Winning is impossible. :: World of Tanks Blitz General Discussions

anytime Nov 27, 2024 @ 8:44am

Originally posted by Neithan:
Originally posted by milan.mrlian:
ask Neithan how build good team/ friends in 3 sec -decision and choice is problem-broke trust
ask him why ally camp in back (many high tier) and low tier fight in mid alone(tension line)?

Go play some singleplayer game, and you have no problems with random teammates. But If you play WoTB, GROW UP AND DEAL WITH IT.

Stop crying like a spoiled brat and man up, it is a GAME, one could get the impression your life depends on..

Here are 5 rating battles, I believe they were 2 in a row, then 3 in a row (last one says 4th battle, which it was, I don't think I SS'd the first battle). Do you see any pattern Neithan?

https://steamcommunity.com/sharedfiles/filedetails/?id=3371035171
https://steamcommunity.com/sharedfiles/filedetails/?id=3371045410
https://steamcommunity.com/sharedfiles/filedetails/?id=3372114882
https://steamcommunity.com/sharedfiles/filedetails/?id=3372116738
https://steamcommunity.com/sharedfiles/filedetails/?id=3372120774

#16

EL_Din_46

Nov 27, 2024 @ 12:16pm

Although the ratings mm matches players by rating points, uncalibrated players are matched based on their regular battle winrate. I haven't seen any mention of balancing uncalibrated players on the two teams. As in regular battles the players would be randomly selected and then tested for whether their tank and rating points/winrate balanced out. Because of this random factor there could be streaks of more or less uncalibrated players on a team.

Whether this is a good or bad design is a different question.

#17

Neithan

Nov 27, 2024 @ 2:33pm

Originally posted by anytime:

Here are 5 rating battles, I believe they were 2 in a row, then 3 in a row (last one says 4th battle, which it was, I don't think I SS'd the first battle). Do you see any pattern Neithan?

*sigh*

ANY mathematician will laugh at your sample size. In small sizes, patterns are ALL OVER THE PLACE, but in a large sample size, patterns become more and more obscured and get lost in the randomness.

Again

I could throw a coin 1 million times, and the expected outcome would be roughly 500.000 tails/heads each. First of all, you would hardly EVER hit that exact numbers when you would throw 1 million coins a billion times.
But instead, I could take any 10 consecutive throws out of that pool, and I would end up with way worse numbers, I could even find a 10:0 streak. So all these sample sizes would be completely off, still the grand scale is that the bigger the sample size, the closer we get to the expected outcome.

Really, noone can help you not understanding randomness and probability...

#18

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 28, 2024 @ 11:10am

#19

Silent

Nov 28, 2024 @ 1:16pm

Originally posted by YASAMATA_HEROWATA:
It is everything fine with randomness and probability.
TS just points that it works slightly differently in different accounts.

https://steamcommunity.com/sharedfiles/filedetails/?id=3370250169
https://steamcommunity.com/sharedfiles/filedetails/?id=3369215591
https://steamcommunity.com/sharedfiles/filedetails/?id=3361207708
https://steamcommunity.com/sharedfiles/filedetails/?id=3359752965
https://steamcommunity.com/sharedfiles/filedetails/?id=3358440088
https://steamcommunity.com/sharedfiles/filedetails/?id=3355198646
https://steamcommunity.com/sharedfiles/filedetails/?id=3352599290
https://steamcommunity.com/sharedfiles/filedetails/?id=3351319468

and

https://steamcommunity.com/sharedfiles/filedetails/?id=3360427832
https://steamcommunity.com/sharedfiles/filedetails/?id=3359488029
https://steamcommunity.com/sharedfiles/filedetails/?id=3353514610
https://steamcommunity.com/sharedfiles/filedetails/?id=3351518186
https://steamcommunity.com/sharedfiles/filedetails/?id=3350682990

You play low tier and you have a sub 50% win rate while maintaining low damage per game. You played 48 games at tier 7 and managed 900 dmg per game average. That's 25% less per game than your own hit points.

The short of it is you are just playing poorly and games feel random because YOU are not influencing them.

If you can't even do 1 tank worth of damage per game you are obviously not benefitting your team very much.

Last edited by Silent; Nov 28, 2024 @ 1:17pm

#20

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 29, 2024 @ 11:46am

One probability theory expert said

So you get a team that, in pure numbers/statistics is inferior, and the "predetermined" winrate would be 40:60. It could still be YOUR single performance that could tip that scale to 45:55 or better. So you DO have an impact.

In other words, Some players were definitely more times in teams with predefined lower winrate.

you are just playing poorly and games feel random because YOU are not influencing them

Was I playing alone? Did other team member make influence? Are we playing with bots?

I had lost series in upper and middle tiers. It is probably this tanks were weak and bad.

X T57
X TVP T50/51
X Progetto 65
X AMX 50B
X T110E5

VIII Centurion Mk I
VIII Elefant
VIII Chrystler K
VIII Progetto 46
VIII Strv 81
VIII TSL-7

VII T42
VII T25E3
VII FV201
VII T-34-1
VII T-34-2G FT
VII IS-2
VII T-43

VI Skoda P-JS
VI Type 58

V Matilda BP
V Scavenger
V Medjay
V Spike

IV Pz B2
IV Gro?traktor

#21

anytime Nov 29, 2024 @ 12:43pm

Originally posted by Neithan:
Originally posted by anytime:

Here are 5 rating battles, I believe they were 2 in a row, then 3 in a row (last one says 4th battle, which it was, I don't think I SS'd the first battle). Do you see any pattern Neithan?

*sigh*

ANY mathematician will laugh at your sample size. In small sizes, patterns are ALL OVER THE PLACE, but in a large sample size, patterns become more and more obscured and get lost in the randomness.

Again

I could throw a coin 1 million times, and the expected outcome would be roughly 500.000 tails/heads each. First of all, you would hardly EVER hit that exact numbers when you would throw 1 million coins a billion times.
But instead, I could take any 10 consecutive throws out of that pool, and I would end up with way worse numbers, I could even find a 10:0 streak. So all these sample sizes would be completely off, still the grand scale is that the bigger the sample size, the closer we get to the expected outcome.

Really, noone can help you not understanding randomness and probability...

Nice troll. This has been something I have observed for years over time. Quite commonly, especially in rating battles where my team loses, the opposing team has 1 or more uncalibrated players, who usually are top 3.

LOl you're idiotic comment about sample size.... fing re****.

#22

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 29, 2024 @ 1:01pm

I knew it would be discussion with weird theories and funny explanations
https://steamcommunity.com/sharedfiles/filedetails/?id=3374079521

#23

Neithan

Nov 29, 2024 @ 6:15pm

Originally posted by anytime:
Originally posted by Neithan:

*sigh*

ANY mathematician will laugh at your sample size. In small sizes, patterns are ALL OVER THE PLACE, but in a large sample size, patterns become more and more obscured and get lost in the randomness.

Again

I could throw a coin 1 million times, and the expected outcome would be roughly 500.000 tails/heads each. First of all, you would hardly EVER hit that exact numbers when you would throw 1 million coins a billion times.
But instead, I could take any 10 consecutive throws out of that pool, and I would end up with way worse numbers, I could even find a 10:0 streak. So all these sample sizes would be completely off, still the grand scale is that the bigger the sample size, the closer we get to the expected outcome.

Really, noone can help you not understanding randomness and probability...

Nice troll. This has been something I have observed for years over time. Quite commonly, especially in rating battles where my team loses, the opposing team has 1 or more uncalibrated players, who usually are top 3.

LOl you're idiotic comment about sample size.... fing re****.

Well, you can think about probability math and all that whatever you wish, does not change the FACTS.

Again, how many coin-tosses (fair and unbiased) would you need to reliably reach the predicted 50:50 outcome? Everytime?

Thousands?

Millions?

Billions?

Just to give you a hint: On 100 coin-tosses, the probability of reaching exactly 50 heads and 50 tails is 7.96%. And the more times you toss the coin, the lower that number gets!

So, what the human mind thinks of the predictable outcome and thus files it under, is 'probability is "certain"/closes in on "1", while it actually has a probability of 0.0796.

Now, with a much-more complicated set of randomization, like we got here, how many tries do you need to reach a predicted outcome like 50:50?

In fact, you NEVER reach a guaranteed "probability" no matter the sample size, as it is RANDOM. You only close in. Probability of outcome is anywhere between 0 (impossible) and 1 (guaranteed), but you will NEVER reach any of both, you only get near them.

So even the best WoTB player on the world can NEVER reach a probability of 1 regarding the outcome of their battles. Neither would the worst ever reach probability 0.

in rating battles where my team loses

Yeah, right, exactly. You know what biased perception is? What about all the rating battles you won?
Right, you don't check on those, as you don't need no confirmation for being the better player, but if you lost, suddenly you get all Sherlock on what the team did wrong/how MM screwed you.

Last edited by Neithan; Nov 29, 2024 @ 7:01pm

#24

anytime Nov 29, 2024 @ 7:22pm

Hey Neithan...what do want me to do? Post every replay from every battle I do? I'll tell you what, IF YOU WILL WATCH THEM, I will POST THEM. So I won't bother posting them. I'll type a random phrase at some point in the video, in which you will have to recite as evidence of watching it. First time you fail, it's over. Dea;?

You're ...I dont' even know. Random on computers is not the same thing as random in "the real world". I'm a degreed software developer, with over 25 years of IT and software development experience. Implementation of mathematical formulas are only as valid as the specifications a developer is given.

When it comes to WG and "random", you can toss most mathematics out the window. This game isn't even physics based (it doesn't require, nor use physics acceleration), never mind the multitude of reasons it benefits WG to manipulate battles. Maybe the Unreal 5 version will be based on "real world physics" and have an actual physics engine. I don't know.

Let me tell you a story. When I was in college, I worked at a retailer...we'll just call it "BB". One thing I was trained SPECIFICALLY, WITH NO DOUBT 100% OF THE TIME, THEY DO NOT ACCEPT RETURNS FOR MUSIC (CDS) THAT HAS BEEN OPENED. So I'm near the office one day and I over hear a conversation between the service clerk and a manager...about what? A customer wanted to return a CD. The service desk person, she said the policy. It turned into a huge thing. And then the manager told me the rest of the story. He said, this customer spends $20,000 a month here at this store, buying for his business. And this really pissed him off. What would you do. Myself...I thought...well I know the policy..and I know the clerk is saying "the policy"...but I said "I would've let them return it." I'm not stupid.

What's the moral of the story? Money talks, BS walks. WG does not provide a game. WG sells a product. Old Smokie Joe can stand being a 40% player, and buying $100 crates every 2 weeks. Old Smokie Joe can't stand being a 10% win rate player, and will stop "playing" the game.

Among all the BS, I will promise that what WG does as far as the "game" are based heavily on mined data and what is most likely to encourage players to spend money. If that takes putting Old Smokie Joe on a "high probability" of winning team and putting some other players on a "low probability" winning team, they're going to do it if they need to. It's even in their patents, that players winning and losing may be tracked and used during the match making process.

#25

Neithan

Nov 29, 2024 @ 8:09pm

I really start to dig into ChatGPT..

Input:
imagine a skill scale 1-10. 1000 players. Ten at each skill level.* game is 7 vs 7. the team with the higher combined skill level wins. Player A at skill level 1 and player B at skill level 10 play 1000 matches each. The teams are randomized from the pool each time. what would be the winrate of player A and player B?

Originally posted by ChatGPT:
]To calculate the win rates of Player A (skill level 1) and Player B (skill level 10), we need to consider how the randomized team compositions affect the overall team skill levels. Here's a step-by-step approach:

### 1. **Assumptions**
- Each match involves random selection of 7 players per team from the pool of 1,000 players.
- The combined skill level of a team is the sum of the skill levels of its 7 players.
- The team with the higher combined skill level wins.
- In the case of a tie, assume an equal probability of winning for both teams.

### 2. **Expected Team Skill**
The average skill level of a randomly selected player from the pool is:

\[
\text{Average Skill} = \frac{1 \cdot 10 + 2 \cdot 10 + \dots + 10 \cdot 10}{1000} = \frac{5500}{1000} = 5.5
\]

For a team of 7 players, the **expected team skill** is:

\[
\text{Expected Team Skill} = 7 \times 5.5 = 38.5
\]

The skill levels of teams will vary around this average depending on the random composition of the teams.

### 3. **Impact of Player A and Player B**
#### Player A (skill level 1):
If Player A (skill level 1) is on a team, that team’s skill is slightly reduced compared to the average team. Specifically, Player A contributes 1 instead of the average skill of 5.5, so their team will tend to have a skill level:

\[
38.5 - 5.5 + 1 = 34
\]

#### Player B (skill level 10):
If Player B (skill level 10) is on a team, that team’s skill is slightly increased. Player B contributes 10 instead of the average skill of 5.5, so their team will tend to have a skill level:

\[
38.5 - 5.5 + 10 = 43
\]

### 4. **Randomization Across Matches**
Over 1,000 matches, Player A and Player B will each play with and against random teammates. This means that:
- **Player A's team** will tend to have a lower average skill than their opponents because their skill level is below the average.
- **Player B's team** will tend to have a higher average skill than their opponents because their skill level is above the average.

### 5. **Win Probability Approximation**
The win rate for Player A and Player B can be approximated using the difference in their team's expected skill level compared to their opponents. Assuming a normal distribution of team skills, the win probability can be estimated using the standard normal cumulative distribution function ($\Phi$):

\[
P(\text{win}) = \Phi\left(\frac{\text{Difference in Skill Levels}}{\text{Standard Deviation of Team Skills}}\right)
\]

The **difference in skill levels** due to Player A or B is:

- Player A: $-4.5$ (below average)
- Player B: $+4.5$ (above average)

The **standard deviation of team skill** can be computed from the variance in individual skills (details omitted for now, but let's assume it).

Let me compute these probabilities numerically.

The win probabilities for Player A and Player B are:

- **Player A (skill level 1):** ~27.7% win rate.
- **Player B (skill level 10):** ~72.3% win rate.

These probabilities reflect the impact of each player's skill level on their team's overall performance when playing 1,000 matches with randomly assigned teammates.

Let's change the skill level of player A and B to a more reasonable 3 and 8...
The simulated win rates are as follows:

Player A (skill level 3): 42.5%
Player B (skill level 8): 60.05%

---

Now, your winrate, as we have seen, is even above the "skill level 8" mark, but you still think they are screwing you with rigging matches, thus you should even have a higher winrate.

Does make you seem pretty entitled, eh?

I mean, you are already way above the average of player, but you STILL think they are screwing with you?

And pls, spare me "You need to be on top to see what is going wrong at the bottom" or whatever BS, it just doesn't add up.
If it was as you said, and WG is making you get bad teams on purpose, then you should actually be proud as heck to STILL achieve such a score. But the more realistic approach is to acknowledge that you are a really good player, but RANDOM TEAMS ARE RANDOM, and you can encounter all ♥♥♥♥, but if you see the whole (laaaaarge sample-size), it all fits in.

There is plenty of "experts" examining replays and statistics, datamining everything about WoT(B), and they are way more proficient about the matter than you or me, yet noone has found hard evidence for all the accusations around. All that comes up is "personal experience", meaning "extremely limited sample sizes from a bias pov". No matter how much you think you are good at observing patterns and deviation from "randomness", THEY ARE BETTER.
So if YOU see a pattern in their MM, why do THEY not report this?

Also, contrary to some belief here, WG are in fact liable for something like rigging matches, and the inevitable fallout from being proven to do so would be extreme. So they not only have to be evil but stupid as well, allthewhile they are pulling the most exquisit of scams undetected by everyone but some guys observing "patterns" in their matchmaking.

It's like 9/11 truth movement or similar (not wanna start a discussion, just make a point), where people think that the same incompetent government they complain about every day, where "internals" get leaked all the time, that screws up even the simplest matters, stumbles over their own feet the whole time, suddenly is capable of pulling the most complicated scheme, involving hundreds, if not thousands of persons, without leaving the slightest trace or a single person involved blowing it.
Common sense like what?

*[Edit] seems I screwed up the input, pool should be 100 players, or 100 at each skill level, but dunno how it screwed up the calculation as I reached the daily limit of free ChatGPT. Seriously thinking about going pro. ^^

Also, seems like Steam forum does not like some of the syntax ChatGPT uses, perhaps due to it being in a quote, will look into it.

Last edited by Neithan; Nov 29, 2024 @ 8:54pm

#26

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 30, 2024 @ 2:29am

some pictures about randomization

Originally posted by anytime:
Another rating battle, another BS "match making". Why doesn't MY TEAM have the uncalibrated player who did top damage? No...can't be...EVER.

Just wait a minute, it was mine (⊙ᴗ⊙)
https://steamcommunity.com/sharedfiles/filedetails/?id=3374503101
It seems very common and the randomization rules becomes now more clear

"uncalibrated player" did 7 shots with 4 damage and 3 kill in last fight
https://steamcommunity.com/sharedfiles/filedetails/?id=3374512491

"uncalibrated player" had even 7 kills in one fight
"uncalibrated player" is a clan member but didn't play any tournamen
"uncalibrated player" has 140 tanks and 45 tanks never played
https://steamcommunity.com/sharedfiles/filedetails/?id=3374529786

#27

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 30, 2024 @ 2:58am

some pictures about randomization
https://steamcommunity.com/sharedfiles/filedetails/?id=3374551175
It is a real killing machine with 103 fights and 22 tanks 5 newer played
https://steamcommunity.com/sharedfiles/filedetails/?id=3374551842

#28

YASAMATA_HEROWATA has World of Tanks Blitz

Nov 30, 2024 @ 4:10am

All three profile have some common parts
https://steamcommunity.com/sharedfiles/filedetails/?id=3374599666
https://steamcommunity.com/sharedfiles/filedetails/?id=3374512491
https://steamcommunity.com/sharedfiles/filedetails/?id=3374551842

#29

anytime Nov 30, 2024 @ 8:31am

Neithan I hope you're not too d*** to understand this. First off, again, AI answers your getting do not nearly impress me as much as they impress you. But much of that is the question being asked of it.

You keep stating "random". But we're not dealing with "random".

In all modes, teams are minimally balanced for tank types, and in some cases individual tanks (for example, Annihilator).

So what's left? Players? How are player chosen? First come first serve? FCFS is NOT random.

How are players chosen? In the example of battle with a platoon, either the match maker waits until another, and again, matching platoon minimally tier and also seemingly tank type (2 heavies might be acceptable vs 1 heavy and 1 TD, or a medium). Then all the other players also have to be factored into battle.

Once in a while, even individually, in an "experienced tank" (not "new tank matching mode") you might be in the queue for 3 minutes. And during that time, it's obvious matches are being built and sent to a game server. So if it's FCFS, then how am sitting here where I've watched 10-20 battles pass by? Certainly ONE of those battles I must've been a "fit" for, and certainly players entering the queue after "place" are getting battles. Certainly, not everyone in queue, has been waiting for me to get a battle.

The other day I was in a clan platoon. And we lost 4 in a row. I could tell, the matches were being manipulated by how long we were queued and how carefully it seemed match maker was being. Certainly, we we'rent the only platoon in queue. One battle there ended up being 2 platoons! So MM was waiting FOR SOMETHING. Exactly what? I don't know.

#30

World of Tanks Blitz

Report this post