/4

8/7

4/2

61/69 -huge 'can't buy a win day, I would be down by 5-10, crawl back get close to 50%, then bam, back down.

23/22

4/1

3/2

8/7

94/113 -huge 'can't buy a win day, I would be down by 5-10, crawl back get close to 50%, then bam, back down.

4/0

5/4

6/0  This might be considered an 'anti-hole' day, but it's not, it's only 6 battles.  A hole day happens when I play 20-ish plus battles...

2/0

10/21 -huge 'can't buy a win day, I would be down by 5-10, crawl back get close to 50%, then bam back down.

4/3

21/21

3/0

6/3

3/3

4/1

3/3

23/26

2/2

-

So we can see the exact same pattern at tier 9 as we saw at tier 8.  Mostly stretches winning at about the same win rate (in this case around 50%), punctuated by a couple of days that were just devastating.  And this is the MAIN point, I'm consistent.  It's not about what level I'm winning at, it's that I'm consistent, and that every couple of days, whamo, huge losing streak.  I didn't suddenly 'get bad', nor was I drunk, or tired.  My level of gameplay was consistent.  It's not that I think I should magically get 'some specific' win rate, it's that if I'm consistent, my winrate should also be consistent.   In the mutant I hover at 55%, in my t-54 I hover at 50-52%.  But every so often, in any tank I play, I get a day that is just horrible.  Now if it was truly random, I should see a bell curve, I should have (3) t-54 days, where I play 20+ battles that day, where I 'can't lose'.  Yet those days never happen.  Have never happened in over 20k battles.  It's only the tragic loss days, never the freebie uber win days.  So it's not random, if it's not random, it's rigged.

-

EDIT 02:

Here's yet another data set from yet another tank/tier (from another post).

-

So I'm starting up the British line, I'm on the Cruiser IV.  Fun little tank.  Was up to 67% win rate (at 200ishB battles)  with it, until tonight.  Dropped down to 64% after 50ish battles played tonight.  Was under 50% several times...

-

May not sound horrible but here are the nights before:

w/l

14/7

16/4

18/10

8/4

8/4

15/3

SEE THE PATTERN HERE?  There was a reason, after 200ish battles I had a 66% win rate.

Then tonight...

L

W

W

W

W

L

W

W

W

W

L

W

W Now up until this point, this night was going as every other night, nice win rate.  Then rigged MM kicks in...

L

L

L

L

L ace tanker, over 1k dmg - at tier 3, 6 kills

L Longest losing streak I've ever had in this tank.

W

D

W

L

W

L

W

L

W

W

L

W

L

L

W

L

L

L

L

L This is the second longest losing streak I've had in this tank.. the longest was earlier tonight.

W

W TOONED UP

W

W

D

L

W  At this point I'm just at 50%, well below the normal 70% I usually average.

L

L

W

W

L

L

L  so now it's fkn 25w to 29l, and the fkn server goes down for maintenance.  so I can't even get to 50%.  f wg, f the sheephole who wrote the mm, f the sheephole who's in charge of it now.

-

EDIT 03:  As you read through this you will hear a lot about the pros and cons of skill based matchmaker (sbmm).  Here is the only thing you need to know about sbmm vs. current mm:

-

Current MM gives you three possible team setups:

*unskilled team vs skilled team (even the best players don't win these, these are 'guaranteed' losses)

*similar team vs similar team (good vs good, or bad vs bad)

*skilled team vs unskilled team (these are gifted wins, sometimes you have to fight to get damage the enemy dies so quickly)

So, no matter what your skill, you WILL BE given one of the three options above (option 1 20% of the time, option 2 60% of the time, option 3 20% of the time).

-

SBMM only gives you one possible team setup:

*similar team vs similar team (good vs good, or bad vs bad)

So, no mater what your skill you will NEVER have to carry a team, nor will you ever be 'gifted' a win.

-

Next, please note the current 3 options do not happen equally.  The first option, where you are on the crap team, happens 20% of the time.  The second option, where both teams are similar, happens about 60% of the time. And the third option, where you are on the much more skilled team, happens about 20% of the time.  Why should you care?  Wasted time.  If the average battle is +/- 6 minutes, do the math on every 1,000 battles.  If 40%, or 400, of them are crap - you spend 6 minutes times 400 battles or 40 HOURS of wasted gameplay.  Now think about how much wasted gameplay MM has forced on those of us with 10k, 20k, 30k...  battles.  WG needs to balance skill.

-

EDIT 03:

Some people were still not getting the above.  So the edit below is a graphic response to the notion that 'sbmm will punish better players by saddling them with more bad players on their teams'.  That's just false.  There is only one playerbase, so over time you'll have be saddled with the same number of bads.  The only difference is gameplay.  Current MM creates very unbalanced teams, which are either very frustrating or very boring.  Sbmm would only create level playing fields of well balanced battles.  Here goes...

-

So, many people have posted tons of data on their battles.  All of them have come in very similar.  'About' 20% of all your battles, you are placed on a very bad team vs. a very good team.  'About' 60% of all your battles both teams are about equal.  This can mean both teams are pretty bad, or both teams are pretty good, the important part is the skill level is close between both teams.  And 'about' 20%  of your battles you are placed on a very good team vs. a very bad team.

-

So let's show actual number here so you can see it in black and white.  I'm going to show 10 battles worth of data for you.  I'm going to organize it in 'groups of occurrence', similar mm will be placed next to each other (in real life they don't come in any particular order).  But for clarities sake I'm showing it this way.  Also, 10 battles is roughly 1 hours worth of play (6 minute average game time).  So it's a reasonable representation of someone sitting down in the evening and getting a few games in.   For this conversation, 'you' will be on team 1.

-

                       team 1                       team 2

battle 1           5 good/10 bad           10 good/5 bad   This is the 20% where mm puts you on a crap team vs.

battle 2           5 good/10 bad           10 good/5 bad   a good team

battle 3           7 good/8 bad              7 good/8 bad    This is the 60% where both teams are 'even'.  I show more

battle 4           7 good/8 bad              7 good/8 bad     bads, 8 vs 7because there are more bads on the server than

battle 5           7 good/ 8 bad             7 good/8 bad     goods.  So even with'even' teams, good players will still be

battle 6           7 good/ 8 bad             7 good/8 bad     'carrying' bad players.

battle 7           7 good/ 8 bad             7 good/8 bad

battle 8           7 good/ 8 bad             7 good/ 8 bad

battle 9           10 good/5 bad            5 good/10 bad   This is the 20% where mm puts you on a good team vs. a

battle 10         10 good/5 bad            5 good/10 bad    crap team.

-

Now lets count the number of 'bads' you had to carry across all games.  You had to carry 78 bads.  Now it's pretty easy to calculate how many bad you would have to carry if ALL your battles were balanced.  Under sbmm, all your games would have been like battles 3-8.  So 8 bads on a team, times 10 battles = 80 bads.  OH MY DEAR LORD, 2 whole extra bads.  The humanity!  The punishment!

-

Here's the real point.  Under the current MM 2 of your games were guaranteed losses.  Those are frustrating as hell.  2 of your games were 'gifts' where you couldn't lose.  Those are boring as hell.  BUT with sbmm, every game is competitive.  Competitive games are much more challenging, and thus much more fun.

-

The final piece of the puzzle is luck.  With the current mm you are subject to extreme bouts of bad luck.  Sometimes I play a couple hours in a day.  I know for a fact that some days start out terrible, and that if I had ONLY PLAYED for the first hour, my win rate would have taken a huge hit.  And this happens all the time.  As the data in my original post shows.  I get several 'normal' days of fairly balanced game play.  Then one very bad day thrown in.  Had I not played on those 'bad days', my win rate would be much much higher.  Oh, I'm sure you are thinking 'what about 'extreme bouts of good luck'.  Those don't happen, and they don't happen for a very specific reason.  The quality of the player base is terrible.  When 20%ish of the player base is good, and 80% is terrible.  On any given day you are far more likely to be saddled with tons of bad players.  There just aren't enough good players on the server to allow an extreme bout of good luck to happen.

-

Finally, I want to make sure to point out that much of the discussion above focused on win rate.  That's just how you have to frame the discussion.  The heart of the matter at hand is quality of gameplay.  When you jump into a battle and you are the only non-red on your team, and the enemy is a pretty rainbow - there is no fun in that.  Don't lie and say there is.  There is also no learning from a whipping like that.  It's pointless.  It's fun-less.  It shouldn't happen.  It's well within WGs power to make those kinds of games disappear.

-

EDIT 04:

-

Of all the bogus statistical arguments you're going to read (if you attempt to read this entire thread), only one had any merit.  That argument came from Neatoman.  He put forth my Mutant data was 'tainted' because my initial gameplay in it was so bad, and that I'd gone away from the tank and got much better at playing, then came back to the tank.  So XVM was seeing/using my old crap stats to calculate chance to win, but I was a better player than my stats.  So I had 'xvm camo'.  Well, I've gone back and added results up to 2,000 battles past the original 112 crap battles.  MORE than enough to erase any camo...

-

Below is the updated battle results for my Mutant.  I have added the next 817 battles I played.  This brings the total to 2,000 battles.  These battles occurred after a long break in gameplay of the mutant.  I had pre-ordered the tank before the games release back in 2011.  I had played 2 weeks of beta before release.  So I jumped into my new shiny tier 8 and thought I would be invincible.  Well 112 battles later, I had a 32% win rate.  Took a long break from this tank.  Went back and played mostly tier 2 tanks for quite some time.  Then in 2016 started driving the mutant for fun. Of course I had gotten much better over time.  Which neato likes to call 'xvm camo'.  His theory was that xvm saw the old stats from when I was a crappy player and thus 'falsified' my actual chance to win.  I agree with him, at least for maybe the next 100-200 battles.  By then I feel xvm was correctly assessing my current skill.  He seamed to think 1,183 battles weren't enough to shake off the camo.  So now I've finished inputting the data for 2,000 battles beyond the initial 112.

-

This, and the original, data clearly shows that sbmm would not have any effect on win rate.  How so you ask?  Well the difference between the current mm and sbmm is the outlier chance to win battles.  Battles with chance to win 0-39%, and 61-100%.  Current MM and sbmm perform exactly the same for chance to win 40-59%, as both provide balanced teams.  So my point was that if sbmm eliminated the crappy outlier games, your win rate would NOT go to 50%.  In my initial data, my win rate for 40-49% was 51%.  So I was outperforming this range.  for 50-50%, my win rate was 63%, definitely outperforming at this range.  But  Neato cried 'nay nay', xvm was underrating me because of the early games.  His position was if xvm was correctly valuing my skill, I COULD NOT outperform chance to win in the 40-59% groups.  Well:

-

-

Notice my average overall chance to win went from 49.88 to 50.68.  No difference.

-

Notice my win rate for 40-49% went from 51.04 to 48.51.  A small dip, less than 1%.

-

Notice my win rate for 50-59% went from 63.74 to 64.40.  A small bump, less than 1%.

-

So EVEN WHEN XVM WAS CORRETLY ACCESSING MY SKILL, I WAS STILL ABLE TO PLAY ABOVE 50%. (in the 40-59% range).

-

So, if we have sbmm, and it only creates balanced teams (between 40-59% chance to win), a player can still be better than 50%.  This is obvious (or should be), because I'm running 56.3% win rate, there just has to be some scrub running 43.7% win rate.  And the holy holy statistics are kept in balance.  Are you telling me there aren't plenty of people running that win rate????

-

So the results above were all done under the current random mm (duh).  What most fail to see/acknowledge is that random mm creates balanced matches.  These are all the battles in the 40-49% and 50-59% brackets.  Both those brackets were battles where both teams had 'about' the same number of good/average/bad players.  This is exactly, and only, what sbmm would do.  Sbmm would simply eliminate the very low/very high chance to win battles, and EVERY battle would fall between 40-59%.

-

So we can look at my results in just the 40-59% bracket and it's an exact model of sbmm.   Some will claim that if all battles were balanced, you could only have a 50% win rate.  Let's do the math.  (538) 40-49% battles + (604) 50-59% battles = 1,142 balanced battles.  I won (261) 40-49% + (389) 50-59% = 650 battles.  This is a (650/1,142) 56.9% win rate.  You can have a better than 50% win rate with sbmm.  It's a fact.

-

Additional discussion on the individual brackets.

-

0-19% and 80-100%.  There are almost no battles in these categories.  The fact that sbmm would eliminate these is of no value.

-

20-39% and 61-79%.  There were 821 of these battles, or 41% of all battles.  The low battles are pure frustration.  As getting saddled with 10 red shtters is just not fun (if the enemy team isn't saddled with 10 red shtters).  While the opposite end of the spectrum is equally bad, while not frustrating, it's boring as all get out.  Sbmm would eliminate these entirely, "and that's a good thing"...

-

40-59%.  Guess what all you 'I love random mm' folks?  40-59% is sbmm.  That's right, you are already playing sbmm, just only 60% of the time.  So no, sbmm will not 'break the game', 'chase better players away', or drive all players to 50%.  We all already play it for the bulk of the time.  BUT, 60% is a D- grade, practically an F.   WG can and must do better.

-

We all know this, balanced teams equals fair play.  Anyone with a shred of pride/integrity/sense of fair play wants to 'earn every win'.  Sbmm is the only way this can happen.  As the data clearly shows, the current MM is stacking the teams/rigging the games/creating un-level playing fields.  Don't let the trolls/stat wh*res/stat nassi's fool you otherwise.

-

WG has recently shown they can be very precise in balancing the mm.  They can now put exactly the same number of lights, TDs, Arty, heavies, and mediums, balanced for tiers - without any delay of battle creation.  They could easily add one more thing to balance, a simply 'second sort' of both teams to make sure the teams were about even.  Simple, and will make this game great again....


 


 


 



DiePanzerGeist #2 Posted Jul 24 2016 - 07:36

    Major

  • -Players-
  • 7211 battles
  • 5,495
  • [DHO6] DHO6
  • Member since:
    04-28-2014

View Postbikebudha1, on Jul 24 2016 - 01:32, said:

 

Hey all, it's time for another of the budha's 'MM is rigged' posts.

-

So in this post I'm going to:

Explain what 'Rigged MM' really means (it's not what most of you think).

Provide documentation as proof.

Explain how mm rigging works.

Show you how to beat it.

Here we go...

-

Spoiler


 


 


 

 



Windows_Are_Tasty #3 Posted Jul 24 2016 - 07:40

    Major

  • Players
  • 58671 battles
  • 3,909
  • [-WLU-] -WLU-
  • Member since:
    04-15-2011
Me thinks OP has an IQ and shoe size that are about the same number...

saru_richard #4 Posted Jul 24 2016 - 07:43

    Major

  • -Players-
  • 26838 battles
  • 2,239
  • [SUX] SUX
  • Member since:
    08-19-2015
looks like MM post number 1341431431234 has nothing original about it... NEXT!!!

abandoned_spirit #5 Posted Jul 24 2016 - 08:07

    Major

  • Players
  • 26628 battles
  • 2,481
  • Member since:
    08-19-2012

ooooooo

so thats what it looks like

when hamsters try to do math

cool!



SwedishEOD #6 Posted Jul 24 2016 - 08:12

    Captain

  • Beta Testers
  • 61180 battles
  • 1,662
  • Member since:
    07-24-2010
If your replays you have embedded in your post is an accurate depiction of your actual gameplay and graphics settings i've seen enough.

aswitz87 #7 Posted Jul 24 2016 - 08:14

    Major

  • Players
  • 5921 battles
  • 6,974
  • Member since:
    06-10-2012

Note: I will not comment like the above posters, just dismissing because they can.  Wastes everyone's time scrolling.

 

Problem, you are drawing conclusions on samples sizes significantly too small.

Yeah in the short run MM will have a decent effect on a persons WR.  So will having a cold.  But a person cannot reasonably blame a cold on years worth of play.
Likewise, MM is not responsible for significant differences in WR over a large sample.

 


Edited by aswitz87, Jul 24 2016 - 08:15.


abandoned_spirit #8 Posted Jul 24 2016 - 08:19

    Major

  • Players
  • 26628 battles
  • 2,481
  • Member since:
    08-19-2012

View Postaswitz87, on Jul 24 2016 - 07:14, said:

Note: I will not comment like the above posters, just dismissing because they can.  Wastes everyone's time scrolling.

 

Problem, you are drawing conclusions on samples sizes significantly too small.

Yeah in the short run MM will have a decent effect on a persons WR.  So will having a cold.  But a person cannot reasonably blame a cold on years worth of play.
Likewise, MM is not responsible for significant differences in WR over a large sample.

 

The statistics he is pulling are also completely meaningless 

xvm chance to win % pulls overall stats to do calculations and overall stats mean essentially nothing.

Not to mention OP is biased, and is only pulling the statistics from his own games which on its own would completely invalidate any test process by itself.

and if you have seen any of his other hilariously clueless threads you too would dismiss this as anything other than hilarious. 



SoTrue #9 Posted Jul 24 2016 - 08:21

    Major

  • Players
  • 33313 battles
  • 3,302
  • [ATTOP] ATTOP
  • Member since:
    04-01-2011

View Postaswitz87, on Jul 23 2016 - 23:14, said:

Note: I will not comment like the above posters, just dismissing because they can.  Wastes everyone's time scrolling.

 

Problem, you are drawing conclusions on samples sizes significantly too small.

Yeah in the short run MM will have a decent effect on a persons WR.  So will having a cold.  But a person cannot reasonably blame a cold on years worth of play.
Likewise, MM is not responsible for significant differences in WR over a large sample.

 

 

+1 for an actual comment.  But I'll push you on it.  What would you consider a 'reasonable' sample size?  I had tracked roughly 1,200 battles (all in the same tank).  My opinion is that is a pretty decent amount.  If you could see the spreadsheet on those 1,200 battles, you would see the pattern shown in previous 5-days worth of battles.  Several days of 55% win rate - ish.  followed by a horrible day where I couldn't buy a win.  Rinse and repeat.  I know 'sample size' is a common talking point for the MM supporters.  But really, since most people only play 10-20 battles at a time.  Really any repeatable trend over that few battles is relevant because that is the scale people experience the game at.

aswitz87 #10 Posted Jul 24 2016 - 08:22

    Major

  • Players
  • 5921 battles
  • 6,974
  • Member since:
    06-10-2012

View Postabandoned_spirit, on Jul 24 2016 - 02:19, said:

The statistics he is pulling are also completely meaningless 

xvm chance to win % pulls overall stats to do calculations and overall stats mean essentially nothing.

Not to mention OP is biased, and is only pulling the statistics from his own games which on its own would completely invalidate any test process.

and if you have seen any of his other hilariously clueless threads you too would dismiss this as anything other than hilarious. 

 

Not entirely.  XVM uses per tank statistics.  So depending on how long a person has played a tank, it can effect how dated the stats are.

And yes, I know he is biased.  So am I in the opposite direction.  Hence why I won't just dismiss, but point out specific flaws in the setup.  Worst case, he recollects and actual does prove what he set out to prove, and I'm wrong.  No big.


Well, worst case no one does anything but bellyache, but thats what we have already so it can only go up.

SoTrue #11 Posted Jul 24 2016 - 08:24

    Major

  • Players
  • 33313 battles
  • 3,302
  • [ATTOP] ATTOP
  • Member since:
    04-01-2011

View Postabandoned_spirit, on Jul 23 2016 - 23:19, said:

The statistics he is pulling are also completely meaningless

xvm chance to win % pulls overall stats to do calculations and overall stats mean essentially nothing.

Not to mention OP is biased, and is only pulling the statistics from his own games which on its own would completely invalidate any test process by itself.

and if you have seen any of his other hilariously clueless threads you too would dismiss this as anything other than hilarious.

 

How are the stats meaningless?

Why are overall stats meaningless?  Are you suggesting a player who's overall stats are good isn't very good?

Please feel free to document your gameplay and show how your battle results differ.  Oh, like anyone has the time to gather and document thousands of players battle results.

And you have weird sense of humor if you find my posts hilarious.

-

Really, man up and track your next 100 battles in a tank you enjoy playing.  Preferably solo random pub matches to eliminate the platoon effect.  Will be waiting...



aswitz87 #12 Posted Jul 24 2016 - 08:36

    Major

  • Players
  • 5921 battles
  • 6,974
  • Member since:
    06-10-2012

View Postbikebudha1, on Jul 24 2016 - 02:21, said:

 

+1 for an actual comment.  But I'll push you on it.  What would you consider a 'reasonable' sample size?  I had tracked roughly 1,200 battles (all in the same tank).  My opinion is that is a pretty decent amount.  If you could see the spreadsheet on those 1,200 battles, you would see the pattern shown in previous 5-days worth of battles.  Several days of 55% win rate - ish.  followed by a horrible day where I couldn't buy a win.  Rinse and repeat.  I know 'sample size' is a common talking point for the MM supporters.  But really, since most people only play 10-20 battles at a time.  Really any repeatable trend over that few battles is relevant because that is the scale people experience the game at.

 

I wish I could find the post.  Someone did a nice explanation of how many matches one would need to definitively say someone was better than another player, based on win percent; the point where luck is factored out.

Anyway, if there is a 1% difference in WRs 10000 battles would be needed to be reasonably sure that player A is better than player B.  If it is a 5% difference is in the low to mid thousands, and a 10% difference would be a few hundred.
Now this doesn't say HOW much better the player is, just that he is.

How long does it take for luck to be removed as a significant factor?  10000 battles or so.  Is that reasonable for video game data collection from a single player?  Not really.
I'd say about 3 thousand would be sufficient to begin to see the bell curve.  Ballparking this though.
Also interval choice is important.  If your data runs from 40-60 and you use 10 as your interval, you likely will have a lot of trouble seeing anything useful.

As for tank choice: I would think sticking to a single tank could take longer for the bell curve to show up.  This is because you then rely on MM having a healthy player pool in one specific tiering segment, which is made worse by preferred MM tanks.  I could be wrong on that though.

And as for patterns, of course we (we being people, not any specific group) see them.  We see them even when they are not there.  That is why tests actually exist to check for patterns (and they even tell us how strong the pattern is).  I don't remember them well from Stats class a decade ago and would have to look them up.

Edited by aswitz87, Jul 24 2016 - 08:38.


abandoned_spirit #13 Posted Jul 24 2016 - 08:36

    Major

  • Players
  • 26628 battles
  • 2,481
  • Member since:
    08-19-2012

View Postbikebudha1, on Jul 24 2016 - 07:24, said:

 

How are the stats meaningless?

Why are overall stats meaningless?  Are you suggesting a player who's overall stats are good isn't very good?

Please feel free to document your gameplay and show how your battle results differ.  Oh, like anyone has the time to gather and document thousands of players battle results.

And you have weird sense of humor if you find my posts hilarious.

-

Really, man up and track your next 100 battles in a tank you enjoy playing.  Preferably solo random pub matches to eliminate the platoon effect.  Will be waiting...

 

let me give a very simple situation where overalls mean nothing

My overall rates me at about borderline light purple/dark blue

XVM would rate me at the same level someone who has platued and played at borderline light purple/dark blue for long periods of time and has not improved past it. In an actual fight though, there is an excellent chance I will wipe the floor with the other players face without much issue. And even if that was taken into account. There is an endless amount of factors that aren't. XVM chance to win is essentially worthless. Of no more value than a quick glace at the players names for stupid names would be.

and I find your post hilarious because its meaningless drivel followed by claims of value. " 'Rigged MM', proven "

If I go pull a bunch of numbers out of the stock market, toss them on a graph and say "this proves global warming is a conspiracy" I would have about as much validity to my statement as the original post here does.

 



kebab6597 #14 Posted Jul 24 2016 - 08:44

    Major

  • Players
  • 58310 battles
  • 9,107
  • [D-DAY] D-DAY
  • Member since:
    04-16-2011

So in a nut shell OP what your basically saying with all the research and evidence you have collected is exactly what one of the Russian devs went on video record saying about 3 years ago about MM manipulating players win/loss ratio to keep the players interested 

 

I just wish i could find that video to back up what i have just said perhaps someone else can recall it and post it here  



SoTrue #15 Posted Jul 24 2016 - 08:57

    Major

  • Players
  • 33313 battles
  • 3,302
  • [ATTOP] ATTOP
  • Member since:
    04-01-2011

View Postabandoned_spirit, on Jul 23 2016 - 23:36, said:

 

let me give a very simple situation where overalls mean nothing

My overall rates me at about borderline light purple/dark blue

XVM would rate me at the same level someone who has platued and played at borderline light purple/dark blue for long periods of time and has not improved past it. In an actual fight though, there is an excellent chance I will wipe the floor with the other players face without much issue. And even if that was taken into account. There is an endless amount of factors that aren't. XVM chance to win is essentially worthless. Of no more value than a quick glace at the players names for stupid names would be.

and I find your post hilarious because its meaningless drivel followed by claims of value. " 'Rigged MM', proven "

If I go pull a bunch of numbers out of the stock market, toss them on a graph and say "this proves global warming is a conspiracy" I would have about as much validity to my statement as the original post here does.

 

 

Ok a real response, thank you.  I agree, overall win rate 'can' be meaningless when looking at two specific individuals as in your scenario.  But be real.   When you see 12 tomatoes on your team, and the enemy only has 4 tomatoes, and xvm says your win chance is 28%.  Do you really have any doubt that your team will lose?  (Or flip it, you're on the team with only 4 tomatoes, and the enemy team has 12 tomatoes, and win chance is 72%, is there any doubt you're going to win?).  XVM is generally pretty accurate over time.  Espeicially as you move away from the 50% chance to win.  I.e. it is much more accurate as chance to win gets very high or very low - at least to the extent that you will lose 95% of all your 20%-29% battles, etc.

-

How is my data drivel?  I've tracked over a thousand battles in the same tank, over a couple of months of gameplay.  The data clearly shows trends and patterns.  All I'm really doing is coming up with a hypothesis to explain these trends and patterns.  Your stock market, global warming comment really doesn't make any sense.  Please, track your next 100 battles and get back to us.



SoTrue #16 Posted Jul 24 2016 - 09:00

    Major

  • Players
  • 33313 battles
  • 3,302
  • [ATTOP] ATTOP
  • Member since:
    04-01-2011

View Postkebab6597, on Jul 23 2016 - 23:44, said:

So in a nut shell OP what your basically saying with all the research and evidence you have collected is exactly what one of the Russian devs went on video record saying about 3 years ago about MM manipulating players win/loss ratio to keep the players interested

 

I just wish i could find that video to back up what i have just said perhaps someone else can recall it and post it here

 

I wish you could find that video too.  But to the main thrust of my point.  The current MM does not factor in player skill, the current MM routinely creates pre-determined battles (wins and losses), it's a source of frustrating for a large portion of the player base.  I'd like to see them fix it by incorporating skill based MM.  All I want is for every battle to fall withing 45%-55% chance to win - you know, every battle to occur on a level playing field.  I don't mind getting beat by a team of equal skill, what made the skunk battle tolerable was it was a 46% chance to win battle.  We had a chance, but we blew it.  Had it been a 15% chance to win, that would have been extremely frustrating, as we wouldn't have had any chance.

abandoned_spirit #17 Posted Jul 24 2016 - 09:18

    Major

  • Players
  • 26628 battles
  • 2,481
  • Member since:
    08-19-2012

View Postbikebudha1, on Jul 24 2016 - 07:57, said:

 

Ok a real response, thank you.  I agree, overall win rate 'can' be meaningless when looking at two specific individuals as in your scenario.  But be real.   When you see 12 tomatoes on your team, and the enemy only has 4 tomatoes, and xvm says your win chance is 28%.  Do you really have any doubt that your team will lose?  (Or flip it, you're on the team with only 4 tomatoes, and the enemy team has 12 tomatoes, and win chance is 72%, is there any doubt you're going to win?).  XVM is generally pretty accurate over time.  Espeicially as you move away from the 50% chance to win.  I.e. it is much more accurate as chance to win gets very high or very low - at least to the extent that you will lose 95% of all your 20%-29% battles, etc.

-

How is my data drivel?  I've tracked over a thousand battles in the same tank, over a couple of months of gameplay.  The data clearly shows trends and patterns.  All I'm really doing is coming up with a hypothesis to explain these trends and patterns.  Your stock market, global warming comment really doesn't make any sense.  Please, track your next 100 battles and get back to us.

if my team is all tomatos, it means nothing. As long as they dont all die in the first 2 min (and I can do things to prevent that in most tanks) I can play off them and win regardless.

And how do you hope to use my games to prove anything? I could toss my last 100 battles in my abominations at you in a zip but I doubt it would do you any good in proving that against all reason, the programming team that struggled to build a functioning match maker went through the trouble to build a far more complex system to rig the games against players that works.



Ptchart #18 Posted Jul 24 2016 - 10:21

    Corporal

  • Players
  • 34754 battles
  • 36
  • Member since:
    03-09-2014
You know when you played 250 games and have 42 % win rate on T34 and usually being top with damage done something is wrong with game. Take any tank you want  any win is a struggle. No matter you can make top damage can get  high caliber you still are going to loose. Already uninstalled this rigged crapgame. And no not giving anyone mine account. And lol I like these posts  I can win no matter what. How can you win no matter what if your team keep dieing faster than you can shoot. First 20seconds in game and already someone managed to get killed.

Edited by Ptchart, Jul 24 2016 - 10:24.


Panzer_74 #19 Posted Jul 24 2016 - 10:23

    Staff sergeant

  • Players
  • 22039 battles
  • 488
  • Member since:
    06-23-2012

What do you hope to achieve by posting these threads every couple of weeks? They are repetitive and monotonous.

You have a theory and no one is going to dissuade you. Good for you.

Who do you want to convince? Wargaming does not give a crapabout your opinion even if it was valid.

This is a company that cannot program MM to prevent 10 vs 5 top tier match ups, and you think they write code complex enough to do what you are suggesting... 



xXx_Dead_Zombie_xXx #20 Posted Jul 24 2016 - 10:27

    Major

  • Players
  • 41405 battles
  • 3,644
  • [SUX] SUX
  • Member since:
    10-09-2011

I find it comical people are actually debating this topic... 

 

Just do like i started to do, sell a tank after 2 consecutive losses in it. once you have them all sold. move on to a better game then this game and its MM, that is broken.

 

WG wonders why the game is dying.. only the people who sink all their money into it, will claim it works perfect.;. and CW 3.0 can suck a giant one.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users