Current State Of Match Making - Bad Matches - Why? - Discussion

[Stand The Wall.6987]

i wonder how games with similar match making systems fare, such as: csgo, team fortress 2, and dota underlords. the failure might be mmorpg related with i'm guessing the complexity as the main culprit. i don't put much stock into complex match making algorithms since the more complex it is the more ways it can fail or be manipulated. i also wonder how other games with their own self made algorithms are doing, and if any mmorpgs with complex algorithms are doing better then gw2.

[Ryan.9387]

https://en-forum.guildwars2.com/discussion/96429/matchmaking-may-be-bugged-if-wiki-is-accurate

They say it is pseudocode but its python code. Its probably been copy pasted and its absolutely bugged as written.

This would mean matchmaking is bugged and isn't creating teams properly. The rating algorithm would sort of smooth over the bad matchmaking and put people generally in the correct bracket, just after any number of wildly imbalanced matches.

[Luthan.5236]

I play only at lower skill tier. (Population might not be an issue here yet.) For the duo que mixed with solo queue players I actually see teams without duo queued players winning more often here in the low skill tiers. I guess on the team with the duo queue those 2 (2 duo + 3 solo often) not rarely play in their own way ... good coordinated. But ignoring the other players in their team unless the - by chance - play the way they'd like them to play. (Sometimes raging at them.)

Then there is balance between classes. (Team setup and not always possible to get an exact mirroring of the class setup on the enemy side.) Also people can switch classes. And can have different builds that might couter reach other. That can get changed before match starts.

And the system does not reward fighting until the end. Barely loss gets same - in ranked - rank loss than fully getting roflstomped. Might lead to players playing worse than thtey actually could play. If they get demotivated easily. (After a long win streak by one loss getting demotivated and playing at lower performance and getting into a loss streak because of it.) And of course no different rating used for different chars/professions. I can't just try out another class where I might play better (or worse) than on my main. Unless I try unranked but I don't know when I'm ready and at same still level than my main unless there is a measurement for it. I mean ... I might even get better at Necro - cause it is considered easier to play - than on some other class. Playing x matches on unranked might not lead to same skill level for every profession. Therefor I just straight go to ranked.

Of course the matchmaker "thinks" I'm at a different rating. Matchmaking based on that.

[Leonidrex.5649]

@Stand The Wall.6987 said:i wonder how games with similar match making systems fare, such as: csgo, team fortress 2, and dota underlords. the failure might be mmorpg related with i'm guessing the complexity as the main culprit. i don't put much stock into complex match making algorithms since the more complex it is the more ways it can fail or be manipulated. i also wonder how other games with their own self made algorithms are doing, and if any mmorpgs with complex algorithms are doing better then gw2.

all games have their own issues. in league it takes too long to reach the ranking you deserve, I can be grandmaster, make new acc and it will take me several months to get to grandmaster again, I just dont think people have this much time.in csgo, the biggest issue is lack of communication and hackersin r6 siege I never had any problems with matchmaking, but im low ranking there since me and my buds dont play much so the problems might begin at higher ranking where having premade team and not having remade could be a big deal.I personally think that the biggest gw2 problems are as follows1 lack of players2 games doesnt teach you how to play3 most classes hold your hand and play for you ( thief, guard, necro cough cough ) -> training wheel classes ( add in bunkers like scrapper and other shit )In most if not all games, even if you do the easiest strategy imaginable there is still insane numbers of ways enemies can fuck you over, but not here.You cant punish thief that just rotates.You cant bunish bunker guard

[Marxx.5021]

I think there are bad matches because match making is garbage. It is often not able to create balanced matchups. There are several reasons for bad match making I guess. Just to name a few : lack of player, changing class balance, no proper way to rate duo queuers, favour fast queue times over close rating, incorrect rating (due to reset and number of matches played). There is an overall lack of incentive to play properly when match making fails. So player just throw or go afk. I would say match making discourages new players from playing PvP which is not exactly helpful if there is a lack of players anyway.

[vorpal.1497]

I think everyone has given up on the entire "e-sports" concept of GW2 so why do we even bother with matchmaking at all really, when the purely random matchups are much faster and much more enjoyable/satisfying/fun

[Trevor Boyer.6524]

@"vorpal.1497" said:why do we even bother with matchmaking at all really, when the purely random matchups are much faster and much more enjoyable/satisfying/fun

This is pretty much the bottom line.

At this point, the match maker is actually functionally working to create imbalanced matches, for the reasons of the phase1ing and phase2ing I mentioned. Most people nowadays can give testimony that you win 500 to 100 and then you lose 100 to 500 and it just keeps doing this over and over. This match making isn't random or volatile or bad, it's actually doing EXACTLY what it is supposed to do, which isn't working anymore in low population. You'd think it would grant a balanced match 500 to 400 every once in awhile but somehow it reliably, key word there "reliably", will 500 to 100 win, and then 100 to 500 lose, over and over and over. This is a PATTERN and isn't luck. The system is very good at knowing how to shove stronger players on one team and weaker ones on the other. I cannot stress enough again, that something within the algorithm is definitely making this happen. So in other words, the match maker really isn't broken it's just neglected and hasn't been tuned for a lower population in quite some time. Whatever that function is, it needs to change to 2020 so we can once again get matches that look more like 500 to 400 or even 500 to 300. But this 500 to 100, 100 to 500 bologna needs to stop.

I'm telling you, it's that phase 1 and phase 2 stuff. It once worked when enough people were online playing who were closely above or bellow a player's rating, but now in 2020 with low population that stupid function starts reaching out to put plat 2s against gold 2s for their challenge phase, rather than putting gold 3s against gold 2s. I'd place money at this point, that this is exactly what's happening.

And yeah you're right, completely organic random match making such as GW1 RA, would actually make for better quality of matches over the course of time than a match maker that is doing what ours is doing now.

People want matches to be winnable and losable, as in the players are so evenly matched skill wise, that the match could go either way. This is the best case scenario. What people are tired of, is a very predictable win/lose/win/lose/win/lose, there is no fun in that. When you're put into a lose match where it is quite clear that the other team is stacked with players several divisions higher than the players on your team, you already know what's going to happen. And even when you're put into a win match, it's just kind of boring to stomp nabs 500 to 100.

The match maker needs to stop phase1ing and phase2ing, and just put people into matches completely solely based on average rating vs. average rating. Let organic wins and loses sort things out from there. We DO NOT need the match maker to be attempting to mitigate wins & loses with phase1ing and phase2ing. It's making the game feel too predictable and bad to play. There is nothing exciting about that.

I really need to dig up that old quote from the game dev, so you guys can see what I keep referring to here.

[Shiyo.3578]

@Leonidrex.5649 said:3 most classes hold your hand and play for you ( thief, guard, necro cough cough ) -> training wheel classes ( add in bunkers like scrapper and other kitten )In most if not all games, even if you do the easiest strategy imaginable there is still insane numbers of ways enemies can kitten you over, but not here.You cant punish thief that just rotates.You cant bunish bunker guard

This is my biggest issue with this game in general and what makes it so hard to carry.

[Trevor Boyer.6524]

@"Ryan.9387" said:https://en-forum.guildwars2.com/discussion/96429/matchmaking-may-be-bugged-if-wiki-is-accurate

They say it is pseudocode but its python code. Its probably been copy pasted and its absolutely bugged as written.

This would mean matchmaking is bugged and isn't creating teams properly. The rating algorithm would sort of smooth over the bad matchmaking and put people generally in the correct bracket, just after any number of wildly imbalanced matches.

Yup you know what, after reading through this, some things kind of make too much sense to ignore. Aside from direct bugs or exploitations, it really looks like old accounts are definitely being punished for total games played in 2020.

Look at this example:

• Two games are being made in the match maker
• One game has: RED (A Team USA main account duo with 20,000+ games played each) - 1500 - 1500 - 1500 vs. BLUE (duo not yet selected) - 1500 - 1500 - 1500
• Other game has RED (A 1600 duo with around 2000 games played each) - 1500 - 1500 - 1500 vs. BLUE (duo not yet selected) - 1500 - 1500 - 1500
• There is a duo who just joins the NA server queue who is another 1600 duo with less than 2000 games played
• Then there is my duo who is 1600 each but we both have 20,000+ games played each, we enter NA queue same time as other duo

The algorithm is being forced to take our two duos just joining the queue and put them somewhere into these matches. From what the algorithm notes are saying, if any of it is accurate at all, it's going to give my 1600 duo the hard match against Team USA due to the matching of total games played, and it will give the easier match to the other 1600 duo who has less than 2000 games played. This is because it directly mentions that it attempts to match players into matches who have similar total games played.

So in the event of a situation like this, the old main accounts will always be getting the short end of the stick, every time. This actually explains a lot about why old main accounts with tens of thousands of games played, get ridiculously difficult matches more often than not.

This also shows some evidence towards the feeling of there being a "good list" and a "bad list" where some accounts are being given easier matches. That list is sorted by total games played. It isn't as simple as "good list" "bad list" but rather a cascade of different tiers of that good to bad. A fresh alt that is only glory rank 20, entering ranked for its first season with only 100 unranked games played or something like that, is going to be granted massively lenient match making compared to the guys with 20,000 games played who keep getting thrown against top 20 players during their placements over and over and over.

This shows that things are starting to not make sense within the leaderboards. Take both of those 1600 duos and we'll call them duo (A) and duo (B). Duo (A) is maintaining that 1600 while having to fight top 20 players every match. And then duo (B) is maintaining their 1600 by more often than not, being placed against other 1600 duos, who have a similar 2000 games played or close to it, while often dodging the bullet of being made to go against guys with 20,000+ games played. Regardless of the leaderboards saying duo (A) & (B) are the same rating, there is a massive disparity in actual skill level between these two duos. The rating system in glicko is failing hard in terms of displaying true skill value through the rating, and it would seem to be mostly due to the algorithm using total games played and glory rank for match making. It may have been a functional system for sorting players 4+ years ago and before f2p accounts happened, but now in 2020, 9 years into the lifespan of Guild Wars 2, when old veterans are memeing on alts for superior match making, this old algorithm function is just punishing old main accounts in harsh ways. This certainly would seem to explain why we'll see pve guys come into ranked and fumble into top 100 positions for the season, but when facing them in ATs they're atrociously ineffective despite their achieved rankings. I've seen it, you've seen it, we've all seen it, we've all noticed this.

For those of you who are familiar with Mario Kart, what we have going on right now in the leaderboards, is a situation where guys who take 2nd place in 150cc are reflected as lower ranked than guys who take 1st place in 50cc.

Frankly I'm tired of it and I'd like to see it fixed.

[Pimsley.3681]

It's FAR easier to grasp what OP is referring to if one has multiple accounts (say 3-5) acquired at different times. I used to be very active in PVP from 2012-early 2016 but gave up on my main account because I was getting far better yolo set-ups on alt accounts. I saw this pattern as early as 2014-15ish, when there was that first ever pre-season/ranked/whatever....

One alt account made and stayed at top 30. Other EU alt made it to top 50. Main NA account...I gave up on.

Then seasons started - Alts easily made it to Diamond.

BTW, main account's been around since Aug/Sept 2012...I literally have the GW2 t-shirt!!!

Alts got WAAAAY better matches. I think it has something to do with either time of acquisition and number of PVP games played - including HOTJOIN trolling. Remember those? Lel.

But I do believe there is a pattern and there are certain unfair algo "baskets" that pre-determine your match-up.

BTW, I study algos for a living, specifically, AI and search engine algos.

[Rose.2593]

Some of this makes sense, Several times this season i have been one win away from moving to a higher division, only to suddenly loose, and then have loose streaks etc.

Also Matchmaking should balance the players, not just make the MMR equal for the teams....Like it should pull 10 roughly equal players, not 2 higher level players, and then 8 average ones to balance the numbers or something. If the players are balanced it would be a fairer game, then having a player "carry" the team.

Still balance the team of course, but not just to get numbers to be equal, attully balance the skill of the players.

A Lot of the matches this season have been 1 sided, where 1 team for some reason is of a much higher level.

Also botting, and afking seems alot worse this season, maybe due to low populations, or the time I have been playing but some matches are no fun when you have no chance of winning due to someone sitting in the home base, which shouldn't count as being active.

[vorpal.1497]

@Trevor Boyer.6524 said:What's going on there? How the hell does that match happen with a 6s queue?

When someone declines a match and there's 9/10 people ready I think it makes yet another concession among many (concessions for duos, concessions for classes, match history, dishonor history etc.) and shoves whatever is in queue into that slot.

[Swagger.1459]

@"Trevor Boyer.6524"

"I have on my team: Me + people I've never seen before who are exploding on contact each engagement."

@"Ben Phongluangtham.1065" said:There is a reason we made both the T1 and T2 versions legendary. The T1 version (slumbering) doesn't require tournament wins. But if you want the version that provides some fancy effects, it's going to take a bit of effort to find a team (maybe put one together yourself!) or brave lfg. T2 is purely a cosmetic reward, so I personally don't feel it's out of line to make it require a bit more leg work and team coordination. GW2 PvP is essentially a team mode and at least some of the higher tier rewards should reflect the effort it takes to build and maintain a team.

Keys sentence of Ben P's post...

"GW2 PvP is essentially a team mode and at least some of the higher tier rewards should reflect the effort it takes to build and maintain a team."

Get on a team of friends that you practice and play with so you can increase your odds of success. THAT is what TEAM PvP is about. Complaining that the match maker is problematic when you ignore all the other factors going on is your flaw in thinking. Get a team if you want to improve ranking because "GW2 PvP is essentially a team mode", and if you want to improve your odds then make the "effort it takes to build and maintain a team.".

[vorpal.1497]

This thread is about matchmaking in the ranked spvp system which hasn't allowed teams for... many years afaik

edit: this thread is also not at all about "winning more" it's about the quality/enjoyability of matches (winning 500 to 50 is not even fun for the winners)

[Zoricus.2439]

@Trevor Boyer.6524 said:

@"Zoricus.2439" said:If there was a much larger player base I don't think the issues we've seen would be as pronounced. My alt account w 200 matches is150 pts above my main account w/10k + matches played. I have to try hard supper carry on my main account. The alt account matches seem to be slightly more balanced.

Let me clear something up, because you mention "seem to be more balanced" rather than saying "seem to be easier".

Do the matches on the alt feel more balanced as in, not just the point outcomes, but over the course of say 10 matches, you are not seeing lopsided point results or even seeing lopsided results as in no clearly visual signs of one side being ultra dominant, nearly as often?

^ Because if that is the case, I'm telling you, this MMR effects match making under rating, total games played effects base MMR, is a function that just need to be reviewed. There is enough evidence adding up, from the players I've spoken with, that shows total games played base MMR inflation is causing some kind of janky problem in the match maker.

Not seeing as many lopsided wins/losses compared to my main account. I don't have to "try hard" to climb in rating on the alt accounts. Some folks Ive played with over the years sit around in gold 3 for extended periods only to maybe squeak out a P1 rating at the end of the season.

[Zietlogik.6208]

I have been having astronomically bad matches for the past month or so, not sure when this all started, but its noticeable for sure

[Lord of the Fire.6870]

@Archer.4362 said:

I've been playing pvp since the beginning, from this classification system, the system almost always looks for you to be in the 50% of wins and losses. You win 3 games in a row, the 4th always takes almost twice as long to jump, even putting for example 1:30 or 1:40, it goes after 4 min, to find a team below your level.

Last game to raise the Tier as before, the waiting time is double and the team is well below your division, therefore you lose 3 or 4 games in a row, it is a loop.

sorry my bad english

Yes but this is naturally your rank rise as a result you get stronger encounters. The problem is more how the game evaluate the round you played after 20 games or so you nearly get the same number of point deducted or added regardless against who you played against or how you performed. This isn't fair at all .

[Rose.2593]

4 times in the last week or so I have been 1 win away from going to a higher division, then all of a sudden the winning streak ends, and match making puts me in a bad match that's not winnable, against higher level players.

Why is it fixed so I can't go up??