Matchmaker Discussion Thread & MM Balance

[ColonelPete]

2 minutes ago, Camperdown said:

Well we all have our preferences. My current WR is 51.5, I wouldn't mind being matched with teams with 47-58% WR and be rid of the utter window licking tools.

And the people above 58%? That is 1.6% of the playerbase. How long are they supposed to wait for a match?

[GarrusBrutus]

17 minutes ago, Camperdown said:

Well we all have our preferences. My current WR is 51.5, I wouldn't mind being matched with teams with 47-58% WR and be rid of the utter window licking tools.

I mean, in other games with bigger playerbases it would make sense. F.e. Counterstrike, Valorant or rocketleague.

But in a niche game like WoWs, that has only a fraction of the playerbase and has on top of that a tier system that already divides players over ten tiers, it is impossible to also implement any form of ELO system.

It divides a small playerbase into even smaller little playerbases. This makes the Matchmakers job neigh impossible.

 

A few other problems that come to mind are:

How are you going to deal with divisions using such a system?

Which metric are you going to judge players by? Overall winrate? solo winrate? PR? Ship winrate?

How do you prevent players from rigging such a system to have advantageous matchmaking?

How do you prevent players from "smurfing" aka making second accounts to stomp low tier players?

How do you keep the game fun for those who are bouncing between being the best in the lower bracket and being the worst in the higher bracket? The "yoyo'ing guys".

How do you measure stats when for example top players only meet top players? Their winrates will drop due to increased resistance, dropping them down in the elo system.

[hardkur911]

12 hours ago, GarrusBrutus said:

Just riddle me this: 

You've won more games the past 21 days than on average before that. Why are you complaining again? 

i found a way to get on the winning side almost every time , but u not gonna believe it cuz its not random at all . 

[Camperdown]

4 hours ago, ColonelPete said:

And the people above 58%? That is 1.6% of the playerbase. How long are they supposed to wait for a match?

They match with the top 25%.

[ColonelPete]

29 minutes ago, Camperdown said:

They match with the top 25%.

So a 58% WR player gets matched from 47% WR to the top (around 59% of the playerbase)...

[GarrusBrutus]

1 hour ago, hardkur911 said:

i found a way to get on the winning side almost every time , but u not gonna believe it cuz its not random at all . 

Oh cool man, what is your trick? 

[DFens_666]

44 minutes ago, Camperdown said:

They match with the top 25%.

 

Naturally you have to put them into 4 groups from 0-25% / 25-50% / 50-75% / 75-100%.

So if you are >51% you would only face people who have 51-100% WR, as thats the group with the top 25% players.

[AkulaTSR]

8 hours ago, GarrusBrutus said:

........."If I want to goof around there is randoms."

 

Haha surely thats the sort of mentality that has randoms in the state it is now 

[AkulaTSR]

Obviously `randomly` a lot of good players could end up on the same team , leading to a roflstomp , but roflstomps 8 out of 10 games ??? Whats happening there ?? 

 

 

[DFens_666]

44 minutes ago, AkulaTSR said:

Obviously `randomly` a lot of good players could end up on the same team , leading to a roflstomp , but roflstomps 8 out of 10 games ??? Whats happening there ??

 

Too many decent players left the game within the past 1,5 years due to CV rework, so we have too much bottom dirt left, who will just die within 3 mins?

[Camperdown]

2 hours ago, ColonelPete said:

So a 58% WR player gets matched from 47% WR to the top (around 59% of the playerbase)...

I haven't modelled this. But you don't need to. MM matches you up with a group of a predefined bandwidth, can be 25% of the player population, can be something else. If waiting time gets too much, MM increases the bandwidth. The point is, rather than getting matched up with the entire player base, you only get matched up with a group that is somewhat similar to you in capability. So, instead of super unicums, total tools and bots and average Joe's being in the same battle, as it is now, you get matched up with the 25% or so of the player base which is not too far from you in skill and outlook. Games get more interesting and competitive. 

I would like that, you don't have too. I am just trying to explain how an alternative MM could work.

WG is not going to do it, so .....

 

[Camperdown]

2 hours ago, DFens_666 said:

 

Naturally you have to put them into 4 groups from 0-25% / 25-50% / 50-75% / 75-100%.

So if you are >51% you would only face people who have 51-100% WR, as thats the group with the top 25% players.

There would not be a hard cut off, MM would try to match you with players of roughly comparable skill.

[DFens_666]

4 minutes ago, Camperdown said:

There would not be a hard cut off, MM would try to match you with players of roughly comparable skill.

 

I dont think 51% and 65% is comparable skill. Neither is 45% and 50%.

So basicly, we have lower ~20-30% as one group, then a huge chunk at around 50% WR which is like 60-70% and <10% above that. The average guys will have a relatively easy time to find matches, while the best could sit in queue indefinetely, depending on the tier they pick.

[ColonelPete]

12 minutes ago, Camperdown said:

I haven't modelled this.

That is the problem.

[Camperdown]

7 minutes ago, DFens_666 said:

 

I dont think 51% and 65% is comparable skill. Neither is 45% and 50%.

So basicly, we have lower ~20-30% as one group, then a huge chunk at around 50% WR which is like 60-70% and <10% above that. The average guys will have a relatively easy time to find matches, while the best could sit in queue indefinetely, depending on the tier they pick.

The point is: do you want to have the bots at 35%, the window licking tools at 40% matched up with super unicums, or do you want to have matches a bit more differentiated?

I would like to have a bit more differentiation. If you do not, fine. I am not trying to convince you. I am just trying to say that a different system is possible, and might give us some more interesting game play because the bots and tools get separated from the more serious and skilled players.

[Camperdown]

6 minutes ago, ColonelPete said:

That is the problem.

Whatever lifts your boat, mate 

[ForlornSailor]

1 minute ago, Camperdown said:

do you want to have the bots at 35%, the window licking tools at 40% matched up with super unicums, or do you want to have matches a bit more differentiated?

 

Id like them to learn in the first place. In the second place, have an economy in place, that stops them from playing T9/T10 constantly.

Seriously, the best game, the best experiance for all of us will be achieved, if we have a playerbase, that is interested in winning, knows the mechanics and knows their ship.

If you ease the games for them - you will achieve the opposite. My opinion. Games will still be piss poor, if people dont need to perform to get what they want - which is XP, credits and wins.

[Camperdown]

8 minutes ago, ForlornSailor said:

 

Id like them to learn in the first place. In the second place, have an economy in place, that stops them from playing T9/T10 constantly.

Seriously, the best game, the best experiance for all of us will be achieved, if we have a playerbase, that is interested in winning, knows the mechanics and knows their ship.

If you ease the games for them - you will achieve the opposite. My opinion. Games will still be piss poor, if people dont need to perform to get what they want - which is XP, credits and wins.

I agree. I want them to learn too. But we also have accounts with 10K + battles and 30-35% WR. Bot accounts I assume. And similar accounts with 40-45% WR. Are they still going to learn after 10K games? I assume not. 

I would like to have groups of similar skills and mindset, cause I think it would give more interesting games and less aggravation.

I don't mind if you have a different preference.

[GarrusBrutus]

7 minutes ago, Camperdown said:

I would like to have a bit more differentiation.

Same here bro, same here. Atleast I would like to get rid of griefer-players like this:

Removing this type of players will make games already endlessly more enjoyable for everyone involved.

 

Anyway, each system that balances players based on skill (with the variable being either WR, PR or damage) is doomed to kill itself. Since over time outliers will be either buffed or nerfed (depending on which side of the mean they are) to be average. Superunicums at the start will inevitably become 50%WR players and so will this lovely gentleman of the picture above.

These systems work in games like Rainbow Six Siege because there the playerbase is much larger and the ELO system works continuously in their everongoing ranked mode instead of the seasonal itteration we have in wows.

[eliastion]

17 hours ago, Camperdown said:

This has been explained a gazillion times even on this thread alone. If you have a ELO type of rating this does not happen. The good effect is that good players are no longer burdened with tools on their team and battles can be much more interesting.

You clearly failed to read the post I was responding to. Let me remind you:

 

problem is when game put all super unicums on one side and new players on another , these should be easy fixed just spread them equally among both teams

 

It was NOT about an ELO-type system (this kind of skill-based MM is not viable but, yes, not quite as idiotic at concept level, at the very least).

[Arrive_Alive]

As said countless times in this thread: a real matchmaking is not possible due to different tiers and the small playerbase. At least not outside of the usual prime times.

But this game needs a "hardcap" on the MM. Found myself in 46% average WR teams today, twice in a row. Half the people in my team didn't even have 300 battles on total but were playing t8/t9 ships (getting to high tiers this fast is another bad game design from WG but it happened and we'll have to deal with it) . The other ones, well just the usual "10k battles and nothing learned yet" types of players. Enemy teams were around 52-53% on average. In one of these games i did 220k and 3 kills in roughly 6 minutes in Bourgogne (got killed after this) and yet we were 4 (!!) ships behind in kills. When you see teams like this on your MMonitor, you know you have to make a huge impact right at the start but well, still not doable.

 

To prevent the worst stomps and onesided games the WR difference between teams shouldn't be higher than 5%. Maybe 4% is possible, but i don't have any data This should be doable, even when waiting times might slightly increase. Team with 5% less average WR might still lose 9 of 10 times but might at least put up a fight.

Also putting a 40% 3man-div against a 60% 3man-div should not happen, regardless of any other parameters. This is one of the reasons why weekends are the worst time to play, too many divisions. One side gets the baddies, the other the purple ones. Or just baddies everywhere and whoever suicides first, loses.

 

1 hour ago, GarrusBrutus said:

Same here bro, same here. Atleast I would like to get rid of griefer-players like this:

Removing this type of players will make games already endlessly more enjoyable for everyone involved.

 

We really need an ingame option to report players, not "one click and reported" but something to fill out why we want to report that player and limit it to once a day or something (to avoid abuse of this system). And WG should take this way more seriously, even if it's a whale. In a team-based game, players like this constantly ruin the fun of 11 people. Imagine how many thousand battles this guy just ruined for his team. This is not even trying. As you said, just griefing (or botting).

 

No one wants to bother with an email to the support team, which is the current way to report these people. 

[159Hunter]

7 hours ago, eliastion said:

You clearly failed to read the post I was responding to. Let me remind you:

 

problem is when game put all super unicums on one side and new players on another , these should be easy fixed just spread them equally among both teams

 

It was NOT about an ELO-type system (this kind of skill-based MM is not viable but, yes, not quite as idiotic at concept level, at the very least).

And how often does that happen? 

Every time ppl bring this up, it's specific games (usually lost ones). All it is, is confirmation bias. Bring a large enough data set to prove that game does this systematically and we can have a word.

 

And we're still with the: which WR/PR/... do we need to look at (solo, division, ship, class, tier....). There are people with unicum WR around that play 90% of their games at tier I and fail the moment they hit tier IV...

[ZifTTroll]

11 hours ago, Marble_Eyes said:

We really need an ingame option to report players, not "one click and reported" but something to fill out why we want to report that player and limit it to once a day or something (to avoid abuse of this system). And WG should take this way more seriously, even if it's a whale. In a team-based game, players like this constantly ruin the fun of 11 people. Imagine how many thousand battles this guy just ruined for his team. This is not even trying. As you said, just griefing (or botting).

Exactly, like the people that play worse than bots and either don't care or even think that they did well...

 

There is no way to solve MM with winrate the average is 50% so if 12 75% WR players gets matched with 12 75% WR players then the WR will slowly shuffle back to 50% and everyone will be at the average given some time. A system like PR is much better at sorting because WR can change regardless of player skill and these situations forces that so everybody will techncally have the same WR and subsequently "skill"

[ColonelPete]

4 minutes ago, ZifTTroll said:

Exactly, like the people that play worse than bots and either don't care or even think that they did well...

 

There is no way to solve MM with winrate the average is 50% so if 12 75% WR players gets matched with 12 75% WR players then the WR will slowly shuffle back to 50% and everyone will be at the average given some time. A system like PR is much better at sorting because WR can change regardless of player skill and these situations forces that so everybody will techncally have the same WR and subsequently "skill"

PR will average out too.

• PR  has a win component
• and damage and kills will be more evenly distributed in more homogeneous teams

[eliastion]

8 hours ago, 159Hunter said:

And how often does that happen? 

Every time ppl bring this up, it's specific games (usually lost ones). All it is, is confirmation bias. Bring a large enough data set to prove that game does this systematically and we can have a word.

And why are you asking this of me? You should probably go discuss it with the guy that made the claim *shrug* I was merely explaining that trying to divide the unicums between the teams is an awful idea of solving this, regardless of how often or how rarely it actually ends up happening that one team has all the good players when things are left random.

 

That being said... I'd say the odds of "all unicums in one team" depend a lot on the number of unicums in a match and the ships they play. If there's just two of them (assuming no divisions and that they play different class so that they occupy different "slots") the chances of them landing in one team are as high as 50%. And, obviously, unicum divs are a frustrating thing to play against (although many unicum divs are actually less unicum than it would seem because some of them are less "unicum skill players that play together in a div" and more "blue skill players that are unicum by always playing in a div").

Oh, and if it's a 3 unicum DD division and a solo unicum cruiser, the chances of then all being on one team is also 50%

 

So, well. I personally wouldn't go as far as saying that the problem of vastly skewed matches is non-existent or even rare. The issue is there, it happens, and it does so often enough to warrant acknowledgement. It's just that the solutions to that range from not viable in practice (ELO-like systems, league system etc.) to absolutely disastrous on basic concept level (skill-balancing the teams by equally splitting unicums and potatoes between them).

 

So, basically, it's an issue could be summed up as: "Well, the issue is there but things we could do to amend it are either not going to work or would be worse than the problem, or both."