Jump to content
You need to play a total of 20 battles to post in this section.
Egoist__Inori

the relationship between Matchmaking and Win Rate

102 comments in this topic

Recommended Posts

0
[HAATO]
Members
19 posts
5,064 battles

Hi there.

 

Today I discussed about the matchmaking of wows with my friends, but we didn't get any result.

 

So if our teammates depends on how we play?

 

For example: If I were a player with 70% win rate, then almost all of my teammates will get a win rate near 45% and make the average win rate of my team get close to 50%

 

Because I'm not sure if I've seen any official answer of matchmaking about this.

 

Thanks for your help

Share this post


Link to post
Share on other sites
Beta Testers
4,007 posts
14,539 battles

The Matchmaker does not take winrate into account at all. 

 

Sometimes i really doubt that

 

I get put with teams that are 40% while the enemy team is a 50% avrage

Most of the times i noticed that i am the only one above 50% or there are max 2 more players in my team

I might not be a great player but when you get all the 40%ers you really doubt mm or that's just the server avrage

 

But who knows except wg and sure they won't tell

Edited by Destroyer_WolfgangZenker

Share this post


Link to post
Share on other sites
8
[NOOBS]
Members
14 posts

From what I have experienced YES the MM does take into account your ranking, Winrate is a major component of WG ranking.

Share this post


Link to post
Share on other sites
41
[ZAZEN]
Alpha Tester
404 posts
8,964 battles

 I would love to find out how you know all your teammates and all the enemy win rats? XVM is not available or is it?

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

The Matchmaker does not take winrate into account at all.

 

On the contrary. Though WG has stated it does not take a player's skill into determining a line up, it does, for the most part attempt to equal out a line up based on WR. Most battles will have no more than a combined team average of a 4% difference in win rate. The highest difference I have tracked was just under an 8% difference and came to the conclusion it was due to a low population on the server. "6am PST" (I have tracked over 1,200 battles) Now, on a side note to this, during those 1,200 battles, there have been 29 battles with a player having at least a 70% WR on the other side but I have never had (1) on any of my teams. And for the 60+, 276 on opposing side versus 53 on my side. I can not explain a reason for such a vast difference in playing with Unicum's versus not. I previously made a post similar to this and used to complain about the lopsidedness but have given up. (I have my thoughts but I won't go there) And, for what it's worth, I stopped collecting data before the Christmas holidays.

Share this post


Link to post
Share on other sites
Supertest Coordinator
7,223 posts
14,450 battles

If the matchmaker tries to balance people by winrates, it does no good to sabotage people with bad win rates by piling them into a team with only one high WR person, because that brings the high WR person closer to 50%, and brings the sub-50%s further away from 50%, i.e. 1 step forward, 11 steps back.

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

 I would love to find out how you know all your teammates and all the enemy win rats? XVM is not available or is it?

 

It's collecting the data from

https://worldofwarships.com/en/community/accounts/

and from being a retired old man with nothing better to do and the knowledge of using an EXCEL spreadsheet.

(God, I need a life) :(

  • Cool 1

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

I posted this on another post

 

Here's Wargaming's patent on the MatchMaker and how it operates, supporting observed behaviour:

"According to another aspect, the matchmaking server may store a win/loss percentage for each user (or vehicle) at a given battle level.matchmaker.png As the player's win/loss ratio decreases, the player becomes more likely to be placed in battles having battle levels at the lower end of the allowable range, whereas as the player's win/loss ration increases, the player becomes more likely to be placed in battles having battle levels at the upper end of the allowable range. Thus, when a player has been repeatedly put into too many difficult battles, the balancing is done in favor of easier battle sessions, thereby encouraging the player by providing an easier game environment. Similarly, when the player has been repeatedly put into too many easy battles, the balancing is done in favor of harder battle sessions, thereby keeping the player challenged instead of letting the player become bored with easy games." 

 

And was reminded not to take it too seriously

 

 

Share this post


Link to post
Share on other sites
79
[WOLF2]
Members
252 posts
9,980 battles

I posted this on another post

 

Here's Wargaming's patent on the MatchMaker and how it operates, supporting observed behaviour:

"According to another aspect, the matchmaking server may store a win/loss percentage for each user (or vehicle) at a given battle level.matchmaker.png As the player's win/loss ratio decreases, the player becomes more likely to be placed in battles having battle levels at the lower end of the allowable range, whereas as the player's win/loss ration increases, the player becomes more likely to be placed in battles having battle levels at the upper end of the allowable range. Thus, when a player has been repeatedly put into too many difficult battles, the balancing is done in favor of easier battle sessions, thereby encouraging the player by providing an easier game environment. Similarly, when the player has been repeatedly put into too many easy battles, the balancing is done in favor of harder battle sessions, thereby keeping the player challenged instead of letting the player become bored with easy games." 

 

And was reminded not to take it too seriously

 

 

 

This is arguing for a standard MMR style matchmaker but using W/L over MMR.  That is, when a player loses a bunch of games, they start playing with other players who have lost a bunch of games.  When a player wins a bunch of a games, they start playing with players who also have won a lot of games.  Unfortunately, it uses W/L % as it's metric rather than MMR which means rather than stabilizing at a point where you play people equal to your skill always, all players should eventually reduce to 50%, which means everyone is equal regardless of skill, which means you'll start to get volatility.  The above description is just not a particularly good match making strategy.

 

What it does say is COUNTER to what is being described by the OP.  He is asking if I'm a 70% player, is the rest of my team 50% players to even it out.  What this is saying is that if you're a 70% player, you should end up fighting other 70% players.

 

Edit:  I should note that you could interpret this differently.  I interpreted "lower end of allowed range" to mean against other players who also suck.  It could be interpreted as you are the lower end with other players that are good.  Even in that case, a true MMR strategy just seems far superior.

Edited by real_icebeast

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

 

He is asking if I'm a 70% player, is the rest of my team 50% players to even it out.  What this is saying is that if you're a 70% player, you should end up fighting other 70% players.

 

If 70% players only played with 70% players, the number of 70% players would dwindle dramatically.  Now, does this mean if one side has a 70% player should the other side have one 2? Good question. But at the same time, why does one team have 3 DD's and the other side have only 2? MM is not perfect, far from it. So many things about this game is lopsided and isn't close to being equally matched. The only way they could make it equal is everybody to have the identical ships and then let the marbles fall where they fall.

As another thought, let's say there are 11 players with average WR of 50%, by adding 1 player with a 70% WR, the team's average will only go up to just under 52% WR.

Edited by LoneStormchaser

Share this post


Link to post
Share on other sites
3
[TSF_1]
Beta Testers
9 posts
2,755 battles

 

On the contrary. Though WG has stated it does not take a player's skill into determining a line up, it does, for the most part attempt to equal out a line up based on WR. Most battles will have no more than a combined team average of a 4% difference in win rate. The highest difference I have tracked was just under an 8% difference and came to the conclusion it was due to a low population on the server. "6am PST" (I have tracked over 1,200 battles) Now, on a side note to this, during those 1,200 battles, there have been 29 battles with a player having at least a 70% WR on the other side but I have never had (1) on any of my teams. And for the 60+, 276 on opposing side versus 53 on my side. I can not explain a reason for such a vast difference in playing with Unicum's versus not. I previously made a post similar to this and used to complain about the lopsidedness but have given up. (I have my thoughts but I won't go there) And, for what it's worth, I stopped collecting data before the Christmas holidays.

 

I posted this on another post

 

Here's Wargaming's patent on the MatchMaker and how it operates, supporting observed behaviour:

"According to another aspect, the matchmaking server may store a win/loss percentage for each user (or vehicle) at a given battle level.matchmaker.png As the player's win/loss ratio decreases, the player becomes more likely to be placed in battles having battle levels at the lower end of the allowable range, whereas as the player's win/loss ration increases, the player becomes more likely to be placed in battles having battle levels at the upper end of the allowable range. Thus, when a player has been repeatedly put into too many difficult battles, the balancing is done in favor of easier battle sessions, thereby encouraging the player by providing an easier game environment. Similarly, when the player has been repeatedly put into too many easy battles, the balancing is done in favor of harder battle sessions, thereby keeping the player challenged instead of letting the player become bored with easy games." 

 

And was reminded not to take it too seriously

 

 

 

Sure, they made a patent, and I could see why a company would want to do this.

 

Though I don't believe that World of Warships is doing anything of the sort.

 

Interestingly - the data you collected doesn't support your theory. With your rather lacklustre stats (Sub 47% win rate) if the server really was trying to "Equal out the line up based on WR" then you should be seeing many unicums and grapes on your side trying to drag your WR up, or conversely, trying to use you to drag their WR down. 

 

Gezeder

 

Share this post


Link to post
Share on other sites
229
[-_W_-]
Members
1,089 posts
2,418 battles

I posted this on another post

 

Here's Wargaming's patent on the MatchMaker and how it operates, supporting observed behaviour:

"According to another aspect, the matchmaking server may store a win/loss percentage for each user (or vehicle) at a given battle level.matchmaker.png As the player's win/loss ratio decreases, the player becomes more likely to be placed in battles having battle levels at the lower end of the allowable range, whereas as the player's win/loss ration increases, the player becomes more likely to be placed in battles having battle levels at the upper end of the allowable range. Thus, when a player has been repeatedly put into too many difficult battles, the balancing is done in favor of easier battle sessions, thereby encouraging the player by providing an easier game environment. Similarly, when the player has been repeatedly put into too many easy battles, the balancing is done in favor of harder battle sessions, thereby keeping the player challenged instead of letting the player become bored with easy games." 

 

And was reminded not to take it too seriously

 

Given the term "battle levels," and the way similar terms have been used to describe the tier of a battle (e.g. T5-7 = battle tier 7) I interpreted this as "players with low win rates will be top tier more often; players with high win rates will be bottom tier more often." Especially since nothing in there indicates that "battle level" refers to the skill of the users involved in the battle, but "allowable range" could refer to the range of tiers.

  • Cool 2

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

 

 

Interestingly - the data you collected doesn't support your theory. With your rather lacklustre stats (Sub 47% win rate) if the server really was trying to "Equal out the line up based on WR" then you should be seeing many unicums and grapes on your side trying to drag your WR up, or conversely, trying to use you to drag their WR down.

 

Gezeder

 

 

I can't agree with you more and as I stated, I can not explain the reason for such a huge difference in playing with good players and not playing with good players. Even the total number of games I've recorded and documented shows that less than 25% of those games had players in the 60 WR range. That's not really a huge number. What this says is that the vast majority of players are sub 50%.

 

One more thing. my stats are blocked. WoW stats are not updated. I currently have well over 2,000 battles now.

Edited by LoneStormchaser

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

 

Given the term "battle levels," and the way similar terms have been used to describe the tier of a battle (e.g. T5-7 = battle tier 7) I interpreted this as "players with low win rates will be top tier more often; players with high win rates will be bottom tier more often." Especially since nothing in there indicates that "battle level" refers to the skill of the users involved in the battle, but "allowable range" could refer to the range of tiers.

 

You're looking at it over the long run "too many" (what ever that number is, who knows) The key words in there is "more likely to be placed" and this is very vague.

As I stated earlier, don't take this seriously. Or :coin:

Share this post


Link to post
Share on other sites
52,091
[MAUS]
Members
13,698 posts

Someone brought up that patent again?  Wargaming made several patents (including for the E-50M).  That doesn't mean it's in use.

 

Neither World of Tanks or World of Warships uses any form of skill based Matchmaking.  It doesn't need to.  The RNG of shuffled random teams creates enough variance that (most) people are unable to tell the difference anyway and see any shortcomings in their own statistics as a conspiracy rather than taking ownership of their own failings.  You think Matchmaker is out to get you?  You're not that important or interesting, I'm sorry to say.

 

If you want better teams, get better yourself.  That way you can guarantee you always have at least one good player on your side.  If you really want to skew the odds, division with similarly skilled players.

  • Cool 5

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

Neither World of Tanks or World of Warships uses any form of skill based Matchmaking.  It doesn't need to.

 

You're using the same argument everybody else uses. No where in this post does it say MM is using skill to set the teams. It's talking about win rate. If your going to say that skill results in a higher WR, then you have confirmed, that teams that have a lower WR than the side that has a higher WR has, right off the start, a lower possibility of winning because every battle is not set at 50-50 win rates. If so, then MM is not doing a good job. In fact, most battles are won by the team that has the higher WR. The argument would then be, well, someone with a high win rate lost because they got put on a team with lesser skill. No. Skill is not the issue. WR is.

Share this post


Link to post
Share on other sites
52,091
[MAUS]
Members
13,698 posts

 

You're using the same argument everybody else uses. No where in this post does it say MM is using skill to set the teams. It's talking about win rate. If your going to say that skill results in a higher WR, then you have confirmed, that teams that have a lower WR than the side that has a higher WR has, right off the start, a lower possibility of winning because every battle is not set at 50-50 win rates. If so, then MM is not doing a good job. In fact, most battles are won by the team that has the higher WR. The argument would then be, well, someone with a high win rate lost because they got put on a team with lesser skill. No. Skill is not the issue. WR is.

 

Skill does result in a higher win rate,  But skill and win rate have nothing to do with team assembly.  Matchmaking doesn't care what the player's win rates are when it builds the teams.  Again, it doesn't need to.  RNG provides enough variety that sometimes skilled players will have an easy win.  At other times, they'll have to carry like the dickens to win.  And sometimes no matter what they do, they'll still lose.  In this way, the games are unpredictable and players of all skill levels need to apply themselves in order to give them the best chances of winning.

  • Cool 1

Share this post


Link to post
Share on other sites
229
[-_W_-]
Members
1,089 posts
2,418 battles

 

You're using the same argument everybody else uses. No where in this post does it say MM is using skill to set the teams. It's talking about win rate. If your going to say that skill results in a higher WR, then you have confirmed, that teams that have a lower WR than the side that has a higher WR has, right off the start, a lower possibility of winning because every battle is not set at 50-50 win rates. If so, then MM is not doing a good job. In fact, most battles are won by the team that has the higher WR. The argument would then be, well, someone with a high win rate lost because they got put on a team with lesser skill. No. Skill is not the issue. WR is.

 

Skill results in higher WR because a skilled player can affect the outcome of a battle more in their team's favor. Thus, a skilled player ends up with a higher WR over time because while they don't win every battle, they make enough difference that they win more battles. I don't currently have a 57% overall WR across over 1600 battles because I got lucky with MM; I have it because, over that many games, the games where I have a potato team against unicums and the games where I have a unicum team against potatoes balance out such that the only consistent factor is my own ability to influence the game in my team's favor. Because WR and skill are thus tied together, WR can't be the issue unless skill is also the issue.

 

The thing is, a matchmaker that attempts to produce two teams with equal overall WR (and thus, in theory, equal overall skill) would be a horrible matchmaker, because it would essentially force everybody to a roughly 50% win rate, at which point it defeats itself because it can no longer determine what players are skilled and what player aren't. The end result would be an MM that still has effectively random teams in most games.

 

In essence, the simple fact that some people have a 60% WR over 1000+ games while others have 40% WR over 1000+ games means that MM doesn't balance teams based on skill or WR or anything else.

Share this post


Link to post
Share on other sites
Supertest Coordinator
7,223 posts
14,450 battles

The issue is that the pool of players at any time is finite, and we know this to be true because matchmaker launches games as soon as enough players are available. Given 24 players, there's only a finite way to split them because you only have at most one of each carrier and potential tier and ship type restrictions. If one team ends up with a faildiv of two 43% Iowas the other team will likely get reasonable 50% ships. If matchmaker were given several hundred people to work with and tried to balance that it would be more reasonable but matchmaker just ejects players into matches ASAP based on simple rules, as best as it can.

Share this post


Link to post
Share on other sites
120
[-V-]
Members
830 posts
5,254 battles

 

the games where I have a potato team against unicums and the games where I have a unicum team against potatoes balance out such that the only consistent factor is my own ability to influence the game in my team's favor. Because WR and skill are thus tied together, WR can't be the issue unless skill is also the issue.

 

The thing is, a matchmaker that attempts to produce two teams with equal overall WR (and thus, in theory, equal overall skill) would be a horrible matchmaker, because it would essentially force everybody to a roughly 50% win rate, at which point it defeats itself because it can no longer determine what players are skilled and what player aren't. The end result would be an MM that still has effectively random teams in most games.

 

This statement contradicts with itself or what your saying, you just happened to be put on the better team.

 

Skill does result in a higher win rate,  But skill and win rate have nothing to do with team assembly.  Matchmaking doesn't care what the player's win rates are when it builds the teams.  Again, it doesn't need to.  RNG provides enough variety that sometimes skilled players will have an easy win.  At other times, they'll have to carry like the dickens to win.  And sometimes no matter what they do, they'll still lose.  In this way, the games are unpredictable and players of all skill levels need to apply themselves in order to give them the best chances of winning.

 

No one has any proof of how MM really works. I have never seen any statement by WG stating how it works, nor would I expect it to.

 

I can say that personally, greater than 63% of the time, those teams I was placed on, before the battle started, had, on average, nearly a 4% less WR than the opponents team Also, over the long haul, if MM was based on pure random line-ups, the number of 70% and 60% players I have played with and against would not be so lopsided. (See my post above)

Are lineups set all by chance? We can all, only guess. Is there some mathematical formula which decides, who knows. This post is not the only post that has questioned how MM works. Is there a record of a WR/skill on file somewhere which determines who people play with?

Most of the proponents of the current MM system seem to be those that have a high WR and the only ones who come to its defense.

Sorry     Egoist     for getting off topic.

 

Edited by LoneStormchaser

Share this post


Link to post
Share on other sites
Beta Testers
5,082 posts
5,575 battles

Neither World of Tanks or World of Warships uses any form of skill based Matchmaking.

 

How do you know this? Without possessing some reliable insider information it's very difficult to determine either way. This is merely repeating the official position on the matter.

Share this post


Link to post
Share on other sites
52,091
[MAUS]
Members
13,698 posts

 

How do you know this? Without possessing some reliable insider information it's very difficult to determine either way. This is merely repeating the official position on the matter.

 

And what's wrong with the official position?  Healthy skepticism is fine but I have never seen anything in the four years I've been playing Wargaming's products to see anything that goes contrary to the official position -- just a lot of wishful thinking of conspiracy theorists and those who want to blame someone else for their poor performance in game.  By contrast, I've seen a lot of well written discourses on how random matchmaking easily demonstrates all of the events we see in game without special need to create some convoluted matchmaking program that "targets" select players to make them perform poorly.
  • Cool 4

Share this post


Link to post
Share on other sites
Beta Testers
5,082 posts
5,575 battles

And what's wrong with the official position?  Healthy skepticism is fine but I have never seen anything in the four years I've been playing Wargaming's products to see anything that goes contrary to the official position -- just a lot of wishful thinking of conspiracy theorists and those who want to blame someone else for their poor performance in game.  By contrast, I've seen a lot of well written discourses on how random matchmaking easily demonstrates all of the events we see in game without special need to create some convoluted matchmaking program that "targets" select players to make them perform poorly.

 

And? Just because you assign labels to dissenting opinions by calling them conspiracy theories and elevate the opinions of your own by citing some nebulous "well written discourses", it doesn't make either of the two opinions any less valid. It's all still speculation on both sides in the absence of proof. You can't bull[edit] a bull[editor].

 

What's wrong with the official position? I'm disappointed that I have to spell it out, but what's wrong with it is that its sole origin has a financial conflict of interest. Specifically, it would hurt their business if substantiated proof were to be presented demonstrating that the MM is in fact skill or win rate based to some degree. Doubting an official position that is burdened with such a conflict is I believe what you earlier called "healthy skepticism".

  • Cool 3

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×