Skill based match making.

[thornzero]

7 minutes ago, Asym said:

We are a game of 6,000 or so players whom are not on at the same time.

Isn't that just NA server? EU is more than double that usually but not to-day. I saw earlier just under 1o k. Then there is still Asia and Russia which I don't know their numbers. IF they combined the servers, there still would be long queues for sufficient match ups, is what I think.

 

Look, I am trying to figure out what is going on the past few weeks. Because I am playing better and luck doesn't last for weeks. The old forum goer I can't remember says how unlucky for you to lose all the time. How unlucky he says, and he has unicum stats -no problem. Yester-day I couldn't do anything right and was at 25% for the day. How unlucky for me - he would say it.

 

I really think time and effort are involved. The player base is going to ebb and flow with and without skill. Randoms to me is; difficult mode. [Brawls with a bunch of ships, not 1 or 2 or 3 is more like randoms lite to me.] Randoms is therefore random in most ways including combatants. 

Or you could think I got carried for three weeks in a row. [I was that bad a player not too long ago. But when I was 41% wr it showed. I had no idea what I was doing.]

[Asym]

2 minutes ago, thornzero said:

Isn't that just NA server?

Yes, because I play on the NA server. 

[thornzero]

Fair. I didn't think hard enough that any change would be isolated to each server. 

[Type_93]

1 hour ago, thornzero said:

I really think time and effort are involved. The player base is going to ebb and flow with and without skill

That’s the ticket right there. I spent the first few years after the open beta just enjoying shooty pew pew pixel boats. It wasn’t until the first few seasons of ranked and CB were I decided to put work into becoming a competent player. It doesn’t take much effort really. I spent the last 6 years keeping my WR above 53% and just recently passed the 54% threshold. It can be frustrating at times and losing streaks will happen. But no matter the bad teams, if you play effectively and have positive battle impact, you will see your stats improve. 

Edited March 6 by Type_93

[Pugilistic]

1 hour ago, Type_93 said:

That’s the ticket right there. I spent the first few years after the open beta just enjoying shooty pew pew pixel boats. It wasn’t until the first few seasons of ranked and CB were I decided to put work into becoming a competent player. It doesn’t take much effort really. I spent the last 6 years keeping my WR above 53% and just recently passed the 54% threshold. It can be frustrating at times and losing streaks will happen. But no matter the bad teams, if you play effectively and have positive battle impact, you will see your stats improve. 

I did over 7k games at 47% before I realized this was something you could actually get better at. I made headway to 50% and then was aided by the first effective treatment I got for the visual stuff associated with chronic vertigo, and have improved somewhat further. Better play really does mean better WR. 

[Utt_Bugglier]

4 hours ago, Pugilistic said:

Better play really does mean better WR. 

Heresy!

Everyone KNOWS it’s the MM!!!

[Colonel Potter]

Really wouldn't work well in my opinion since you can inflate your win percentage by playing low tiers, which usually have more bots. If you're that type to do that, you're usually playing against new players and bots, so your inflated win percentage doesn't reflect on your actual skills since you really haven't the experience against players who have earned a higher win percentage in high tiers.

For instance, I raised my win percentage from sub-48% to around 50.7% (the last time I checked) by playing mostly tier 10s. I never even thought to play low tiers to raise it. Now, I know I am far from being a great player, but I am confident I can do better in high tiers than someone who only has only inflated their win percentage in low tiers against bots and newer players.

You will still get teams that weren't matched up well if there are inflated win percentages. And besides, just because you have everyone with the same win percentages in a match doesn't mean there won't still be blowouts. Most teams don't work together in Randoms, and is why you see blowouts more often than not. Have been on teams that should have been smoked, but have easily won because we worked together.

Basing MM on winning percentages won't cure any "problems".

[Jakeshuffle]

I don't see skill based match making keeping teams from being stomped in other games like Overwatch, etc.

[Schnitchelkid01_]

On 3/6/2024 at 8:14 PM, Colonel Potter said:

Really wouldn't work well in my opinion since you can inflate your win percentage by playing low tiers, which usually have more bots. If you're that type to do that, you're usually playing against new players and bots, so your inflated win percentage doesn't reflect on your actual skills since you really haven't the experience against players who have earned a higher win percentage in high tiers.

For instance, I raised my win percentage from sub-48% to around 50.7% (the last time I checked) by playing mostly tier 10s. I never even thought to play low tiers to raise it. Now, I know I am far from being a great player, but I am confident I can do better in high tiers than someone who only has only inflated their win percentage in low tiers against bots and newer players.

You will still get teams that weren't matched up well if there are inflated win percentages. And besides, just because you have everyone with the same win percentages in a match doesn't mean there won't still be blowouts. Most teams don't work together in Randoms, and is why you see blowouts more often than not. Have been on teams that should have been smoked, but have easily won because we worked together.

Basing MM on winning percentages won't cure any "problems".

The problem now with going to low tiers for me at least to "take a break" or try to win some games is that they are filled with so many bots that they decide the match by seemingly how well the DD bots play and sure I will have impact but it usually becomes a 50-50 sort of deal.

[Bumblegoose]

28 minutes ago, Schnitchelkid01_ said:

The problem now with going to low tiers for me at least to "take a break" or try to win some games is that they are filled with so many bots that they decide the match by seemingly how well the DD bots play and sure I will have impact but it usually becomes a 50-50 sort of deal.

A very true and relevant observation. Often on this forum and elsewhere, the impression is given that winning at low tiers is quite easy. However, there are poweful 'bot rolls' to navigate, in which you can easily find your team four, five or six ships up or down after a couple of minutes of play. Many of these games cannot be saved. In addition, you deal with seasoned veterans at these lower tiers along with lower tier unicum divisions. Certainly on the Asia server, inexperienced new players ('baby seals') are rarely seen at these lower tiers.

[Kruzenstern]

On 3/7/2024 at 5:14 AM, Colonel Potter said:

Really wouldn't work well in my opinion since you can inflate your win percentage by playing low tiers, which usually have more bots. If you're that type to do that, you're usually playing against new players and bots, so your inflated win percentage doesn't reflect on your actual skills since you really haven't the experience against players who have earned a higher win percentage in high tiers.

For instance, I raised my win percentage from sub-48% to around 50.7% (the last time I checked) by playing mostly tier 10s. I never even thought to play low tiers to raise it. Now, I know I am far from being a great player, but I am confident I can do better in high tiers than someone who only has only inflated their win percentage in low tiers against bots and newer players.

You will still get teams that weren't matched up well if there are inflated win percentages. And besides, just because you have everyone with the same win percentages in a match doesn't mean there won't still be blowouts. Most teams don't work together in Randoms, and is why you see blowouts more often than not. Have been on teams that should have been smoked, but have easily won because we worked together.

Basing MM on winning percentages won't cure any "problems".

Well, obviously the win rate (or whatever other ratio is used to determine 'skill') to work from would have to be that of the ship you selected or at least the tier you selected, not the global one.

Still, I find the general idea of skill-based matchmaking unattractive. I like winning. And if I git gooder, I want to win more. Not get harder opponents or worse teammates (though I find it hard to imagine how THAT would be possible, seeing most of the teams these days are already rock bottom) and win less and eventually end up near 50% with everyone else.

[Jakob Knight]

Bah.  Just give me more ammo and fuel.  I'll go kick the enemy's kneecap out while everyone else worries how tall they stand.

 

 

Edited March 14 by Jakob Knight

[Unlooky]

5 hours ago, Bumblegoose said:

A very true and relevant observation. Often on this forum and elsewhere, the impression is given that winning at low tiers is quite easy. However, there are poweful 'bot rolls' to navigate, in which you can easily find your team four, five or six ships up or down after a couple of minutes of play. Many of these games cannot be saved. In addition, you deal with seasoned veterans at these lower tiers along with lower tier unicum divisions. Certainly on the Asia server, inexperienced new players ('baby seals') are rarely seen at these lower tiers.

Winning at low tiers is easy. Don't pretend otherwise. Every problem you've listed with winning there exists to a greater degree at higher tiers, except for bots. Bots exist mostly as XP pinatas for both teams anyways, so it's really not an issue. It's not a 100% guaranteed win like co op, but its certainly easier the lower you go. 

[YouSatInGum]

On 3/5/2024 at 8:50 PM, Snargfargle said:

The game already has skill-based matchmaking of a sort. Most people in the higher tiers of Ranked generally are better than the average player. You can select teams for Clan battles, ensuring that those on your team are at least competent. If you can get into tournaments you will be expected to be good and should expect to face equally good players. And then, there is always the Training Room, where you can organize a match to your heart's desire.

The problem with skill-based matchmaking is that it increases queue times. Random games during non-peak hours already are either limited in player numbers or full of bots. In Age of Empires III, the maps were chosen by the game host, who then could accept or remove individual players as well. There were only three players per team and two teams but sometime it took over an hour before everyone finally decided that the teams were balanced enough to proceed.

I'm in a lot of stomps. In fact it's the usually mode of play for a WOWS match, However, I'm not on the losing team all the time. With my Massachusetts, after 1500 games, my win percentage is 57%. A steamroll oftentimes occurs for one or more of the following reasons.

1. Matchmaker placed better divisions on one team than the other.

If you see a [SALTY] division on one then that team is probably going to win unless there is maybe there is an [07] division on the other team.

2. Matchmaker placed a terrible division on your team.

 

While high tiers maybe separate the more experienced players (relatively speaking) from the those that only have 500 games or or less, once you are past that hurdle, players at t10 really aren't any better than what you find at T7-9.

For the rest of this post... I'm going to respond to perceptions that some here seem to have.

First one...  SBMM will add to queue time.  The answer to this one ranges from absolutely yes to practically no extra time needed.  It all depends on the form of implementation.  For example, if the MM tries to group players into brackets or tries to add restrictions to current "bucket" system that WG uses, then yeah, you can bet that wait time will go up especially during non- peak times.  On the other hand, a SBMM "light" type system that is only looking to prevent obvious mismatches and only adjusts teams after the roster has been populated, will only add the amount of time it takes for a server to do some simple arithmetic and swap players between rosters, then you are talking about milliseconds of added time.... or practically speaking no added time.  No doubt there are different flavors in between

As for there not being enough players to run a SBMM, the answer to that is going to be very similar to above.  It all depends on the implementation.  One that is only doing roster swaps after team formation to prevent large skill disparities, pretty much won't add any time.  Let me also mention here, that I have taken randoms results and recorded outcomes versus what MMM showed me.  I did this for 312 battles from T6 -  T10 (t8 being most common), however this is now over 2 years ago.  Obviously dated data and results, but I still think the core conclusion hold which is that once you pass a 3% or 4% team average WG differential, the top team wins with at least a 90% confidence level. 

Since I'm on the subject, some here have suggested other measures of skill or performance.  For example, what about average XP?  Well, there are 2 major problems with this measure.  First is that WG has mixed in premium time bonuses such that the XP is saved in the data base after the bonus, but doesn't identify those accounts with premium much less time periods with bonus versus not.  Basically the data is complete garbage.  I don't have a clue as to why WG does it this way.

Anyway, the second issue is the formula for XP measurement has had major changes and mixing the two would invite major data conflation issues as two identical games would result in two very different scores.  Mixing large amounts of historical data like this would give very skewed results.... not to mention adding this garbage on top of the premium time garbage problem.  At one time, XP probably was the best overall measure of individual game performance, however the formula change from a year ago took this several steps in the wrong direction.

Measures like PR, have their own issues as well as PR is fairly easy to manipulate.  It's not a bad measure when uses for personal growth and looking at recent games.  The older the game, the less relevant it is to current performance.  PR only relies on 3 basic stats such that many activities that matter such as capping and map control.  For this reason, I tend to ignore PR as global measure and only use it as relative measure between ship types or for personal growth.

 

I do not care for the current "random" matchmakers system.  It's poorly named as it's not really all that random since divisions are the antithesis of randomness.  In fact, "solo" players are at a measurable disadvantage due to effect of divisions.  I have broken the reasons for this in past posts, but on average it works against solo players.

Unfortunately, all of this is sort of moot point as WG has made it very clear that they are not interested in anything resembling a SBMM.  They won't even experiment on the test server and go so far as repeating misleading information to support their position. 

 

 

 

 

[Snargfargle]

3 minutes ago, YouSatInGum said:

WG has made it very clear that they are not interested in anything resembling a SBMM. 

The description of Brawl says that "Players are matched based on their skill level" but I've no idea how this is implemented.

 

[Wolfswetpaws]

🍿

[YouSatInGum]

1 hour ago, Snargfargle said:

The description of Brawl says that "Players are matched based on their skill level" but I've no idea how this is implemented.

 

It's the same concept in government where you name the bill or law to be passed the opposite of what it really does... ie. Clear Skies Act really weakens EPA rules or Patriot Act reduces privacy protections.

Depending on the restrictions and tiers, Brawl will produce some of the most unbalance teams that you will ever see.  Besides, in the past WG has said that Brawl MM is based solely on wins.  How that has more to do with skill rather than the number of time you hit the battle button is beyond me.  I'd love to see WG try to explain it, but considering no more forums, the days of them answering questions seem to be ancient history.

[ArIskandir]

2 hours ago, Snargfargle said:

The description of Brawl says that "Players are matched based on their skill level" but I've no idea how this is implemented.

 

Last I knew it only accounted for number of wins. 

So for my first match I can be matched with a guy who has a record of 0-8. If I have a record of 5-0, I could be matched with a guy with 5-18... perfect match of skills!!!

[Bumblegoose]

4 hours ago, Unlooky said:

Winning at low tiers is easy. Don't pretend otherwise. Every problem you've listed with winning there exists to a greater degree at higher tiers, except for bots. Bots exist mostly as XP pinatas for both teams anyways, so it's really not an issue. It's not a 100% guaranteed win like co op, but its certainly easier the lower you go. 

I disagree entirely with this illusory view of lower tier gameplay. But each to their own, we can agree to disagree and move on.

[3LUE]

Have you seen high tiers? It is a mess full of players with less than 500 games. High tier is a joke right now. 80% of the players up there have either yellow or red stats, and most of them fall on the red stat side. 

[Asym]

34 minutes ago, 3LUE said:

Have you seen high tiers? It is a mess full of players with less than 500 games. High tier is a joke right now. 80% of the players up there have either yellow or red stats, and most of them fall on the red stat side. 

So, only green players can play"?  How can anyone get GuD is they can't play?  Help them.  Or, play PVE with the rest of us.

[Wolfswetpaws]

1 hour ago, 3LUE said:

Have you seen high tiers? It is a mess full of players with less than 500 games. High tier is a joke right now. 80% of the players up there have either yellow or red stats, and most of them fall on the red stat side. 

The ship is the "ticket" to enter the battle.

Have ship?  Then the player can enter the battle while sailng the ship.

Some people have "alt" accounts or give/receive gifts of ships or whale-purchase ships.
So what?

What bothers you and other people, really? 
That your opponents are too easy to sink?  That your teammates are too easy to sink?
That you're not guarranteed an outcome that you desire?

 

Edited March 16 by Wolfswetpaws

[Ensign Cthulhu]

57 minutes ago, Wolfswetpaws said:

That you're not guarranteed an outcome that you desire?

This.

[Unlooky]

3 hours ago, 3LUE said:

Have you seen high tiers? It is a mess full of players with less than 500 games. High tier is a joke right now. 80% of the players up there have either yellow or red stats, and most of them fall on the red stat side. 

Versus low tier literal and figurative bots? I'm not saying Tier 10 lobbies are difficult, I'm saying that they are comparatively more skilled on average. Do you run Potato Alert or MMM? Here's the numbers of the 7 consecutive battles I played in my Kléber in the span of just over an hour from 4pm-5pm (central) on a Thursday afternoon, roughly around NA peak time. 

Game 1:

0 <500 battle players, 2 with red PR

Game 2:

0 <500 battle players, 8 with red PR

Game 3:

1 <500 battle player, 7 with red PR  

Game 4:

1 <500 battle player, 5 with red PR

Game 5:

1 <500 battle player, 4 with red PR

Game 6:

1 <500 battle player, 6 with red PR

Game 7:

3 <500 battle players, 6 with red PR

 

These were all full 24 player lobbies. The worst (red) players clearly do not make up a majority of the matches, and there was never more than one <500 battle player with the exception of one game. 

 

I know 7 games isn't a significant sample size, but this should represent what an average hour long WoWs Tier 10 play session looks like. 

[3LUE]

On 3/16/2024 at 12:02 AM, Unlooky said:

Versus low tier literal and figurative bots? I'm not saying Tier 10 lobbies are difficult, I'm saying that they are comparatively more skilled on average. Do you run Potato Alert or MMM? Here's the numbers of the 7 consecutive battles I played in my Kléber in the span of just over an hour from 4pm-5pm (central) on a Thursday afternoon, roughly around NA peak time. 

Game 1:

0 <500 battle players, 2 with red PR

Game 2:

0 <500 battle players, 8 with red PR

Game 3:

1 <500 battle player, 7 with red PR  

Game 4:

1 <500 battle player, 5 with red PR

Game 5:

1 <500 battle player, 4 with red PR

Game 6:

1 <500 battle player, 6 with red PR

Game 7:

3 <500 battle players, 6 with red PR

 

These were all full 24 player lobbies. The worst (red) players clearly do not make up a majority of the matches, and there was never more than one <500 battle player with the exception of one game. 

 

I know 7 games isn't a significant sample size, but this should represent what an average hour long WoWs Tier 10 play session looks like. 

I run match maker monitor. Red stats players have increased in great numbers in the high tier. It is even worse as more skilled players are leaving the game or just not playing other than clan battles.