Jump to content


1 tank 50 games: Tier 8 MT MM analysis + extra analysis


  • Please log in to reply
46 replies to this topic

struth #21 Posted 11 August 2018 - 08:48 AM

    Crikey

  • Council of Armored Forces
  • 44410 battles
  • 2,046
  • [-1AR-] -1AR-
  • Member since:
    10-17-2013

Is it a trash tank Jordy?

Just wondering because what I have seen of them they don't look remotely OP.

 

Currently doing 100 games in a tier 6 grind, first for me in ages,

running at

50% bottom tier 3/5/7

30% top tier 3/5/7

10% Mid tier 3/5/7

10% bottom tier 2 tier game.

 

Tier 6 tanks in tier 8 games can be a bit rough when say a Skorp G or tier 8 arty can one shot you.

Then try penning a tier 8 HT in a tier 6 MT without prem rounds.



Bash_sh #22 Posted 11 August 2018 - 10:07 AM

    First lieutenant

  • Member
  • 26965 battles
  • 947
  • [PBKAC] PBKAC
  • Member since:
    05-25-2013

View PostFramFramson, on 11 August 2018 - 10:08 AM, said:

I would be really interested if there were some way to take a random sampling of a reasonable number of players  (say 100+) to see the length and frequency of their loss streaks as compared to their win streaks over a period long enough to provide good data (say 1000 matches?). I think the results would be quite revealing.

 

been done

there is stats somewhere in the internet that shows you the bell curve of win rate. the average win-rate of players are in the 48% range



mttspiii #23 Posted 11 August 2018 - 10:48 AM

    Major

  • Beta-Tester
  • 29193 battles
  • 16,589
  • [PVP] PVP
  • Member since:
    04-15-2012

View PostFramFramson, on 11 August 2018 - 06:08 AM, said:

I would be really interested if there were some way to take a random sampling of a reasonable number of players  (say 100+) to see the length and frequency of their loss streaks as compared to their win streaks over a period long enough to provide good data (say 1000 matches?). I think the results would be quite revealing.

 

This may be anecdotal, but my gameplay is usually to clear x2's on all my grind tanks. Win a battle in a tank, move to next, and when I've cleared all x2's, stop WoT.

I do get 6-battle losing streaks sometimes with tanks that "just don't want to win today", which is very frustrating, until I bother to look at the tank's WR and notice that my tank's WR didn't tank as badly as I thought, despite having played not so many battles in them.

 

Means that I had a prior winning streak in that tank, but I didn't notice it because to me I was just clearing the x2's, and that my winning streak spread out at 1 win per battle, a day, for a week. A feat of incognition on my part I guess.


I'm fierce and I'm feeling mighty,

I'm a golden girl, I'm an Aphrodite

 

 


FramFramson #24 Posted 11 August 2018 - 10:51 AM

    Major

  • Member
  • 37937 battles
  • 3,237
  • Member since:
    02-22-2015

View PostBash_sh, on 10 August 2018 - 09:07 PM, said:

 

been done

there is stats somewhere in the internet that shows you the bell curve of win rate. the average win-rate of players are in the 48% range

 

Not the win rate. Streak frequency and duration. (for this purpose, I would call a streak anything above say... 3 games?)

LT-playing masochist. It's too much fun to be a mosquito.


FramFramson #25 Posted 11 August 2018 - 10:52 AM

    Major

  • Member
  • 37937 battles
  • 3,237
  • Member since:
    02-22-2015

View PostJordyBro, on 10 August 2018 - 06:46 PM, said:

I have had a similar experience with the 50TP, first tier 8 I have played in ages in pubbies. How do people play 8s? If it is not trash matchmaking its 2-3 arty a game just completely ruining it

 

I play T8s all the time and I just keep wondering why I do that to myself.

 

The problem is that so many of my favourite tanks are T8s.


LT-playing masochist. It's too much fun to be a mosquito.


Matt_B #26 Posted 11 August 2018 - 03:40 PM

    Staff sergeant

  • Member
  • 17225 battles
  • 381
  • Member since:
    10-09-2015

View Postmttspiii, on 11 August 2018 - 03:44 AM, said:

 

https://www.popsci.com/article/science/are-lucky-streaks-real-science-says-yes

 

TL;DR:

When you start winning, your mind starts assuming that your next game will be a loss, so to compensate you start to play safer, and thus by playing safer you are more likely to win your next battle.

When you start losing, your mind starts assuming that your next game will be a win, so you get frustrated/un-mojo-ed and play sloppier, and thus by playing sloppier you are more likely to lose your next battle.

 

Do note that this is more an issue about cognition, so ironically this means that the less you notice (or care) for your winning or losing streak, the less likely you are going to get a winning or losing streak in the first place. The subconscious can be a jerk like that.

 

That's an interesting read, and the phenomenon sounds a bit like tilt, which I'm sure we've all experience at some time even if we couldn't put any percentages to it.

 

That said, with WOT pubbies being the 15 vs 15 slugfests that they are, I'd think that it's going to be pretty hard to drive your short term win percentage down much below 30, because you'll eventually end up on a side that can't lose no matter how bad you're playing. And let's face it, even bots and AFKs are going to tend to around 40. You'd pretty much have to be griefing your teammates to do worse for any length of time.

 

As such, I'd still think that your chances of having a losing streak that extends beyond the lower end of double figures is going to be rather slim.



neokai #27 Posted 12 August 2018 - 04:00 PM

    SealClub Number 9

  • Senior Moderator
  • 38174 battles
  • 8,090
  • Member since:
    06-07-2013

View PostArista_Skyfall, on 10 August 2018 - 02:58 PM, said:

The results or the prediction? the prediction would change, yes, but it has no affect on the results

 

Part of hypothesis testing is that you can predict future behaviour based on your hypothetical model. Since you said you have almost 100% accuracy, tell me what are your key predictors (e.g. tier values, team composition, XVM values) and I can draft you a double blind experiment with a fairly robust dataset. Then we can run your predictive model and test its veracity.

 

View PostJordyBro, on 11 August 2018 - 07:46 AM, said:

I have had a similar experience with the 50TP, first tier 8 I have played in ages in pubbies. How do people play 8s? If it is not trash matchmaking its 2-3 arty a game just completely ruining it

 

With care and aggression in equal parts. Over commit and you get destroyed, but hang back too long and you might miss the breakthrough moments.

 

View PostFramFramson, on 11 August 2018 - 10:51 AM, said:

 

Not the win rate. Streak frequency and duration. (for this purpose, I would call a streak anything above say... 3 games?)

 

You will need to draft a new set of filters and sift through individual player history to compile the streaks.

 

View PostMatt_B, on 11 August 2018 - 03:40 PM, said:

 

That's an interesting read, and the phenomenon sounds a bit like tilt, which I'm sure we've all experience at some time even if we couldn't put any percentages to it.

 

That said, with WOT pubbies being the 15 vs 15 slugfests that they are, I'd think that it's going to be pretty hard to drive your short term win percentage down much below 30, because you'll eventually end up on a side that can't lose no matter how bad you're playing. And let's face it, even bots and AFKs are going to tend to around 40. You'd pretty much have to be griefing your teammates to do worse for any length of time.

 

As such, I'd still think that your chances of having a losing streak that extends beyond the lower end of double figures is going to be rather slim.

 

Pretty much, though tilt focuses entirely on the negative spiral. mttspii's link claims there is a positive corollary where your winning ways positively reinforce your mindset which in turn leads to more wins. Not sure how true that is, but I do believe that improved focus from being relaxed (rather than stressed from losing and "bad" teammates) improves the chances of winning.


Spoiler

Aoyama_Blue_Mountain #28 Posted 12 August 2018 - 07:26 PM

    Is the order a blue mountain?

  • Beta-Tester
  • 68609 battles
  • 7,220
  • [WAT] WAT
  • Member since:
    05-30-2012

However reasons based on personal performance cannot work in a game where 80% of the games are not decided by you and a bot can get 40% winrate.

 

Also, using simple random probability says that the probability of losing 10 games in a row is 2^10 = 1/1024 and hence it should appear once out of 1000 games. However this is seldom observed. I just got another 9-loss streak after yesterday's dunno how many loss streak. which followed after the 10-loss one I posted in YouTube. This afternoon's session stands at 9 / 27 ;)

 

Because we're talking about a small number of samples, a Poisson distribution is more appropriate. And λ is pretty large in this case.

 

Either accept the 20-battle rule (or rather, based on the battles this weak, we're indeed looking at 50-battle rule) combined with winrate compression from MM / meta changes, meaning that outcome is random and we get these streaks because of huge variation, or we would be admitting that outcome is not random despite having only 20% correlation to personal skill, meaning skill-based MM exists.



Mother_Of_All_Rommel #29 Posted 12 August 2018 - 07:49 PM

    Destroyer of WoT

  • Member
  • 20783 battles
  • 2,703
  • Member since:
    12-08-2014
If WG really make sbmm, they should make it public. Lots of ppl would probably love it. No reason to keep it in the dark.

Edited by Mother_Of_All_Rommel, 12 August 2018 - 08:08 PM.

THE ONLY REAL POWER COMES OUT OF A LONG RIFLE


JordyBro #30 Posted 12 August 2018 - 08:16 PM

    Captain

  • Council of Armored Forces
  • 23483 battles
  • 1,178
  • [MYTH] MYTH
  • Member since:
    12-25-2012

View Postneokai, on 12 August 2018 - 08:00 AM, said:

 

With care and aggression in equal parts. Over commit and you get destroyed, but hang back too long and you might miss the breakthrough moments.

 

Would be better if the people in charge or improving/running the game had half a brain and listened to the overwhelming feedback from the player base regarding arty


 


Aoyama_Blue_Mountain #31 Posted 12 August 2018 - 08:31 PM

    Is the order a blue mountain?

  • Beta-Tester
  • 68609 battles
  • 7,220
  • [WAT] WAT
  • Member since:
    05-30-2012

Because SBMM penalizes one group of players for the benefit of the other. Take the type that causes winrate compression for example, it will cause people with >50% winrate to start losing more, but it is always being suggested by <49%-ers for the same reason.

 

However WG would need ways to get 40%-ers to continue paying playing, so they need ways to compress the winrate - We already have multi tier MM and 15v15 for this. There is little benefit to WG to reveal that good players are getting their winrate lowered while bad players are getting more wins not because of their skill.



AlexTheKid72 #32 Posted 13 August 2018 - 05:13 AM

    Corporal

  • Member
  • 14639 battles
  • 35
  • [REIGN] REIGN
  • Member since:
    05-28-2017

SBMM doesn't need to be 15 55%ers vs 15 55%ers.

I would like to see equal numbers of purples, blues, greens on each team though. The number of whitewash battles I see with equal skill distribution is minimal compared to the blowouts where one team has many more +50% players than the other. Those games end blowouts.

How hard can it be for mm to go "ok we have 5 heavies 5 td's 5 mediums each team ..... let's quickly do a "captains pick sides" for our customers, and have the best heavy 1 team, 2nd best other, third best 1st team, 4th best other etc? It worked at school for P.E. team picking ........


The first 9000 were a struggle ... I have fun now.

Matt_B #33 Posted 13 August 2018 - 06:43 AM

    Staff sergeant

  • Member
  • 17225 battles
  • 381
  • Member since:
    10-09-2015
If you want SBMM there's Ranked Battles mode. It doesn't seem too popular at the moment though.

JordyBro #34 Posted 13 August 2018 - 07:20 AM

    Captain

  • Council of Armored Forces
  • 23483 battles
  • 1,178
  • [MYTH] MYTH
  • Member since:
    12-25-2012

View PostMatt_B, on 12 August 2018 - 10:43 PM, said:

If you want SBMM there's Ranked Battles mode. It doesn't seem too popular at the moment though.

 

That isn't why it's unpopular.. ranked is [Content Removed] due to the mentality players are adopting to maintain/gain ranks. 

 

I used to be an advocate for SBMM, however I think just fixing up the tiers you face, tier 8 do not fight tier 10. I do not play lower tiers but I assume 7s coming up against 9s would be similar in it's issues. 

Inappropriate use of Word , Post Edited , User Sanctioned
~Ephys

 


Edited by Ephys, 14 August 2018 - 01:10 PM.

 


Angmar_ #35 Posted 13 August 2018 - 08:41 AM

    Corporal

  • Member
  • 17060 battles
  • 33
  • [LEFT] LEFT
  • Member since:
    12-28-2014
Tier 7 v tier 9 doesn't feel as bad to me as tier 8 v tier 10, maybe that's just me though

Aoyama_Blue_Mountain #36 Posted 13 August 2018 - 11:47 AM

    Is the order a blue mountain?

  • Beta-Tester
  • 68609 battles
  • 7,220
  • [WAT] WAT
  • Member since:
    05-30-2012
RB fails because it doesn't reward playing to win.


mttspiii #37 Posted 13 August 2018 - 12:04 PM

    Major

  • Beta-Tester
  • 29193 battles
  • 16,589
  • [PVP] PVP
  • Member since:
    04-15-2012

View PostAoyama_Blue_Mountain, on 12 August 2018 - 07:26 PM, said:

Either accept the 20-battle rule (or rather, based on the battles this weak, we're indeed looking at 50-battle rule) combined with winrate compression from MM / meta changes, meaning that outcome is random and we get these streaks because of huge variation, or we would be admitting that outcome is not random despite having only 20% correlation to personal skill, meaning skill-based MM exists.

 

If only we could compare your results with games with similar online games that pit X players in 2 teams against each other like Gunbound or even MOBA's, then check for streak lengths and durations, and check for MM differences.

 

Maybe, just maybe, WG's using an old-fashioned RNG that can't really do random very well.


I'm fierce and I'm feeling mighty,

I'm a golden girl, I'm an Aphrodite

 

 


FramFramson #38 Posted 13 August 2018 - 02:41 PM

    Major

  • Member
  • 37937 battles
  • 3,237
  • Member since:
    02-22-2015

View PostAoyama_Blue_Mountain, on 12 August 2018 - 10:47 PM, said:

RB fails because it doesn't reward playing to win.

 

RB fails because it doesn't reward playing at all. Last time I checked the only way to be sure of ANY reward worth the time, you had to be friggin' Rank 9. 

 

If it gave bonds to the members of the winning team at the same rate as regular random all TX-battles, it would be absolutely fine, but unless you're in the top 25% of the server, it's pretty much a complete waste of time.



Angmar_ #39 Posted 13 August 2018 - 03:49 PM

    Corporal

  • Member
  • 17060 battles
  • 33
  • [LEFT] LEFT
  • Member since:
    12-28-2014
Yeah but the rate you get bonds at is rubbish anyway, not that I think they should exist :|

FramFramson #40 Posted 14 August 2018 - 01:15 AM

    Major

  • Member
  • 37937 battles
  • 3,237
  • Member since:
    02-22-2015

View PostAngmar_, on 13 August 2018 - 02:49 AM, said:

Yeah but the rate you get bonds at is rubbish anyway, not that I think they should exist :|

 

Sure, but it's better in regular TX randoms than it is in Ranked - which is [edited]! Should be the same - or MORE - if they want people playing Ranked.

LT-playing masochist. It's too much fun to be a mosquito.





1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users