Matchmaker Discussion Thread & MM Balance

[Aethervoxx]

1 minute ago, ColonelPete said:

WG does not need to roll dice for the MM.

The randomness come from the time players press the battle button.

ROFL. The 'non-randomness' comes after pressing the battle button & when you are waiting that variable amount of time while the MM 'arranges' the teams.

I'm surprised, Pete, you can't/haven't yet figure/d this out. You've played over 16k battles.

[ColonelPete]

8 minutes ago, Aethervoxx said:

ROFL. The 'non-randomness' comes after pressing the battle button & when you are waiting that variable amount of time while the MM 'arranges' the teams.

I'm surprised, Pete, you can't/haven't yet figure/d this out. You've played over 16k battles.

No. That is the time needed for the rest of the team to click in. If this had nothing to do with the players, every match could start one server tick after you clicked the battle button, which obviously does not happen.

 

And I am not surprised that you still do not have a clue how the MM works. You've played over 12k battles.

[Aethervoxx]

18 minutes ago, ColonelPete said:

That is the time needed for the rest of the team to click in.

Wrong! There's usually plenty of players in the queue (at the higher tiers, especially) & the numbers in the queue are continually changing while one waits.

The time in the queue (& that sometimes is 'long' considering how many players are in the queue) is exactly where the MM starts the 'arrangement' results by the wretched team imbalances.

23 minutes ago, ColonelPete said:

And I am not surprised that you still do not have a clue how the MM works. You've played over 12k battles.

Change the 12k to a 16k. This statement of yours works both ways. You can believe your fantasy, Pete, I will believe mine. 

[RuruushuVuiBuritania]

1 hour ago, Aethervoxx said:

Incorrect. WG could easily program the MM to account for player skill when compiling teams but everyone knows the MM does not do this.

 

Perhaps not, however, it could get close to true randomness. Again, this is whether WG programs their machine intelligence to do so. Hence, you either believe or not believe whether WG makes the MM 'random enough'.

 

You might have said, 'Numbers don't lie, but liars don't know how to count'.

 This appears to be the old computer adage (& statiticians retort) about data - "Garbage in = Garbage out".

Well, randomness implies that skill is indeed not counted... So the MM is indeed random then, it sweeps together ships and tiers. True randomness is way too complicated for machine intelligence at present, because the patterns we make for randomness are flawed. I think the matchmaker is random enough and sometimes too random, if my team is filled with too many bad teammates. But that is the name of the gamemode and certainly what it does.

[ColonelPete]

56 minutes ago, Aethervoxx said:

Wrong! There's usually plenty of players in the queue (at the higher tiers, especially) & the numbers in the queue are continually changing while one waits.

The time in the queue (& that sometimes is 'long' considering how many players are in the queue) is exactly where the MM starts the 'arrangement' results by the wretched team imbalances.

Change the 12k to a 16k. This statement of yours works both ways. You can believe your fantasy, Pete, I will believe mine. 

As I said, you do not know how the MM works.

 

The MM tries to fill multiple battles roosters. A Tier VIII ship can wait in line for a Tier X, IX, VIII and VIII-only battle. When you queue up in a Tier X ship and a Tier VIII ship joins after you, you see the queue numbers go up. And when a Tier VIII battle, he is waiting for, gets filled up sooner, he is removed from the queue. Apart from the fact that the queue numbers do not change in real time and they therefore show only a snapshot of the queue. And you completly ignore that when you select a ship class that everyone else is playing (BB), you have to wait a long time. Someone else might join in a less popular class (DD) after you and fill up the last spot in a battle queue to complete a battle rooster (3 DD are waiting and need 1 more DD to fill up two teams) and immediately goes into battle, while you still wait.

 

This was all explained in detail.

 

Looks like you are the one making stuff up, again...

[Aethervoxx]

3 hours ago, ColonelPete said:

As I said, you do not know how the MM works.

Neither do you, Pete, know how the MM works unless you wrote the MM code yourself.

[ColonelPete]

6 hours ago, Aethervoxx said:

Neither do you, Pete, know how the MM works unless you wrote the MM code yourself.

You do not need to code or build something to know how it works. That is not how the world works.

[_Drunk_Sailor_]

I don't mind that MM between hire and lower tiers as long as there are the same number of ships of each tier/class on each side.  However, shouldn´t the players skills also be split 50/50.  I have done some analysis on the matchmaking, and I have found that over the last 674 battles, 47% of the time my team win rate is lower than the enemies' team.

How is t his good matchmaking.  At first I thought that this is some mistake in the coding. But I doubt it.  I have the feeling that I am getting screwed by WG.  I have spent already too mauch on this game and it could be that they see me as a sucker (which I am) and are trying to squeeze more out of me?.  Obviously this isn´t a coincidence.

 

OR... I calculated something wrong.. maybe...  has anyone come across this?

 

I am a 48% win rate. So when I see that my teams only have a 47% chance, maybe I am not as bad as I think?

 

 

[RuruushuVuiBuritania]

9 minutes ago, Intoxicated_Sailor said:

I don't mind that MM between hire and lower tiers as long as there are the same number of ships of each tier/class on each side.  However, shouldn´t the players skills also be split 50/50.  I have done some analysis on the matchmaking, and I have found that over the last 674 battles, 47% of the time my team win rate is lower than the enemies' team.

How is t his good matchmaking.  At first I thought that this is some mistake in the coding. But I doubt it.  I have the feeling that I am getting screwed by WG.  I have spent already too mauch on this game and it could be that they see me as a sucker (which I am) and are trying to squeeze more out of me?.  Obviously this isn´t a coincidence.

 

OR... I calculated something wrong.. maybe...  has anyone come across this?

 

I am a 48% win rate. So when I see that my teams only have a 47% chance, maybe I am not as bad as I think?

 

 

WG has no need to entice you with good matchmaking when 1) you already payed 2) matchmaking is the same for everyone. Lots of whales with 45% winrate and free-to-play unicums with 60% winrate. 48% is in fact average for this game, since there are also draws.

Random does truly mean random in the most random possible way for the matchmaker. Random ships and classes but a fixed maximum of classes. Not taking winrate into account.

[ColonelPete]

33 minutes ago, Intoxicated_Sailor said:

...47% of the time my team win rate is lower than the enemies' team.

Which means your teams winrating was higher 53% of the time. How is that a bad thing?

[GarrusBrutus]

8 hours ago, Intoxicated_Sailor said:

OR... I calculated something wrong.. maybe...  has anyone come across this?

How is this calculated? How do you know everyone's stats from each match? Matchmaking monitor? 

Well... That is highly unreliable data. For multiple reasons that I'll explain if you want. 

Besides, nobody conducting such experiments with their own data is objective. You feel you should win more, as does every player in this game. 

Also, even though 600 battles may seem like a lot, it is in fact very little data to draw conclusions from. Just one frustrated session or a couple of bad games will skew the data. 

What I also fail to understand is this: you're a 48% wr player. Your team has won about 47% of the time in 600+- battles, correct? What % of battles did you expect to win? 

[_Drunk_Sailor_]

16 hours ago, RuruushuVuiBuritania said:

WG has no need to entice you with good matchmaking when 1) you already payed 2) matchmaking is the same for everyone.

Sure they do. A game that is perceived as unfair will not get me to drop any more money into it. What data or evidence do you have that it is the same for everyone?

[_Drunk_Sailor_]

16 hours ago, ColonelPete said:

Which means your teams winrating was higher 53% of the time. How is that a bad thing?

Correct. But I mis-wrote.  What I meant to state is that my teams advantage, in win rate is only 47%. Hence the enemy advantage is 53%. 

[SV_Kompresor]

10 minutes ago, Intoxicated_Sailor said:

Sure they do. A game that is perceived as unfair will not get me to drop any more money into it. 

You didn't actually disagree with what he said though.

 

And then there are players who say that wg made mm unfair because they want them to spend more money. How the hell they think this is supposed to work I have no idea though.

[_Drunk_Sailor_]

8 hours ago, GarrusBrutus said:

How is this calculated? How do you know everyone's stats from each match? Matchmaking monitor? 

Well... That is highly unreliable data. For multiple reasons that I'll explain if you want. 

Besides, nobody conducting such experiments with their own data is objective. You feel you should win more, as does every player in this game. 

Also, even though 600 battles may seem like a lot, it is in fact very little data to draw conclusions from. Just one frustrated session or a couple of bad games will skew the data. 

What I also fail to understand is this: you're a 48% wr player. Your team has won about 47% of the time in 600+- battles, correct? What % of battles did you expect to win? 

It’s a weighted average. Total wins/total battles of each player.

yes

how is it unreliable? Honestly I would like to know so we can suggest improvements. After all, I pay for it. So if it is trash, I would like some evidence.

respectfully… it is not an “experiment”. It an analysis of the data I have. No manipulation. It match the results of the live monitoring.

600 battles I was in.  It’s over 80 million for all players I played against. It’s enough for an indication and we’ll within margin of error.

To your last question, if the matchmaking is 47% for my teams advantage and I am purely average, then I would expect to win 47%, because I would nit generally be better or worse and nit really be able to move the needle in any way.

 

what I have noticed is that the average exp for each team, is almost dead on the same (not shown above).

[ColonelPete]

10 minutes ago, Intoxicated_Sailor said:

Correct. But I mis-wrote.  What I meant to state is that my teams advantage, in win rate is only 47%. Hence the enemy advantage is 53%. 

And now take a guess why your team was inferior

[_Drunk_Sailor_]

3 minutes ago, SV_Kompresor said:

You didn't actually disagree with what he said though.

 

And then there are players who say that wg made mm unfair because they want them to spend more money. How the hell they think this is supposed to work I have no idea though.

I disagreed that they don’t have a reason to entice. Sure they do. I can’t speak for other players.

[_Drunk_Sailor_]

1 minute ago, ColonelPete said:

And now take a guess why your team was inferior

I must be an idiot, why?

 

[ColonelPete]

Just now, Intoxicated_Sailor said:

I must be an idiot, why?

 

Look at your winrate.

Recalculate the teams, remove your winrating and exchange it for a 52% winrate player.

[RuruushuVuiBuritania]

1 hour ago, Intoxicated_Sailor said:

Sure they do. A game that is perceived as unfair will not get me to drop any more money into it. What data or evidence do you have that it is the same for everyone?

Is the burden of proof not on you, considering you were the one questioning how fair it is? If a whale gets 45% winrate and a unicum with 60% winrate is free-to-play, should it not be the other way around? There is no reason to letting paying customers get worse results. The goal is to frustrate, let someone buy something, they leave because they are not skilled enough to use a high tier premium for example, rinse and repeat.

[_Drunk_Sailor_]

2 hours ago, ColonelPete said:

Look at your winrate.

Recalculate the teams, remove your winrating and exchange it for a 52% winrate player.

You are confusing my win rate with which team has a higher advantage in each match. My report above isn´t self explanatory enough and I see some omissions in the report itself and my explanations are being misunderstood...so  let me clarify....

 

47.18% of the time (Matches), my team has a better overall wieghted win rate than the enemy.  

When you add everyone's wins and divided by everyone's total battles, you get a TEAM win rate. One for your team (friendly) and the another for the enemy.  47.18% of the time, the friendly team has a higher overall weighted win rate than the enemy.  Inversely, almost 53% of the time (matches), the enemy team has players who's average weighted win rate is better. That is nearly a 5 percent swing in balance.

Even if my personal win rate was 0%, shouldn´t MM make it so that both teams have an equal chase.  So, if I am 0% win rate, in order to make it even for the rest of the players on my team, to give them a chance despite my crappy historical performance, there should be more better players on my team than on the enemy`s?  In my holy opinion, it should be 50/50 no matter how good or bad one is.

 

Yes, sure, not all battles are going to be exactly 50/50. But after a few hundred battles, wouldn´t you think it would be closer?

 

So my win rate in fact is irrelevant if you are going to have MM being balance.

 

I know I am not great player, maybe not even average, but shouldn't my team have similar chance to win as the enemy 1/2 the time?

 

Side note:  I think it is balancing the matches on average experience per match.  I need to dig down more to see if this how they are balancing the matches.

[_Drunk_Sailor_]

1 hour ago, RuruushuVuiBuritania said:

Is the burden of proof not on you, considering you were the one questioning how fair it is? If a whale gets 45% winrate and a unicum with 60% winrate is free-to-play, should it not be the other way around? There is no reason to letting paying customers get worse results. The goal is to frustrate, let someone buy something, they leave because they are not skilled enough to use a high tier premium for example, rinse and repeat.

Well first, if you want proof, I have a lot of data. will be happy to share it with you or anyone.

Second, I even questioned if I am logically doing something wrong.

Third you said WG has no need to entice with good match making. Why not? So they are going to pick on  some players for no reason?  Why is that MM puts me in teams with a lower overall winrate 47% of the time? If it is random as you say, then it would be 50%. They claim balance. I don't see it. 

 

[ColonelPete]

5 minutes ago, Intoxicated_Sailor said:

So my win rate in fact is irrelevant if you are going to have MM being balance.

No.

You are part of the team. The worse you are, the worse is your team. The only thing you can complain about is you.

And since the MM ignores winratings, that is to be expected.

 

7 minutes ago, Intoxicated_Sailor said:

You are confusing my win rate with which team has a higher advantage in each match. My report above isn´t self explanatory enough and I see some omissions in the report itself and my explanations are being misunderstood...so  let me clarify....

I am not confusing anything. I am telling you that you get a MUCH different result when a players has a better winrating.

You are basicly proving that the player is the deciding factor over a large number of battles.

 

I also disagree with the way you calculated the teamwinrate. A 53% winrate player with 5000 games would be weaker than a 53% with 10.000 games.

Also a 40% winrate player with 50 games would barely make a dent in your teams winrate, because of the low weight, but is in reality a real burden for the team.

 

[triumphgt6]

The problem with the game at present is that the time needed to reach higher tiers is far too short. The way to get a better win rate is to transition slowly, get good at a tier and then move up. When I started very few played T10 and it took a long time to grind up. These days, you can buy a Tier 10 and this will lead to poor results. We did see similar issues when Tirpitz was released and you could almost guarantee the player would be awful. I have not seen any MM bias when playing - I have teams that are great and others that are awful. Play in a division and you do at least get two others who play to a plan. Tier X is not the only tier and is certainly not the most fun.

[RuruushuVuiBuritania]

37 minutes ago, Intoxicated_Sailor said:

Well first, if you want proof, I have a lot of data. will be happy to share it with you or anyone.

Second, I even questioned if I am logically doing something wrong.

Third you said WG has no need to entice with good match making. Why not? So they are going to pick on  some players for no reason?  Why is that MM puts me in teams with a lower overall winrate 47% of the time? If it is random as you say, then it would be 50%. They claim balance. I don't see it. 

 

WG has never claimed balance in anything except mirroring tiers and ship classes. Random also means that skill level is random. And it cannot be 50% by default because draws exist. Thus 48% is the average winrate of the playerbase. You seek for balance in the wrong places. Randomness (or close enough to it, from what machine intelligence can produce) is random ships with fixed tiers and ship classes. It has nothing to do with skill level or winrate. If it was guaranteed to be 50%, everyone would have 50% and there would be no distinction in skill level at all. I dare even say that you are getting the long end of the stick: 53% of the time you have a better team. And some people would kill for a better team 47% of the time. WG is not picking on anyone. Either your model and assumptions are incorrect or you overestimate your abilities. That is not a bad thing, a lot of players do that. To get better will improve your winrate.

You know your data and your numbers, but you simply do not know enough of the workings of statistics to effectively apply them. Same data sets with two different approaches can lead to radically different outcomes.