PSA how placements work

[Faux Play.6104]

For new players, the guild wars 2 system will start you with a rating of 1200 and a deviation of 350. So it is saying it is 95% sure you will be somewhere between a 500 and 1900 rating. As you play more matches your deviation will go down to a point. For returning players, it will "soft reset" your rating so you start at (last season +1200)/s. So someone that was 1600 last season would start the season at 1400.

The example plot shows two players who both went 6-4 over their first 10 placement matches. The blue player ended up in plat 1 while the red player ended up in silver 3. The dots represent their rating, and the error bars represent the 95% confidence interval of that rating. Each player played matches against opponents that were within 150 rating points. The blue player won their first 3 matches while the red player lost its first 2. The first matches have the biggest impact on your rating. Also, there is still overlap between the players predicted skill level even with a ~400 rating difference since they have not played enough matches to have a very accurate rating.

[JTGuevara.9018]

And this entire graph, along with your accurate explanation, shows me--to my face--why the ELO-based league system is hot GARBAGE.

According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.

This pretty much echoes what me and many other players have said: The most important matches are the placements. They determine your season. After those, what is even the point of playing after placing high except to stave off decay. It's the most NONSENSICAL thing ever. Really, you're better off with some sort of win/loss percentage based system or round-robin format, like it's done in sports--assuming you want things to be fair and competitive, that is.

[Stand The Wall.6987]

its a game of luck since you can't choose your team mates, at least for placements. after that your good plays can and will make you climb.

[Exedore.6320]

@JTGuevara.9018 said:According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.You're assuming that the red (orange) player is not approaching their proper rating. The example makes no claim of what each player's expected rating is.

The reality is the red player is likely not as good as the blue player.

[shadowpass.4236]

@"JTGuevara.9018" said:And this entire graph, along with your accurate explanation, shows me--to my face--why the ELO-based league system is hot GARBAGE.

According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.

This pretty much echoes what me and many other players have said: The most important matches are the placements. They determine your season. After those, what is even the point of playing after placing high except to stave off decay. It's the most NONSENSICAL thing ever. Really, you're better off with some sort of win/loss percentage based system or round-robin format, like it's done in sports--assuming you want things to be fair and competitive, that is.

Player 2 isn't screwed from the get-go. Bad placements should really only affect where you start out at. If you're a good player, you can place in bronze and 1v4/5 your way to gold 2-3 then 1v2/3 your way to plat pretty easily in a day or two.

Placements are important matches but they definitely aren't the end-all-be-all. If I lost all 10 of my placement matches, I'd probably be able to climb several divisions in a day simply because almost everyone under (probably?) plat has some severely obvious drawbacks in either mechanical skill and/or rotational knowledge.

I mean look at the build you linked in that matchmaking thread... You're basically a 25k HP meat shield with no way to close/create distance or prevent a burst from hitting you asides from your 2 dodges and an instant fear once every traited 48 seconds in the case of "Fear Me!" Can guarantee you'd get hard clapped by any competent player on a decent build simply due to that alone. To give you more details, your heal is also entirely dependent on whether or not the enemy decides to continue hitting you. So, asides from being on a longer cooldown than most heals, someone could shut down your sustain simply by stowing their weapons when you activate Defiant Stance and avoiding your relatively short range burst skill on axe to stop Adrenal Health from proccing.

Besides, as someone that mostly casually solo ques around plat 2 - plat 3 nowadays, going 8-2 on my placements ends up sticking me somewhere in plat 1 and going 7-3 places me at 1500 rating. Again, this is largely independent from the facts that I've dropped to gold when helping silver-ranked friends of mine learn how to play conquest in ranked and yet have been in the top 10 numerous times. Whether or not I go up or down depends on the matches I play afterwards, how much effort I put into them, and if I'm playing with a good duo or decide to let RNG play a larger factor in my win/loss ratio.

[Faux Play.6104]

@"JTGuevara.9018" said:And this entire graph, along with your accurate explanation, shows me--to my face--why the ELO-based league system is hot GARBAGE.

According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.

This pretty much echoes what me and many other players have said: The most important matches are the placements. They determine your season. After those, what is even the point of playing after placing high except to stave off decay. It's the most NONSENSICAL thing ever. Really, you're better off with some sort of win/loss percentage based system or round-robin format, like it's done in sports--assuming you want things to be fair and competitive, that is.

what you may have missed from the statement is people are playing matches against people rated near their level. anyone that is a plat level player should be able to carry a silver tier match unless multiple people are throwing. Even if they have bad luck in placements they will end up winning 60-70% of their matches. At that win rate you will quickly advance out of silver/gold even after your rating stablizes. The other thing about the math is you only know someone's rating to a certain "confidence level". That is what the bars are on the graph. Even when your rating stabilizes, that band is still fairly large. From my experience, you will have a lot more fun if you have bad luck in placements than if you have good luck. If you end up winning a bunch of matches you shouldn't, you will be one of the people in the forms posting about how they keep loosing matches because they are getting bad teammates.

[Chase.8415]

@JTGuevara.9018 said:And this entire graph, along with your accurate explanation, shows me--to my face--why the ELO-based league system is hot GARBAGE.

According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.

This pretty much echoes what me and many other players have said: The most important matches are the placements. They determine your season. After those, what is even the point of playing after placing high except to stave off decay. It's the most NONSENSICAL thing ever. Really, you're better off with some sort of win/loss percentage based system or round-robin format, like it's done in sports--assuming you want things to be fair and competitive, that is.

Bro it's so easy climbing out of silver that if you're stuck there, quite simply you belong in silver.

I am not even sure if people play with their monitor turned on in silver.

[JTGuevara.9018]

@Exedore.6320 said:

@"JTGuevara.9018" said:According to this graph, player 2 is

screwed

from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.You're assuming that the red (orange) player is not approaching their proper rating. The example makes no claim of what each player's expected rating is.

The reality is the red player is likely not as good as the blue player.

Oh, I'm not "assuming" anything...

I can clearly see with my own two eyes how this whole thing works. It DOESN'T MATTER what the system's so-called "expected value" is, I could care less. The point is that player 2 will struggle to climb while player 1 has a much easier time due to the deviation.

[JTGuevara.9018]

@"shadowpass.4236"

"Player 2 isn't screwed from the get-go."------> Get your eyes checked and look at the graph again..."Bad placements should really only affect where you start out at. " -----> Supposedly....and yet they don't....

sigh....come on guys...just look at the graph. How the fuck is player 1 going to catch up? Player 2 got the RNG matchmaking at the outset and is not going to struggle as much. That's just STUPID. One player gets a free ride while another has to grind hundreds of matches? It's a joke. It's freakin backwards. It's just MIND-BOGGLING that people don't think that there's anything wrong with this.

[Tharan.9085]

@JTGuevara.9018 said:@"shadowpass.4236"

"Player 2 isn't screwed from the get-go."------> Get your eyes checked and look at the graph again..."Bad placements should really only affect where you start out at. " -----> Supposedly....and yet they don't....

sigh....come on guys...just look at the graph. How the kitten is player 1 going to catch up? Player 2 got the RNG matchmaking at the outset and is not going to struggle as much. That's just STUPID. One player gets a free ride while another has to grind hundreds of matches? It's a joke. It's freakin backwards. It's just MIND-BOGGLING that people don't think that there's anything wrong with this.

If you have to grind hundreds of matches to get out of silver you're simply bad.

[shadowpass.4236]

@JTGuevara.9018 said:@"shadowpass.4236"

"Player 2 isn't screwed from the get-go."------> Get your eyes checked and look at the graph again..."Bad placements should really only affect where you start out at. " -----> Supposedly....and yet they don't....

sigh....come on guys...just look at the graph. How the kitten is player 1 going to catch up? Player 2 got the RNG matchmaking at the outset and is not going to struggle as much. That's just STUPID. One player gets a free ride while another has to grind hundreds of matches? It's a joke. It's freakin backwards. It's just MIND-BOGGLING that people don't think that there's anything wrong with this.

Do you even know how to read the graph yourself??? It shows nothing except the rating difference between the same win ratio out of 10 placement matches and the impact the outcomes that the first 2 have on the end result.

Bad placements only affect where you start out at. If I placed in bronze, according to your logic, I would never be able to end in any other division except bronze. Yet that is entirely, 100% false.

Not to mention that the matches after your placements are incredibly volatile in terms of how much rating you gain and lose so it's pretty easy to make up that difference in the games that follow a bad set of placements. Similarly, it's just as easy for someone with good placements to lose that rating in the matches that follow...

Instead of telling everyone else that they're crazy, maybe you should take a look at yourself and realize you have no idea how this stuff works and should not be attempting to lecture others on what's right and what's wrong. Oh, and maybe your placements would've went better if you didn't run a build that doesn't serve any purpose other than to soak damage should you be fortunate enough that your enemies even deemed you enough of a threat to target you in the first place.

Here some more math for you. Assuming you win between 10-15 rating per match, it should only take you a minimum of 24-37 games to make up the 363 rating difference the graph displays. If you have to play hundreds of games to make that up, you deserve to be in that rating. Not to mention that every single match is weighted depending on who you're fighting which it seems you have yet to comprehend.

[Exedore.6320]

I had a friend who is normally platinum1. On an alt account, he intentionally lost all his placement matches to end up in Bronze. He was able to get back to platinum by mid-season playing somewhat casually. You need at least 120 games for a PvP season to be on the leaderboard.

The system is working. If you're not climbing, you're at the correct rating.

[Khalisto.5780]

/grabpopcorn

just watching ppl blaming everything but themselves for the rating they have right now

[AliamRationem.5172]

@JTGuevara.9018 said:And this entire graph, along with your accurate explanation, shows me--to my face--why the ELO-based league system is hot GARBAGE.

According to this graph, player 2 is screwed from the get-go. It will be difficult and next to impossible for player 2 to progress to player 1's level, assuming player 2 even bothers to grind so many matches.

This pretty much echoes what me and many other players have said: The most important matches are the placements. They determine your season. After those, what is even the point of playing after placing high except to stave off decay. It's the most NONSENSICAL thing ever. Really, you're better off with some sort of win/loss percentage based system or round-robin format, like it's done in sports--assuming you want things to be fair and competitive, that is.

It's been a long time since I took a statistics course, but I don't think you're interpreting this graph properly at all.

[AliamRationem.5172]

@JTGuevara.9018 said:@"shadowpass.4236"

"Player 2 isn't screwed from the get-go."------> Get your eyes checked and look at the graph again..."Bad placements should really only affect where you start out at. " -----> Supposedly....and yet they don't....

sigh....come on guys...just look at the graph. How the kitten is player 1 going to catch up? Player 2 got the RNG matchmaking at the outset and is not going to struggle as much. That's just STUPID. One player gets a free ride while another has to grind hundreds of matches? It's a joke. It's freakin backwards. It's just MIND-BOGGLING that people don't think that there's anything wrong with this.

No. You're just trying to make the graph fit your narrative when it simply doesn't. Those error bars indicate that the system has such low confidence at the end of 10 placement matches, that it can't be sure whether these two players are of roughly equal skill or not, despite placing them hundreds of rating points apart.

Let's assume that they are roughly equal in skill. How does the player who placed low catch up? Presumably, if these two players of roughly equal skill were matched against each other enough times, all other things being equal, you would expect to see about a 50/50 win/loss rate for both players. Yet if that did occur, the rating gap between these two players would close because the low-rated player would gain more rating for a win and lose less rating for a loss against the higher rating player.

Of course, you know this. But you seem to want to ignore it in favor of your narrative that the system is unfair. ??

[Faux Play.6104]

The one big thing about the math behind this is the deviation. People are thinking I'm a 1425 rated player, but in reality it is a much larger band that you could be. The deviation is a measure of how accurately you know something. So if you are playing matches against teams with a rating of 1200, 68% will be within one deviation of that (a little more than 3 out of 5 matches), and 95% (19 out 20) will be within two deviations of that. When I was doing the math deviations were settling at around 60 regardless of what I did. That means you are likely to move a hundred points or so over the course of a season in either direction.

I made a sheet that would calculate match results based on an assumed rating of the player. They would be more likely to win matches vs. players lower than them and less likely to win if the team was higher than them. Each time the sheet is recalculated you get a different plot since the results are randomized, and based on the previous match results. As you can see people's ratings will vary quite a bit, and they all had at least one period of win/loss streaks. For the example I used a player with an assumed rating of 1600.

[Exedore.6320]

The deviation is why I find the tier levels and rigid tier boundaries less meaningful. Tier levels are too small to accurately place everyone, and the rigid boundaries means that someone near the end of a tier bounces between tiers often.

I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill. I would also like to see soft tier boundaries where you need to progress a certain amount part the soft boundary before your tier changes.

[Eddbopkins.2630]

If there's one thing anet can not do after 7 years, it's code a unbiased, and fair, competitive game mode.

[Chaith.8256]

@Exedore.6320 said:I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill.

As a community we cried out for years to give us a rating we can see with our eyes. The grass is always greener, don't forget. ArenaNet trying to react to player's wishlists, that's why PvP systems for 8 years have just been removed and replaced constantly, spending the entire development budget flip flopping between multiple equally flawed PvP systems.

[Faux Play.6104]

@Chaith.8256 said:

@Exedore.6320 said:I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill.

As a community we cried out for years to give us a rating we can see with our eyes. The grass is always greener, don't forget. ArenaNet trying to react to player's wishlists, that's why PvP systems for 8 years have just been removed and replaced constantly, spending the entire development budget flip flopping between multiple equally flawed PvP systems.

To me it is important for them to be open and transparent with the rating system. That way people can actually see why the system does what it does. No system is perfect, but the glicko/elo rating seems to be the consensus standard for handling ratings. No matter what rating system that is in effect, players will always think their rating is too low.

I still think ATs and monthly should count towards the season's ratings. That would help with the ratings on the top tier players, and allow people that only want to team queue a level playing to earn their ratings.

[Exedore.6320]

@Chaith.8256 said:

@Exedore.6320 said:I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill.

As a community we cried out for years to give us a rating we can see with our eyes. The grass is always greener, don't forget.I've played under systems with tires and hidden rating. as long as the boundaries are well placed, I prefer that.

ArenaNet trying to react to player's wishlists, that's why PvP systems for 8 years have just been removed and replaced constantly, spending the entire development budget flip flopping between multiple equally flawed PvP systems.

I wish they'd do less of that sometimes. The community is generally bad at knowing what they really want.

[Vancho.8750]

@Exedore.6320 said:

@Chaith.8256 said:

@Exedore.6320 said:I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill.

As a community we cried out for years to give us a rating we can see with our eyes. The grass is always greener, don't forget.I've played under systems with tires and hidden rating. as long as the boundaries are well placed, I prefer that.

ArenaNet trying to react to player's wishlists, that's why PvP systems for 8 years have just been removed and replaced constantly, spending the entire development budget flip flopping between multiple equally flawed PvP systems.

I wish they'd do less of that sometimes. The community is generally bad at knowing what they really want.

I always cringe when someone says "the community", you have many people with different desires you don't have "community".The elo/glicko doesn't work since it is based on single player numbers and we play 5v5 it is probably works for 2v2 when you pick your team mate, but for conquest is really terrible there. The matchmaker constantly puts you in terrible games win or lose only a handful of games are good since the population is low and it is low since the game is not fun and we go into a loop. The glicko system wants you to play huge amount of games to determine what is your skill level and at some point you will end up in a place where you should win all of the games which is statistically impossible and this is where people start playing the meta game, the game outside the game. So you get the Meta gamers creating smurfs to bring down other people down, top players playing in duos while the other top players dodge their games and other shenanigans that have nothing to do with playing the pvp match.That is why the tournaments are probably better ranking system then the ranked one.

[Exedore.6320]

@Vancho.8750 said:

@"Exedore.6320" said:I wish they'd do less of that sometimes. The community is generally bad at knowing what they really want.

I always cringe when someone says "the community", you have many people with different desires you don't have "community".The elo/glicko doesn't work since it is based on single player numbers and we play 5v5 it is probably works for 2v2 when you pick your team mate, but for conquest is really terrible there.You're doing a great job of illustrating my point with your uninformed opinion.

Glicko/Elo may not be perfect, but it's been shown by basically every online game that it's still a reliable system and better than all others to date.

[megilandil.7506]

@Vancho.8750 said:

@Exedore.6320 said:

@Chaith.8256 said:

@Exedore.6320 said:I would prefer hidden rating, allowing ANet to adjust tier boundaries to better differentiate skill.

As a community we cried out for years to give us a rating we can see with our eyes. The grass is always greener, don't forget.I've played under systems with tires and hidden rating. as long as the boundaries are well placed, I prefer that.

ArenaNet trying to react to player's wishlists, that's why PvP systems for 8 years have just been removed and replaced constantly, spending the entire development budget flip flopping between multiple equally flawed PvP systems.

I wish they'd do less of that sometimes. The community is generally bad at knowing what they really want.

I always cringe when someone says "the community", you have many people with different desires you don't have "community".The elo/glicko doesn't work since it is based on single player numbers and we play 5v5 it is probably works for 2v2 when you pick your team mate, but for conquest is really terrible there. The matchmaker constantly puts you in terrible games win or lose only a handful of games are good since the population is low and it is low since the game is not fun and we go into a loop. The glicko system wants you to play huge amount of games to determine what is your skill level and at some point you will end up in a place where you should win all of the games which is statistically impossible and this is where people start playing the meta game, the game outside the game. So you get the Meta gamers creating smurfs to bring down other people down, top players playing in duos while the other top players dodge their games and other shenanigans that have nothing to do with playing the pvp match.That is why the tournaments are probably better ranking system then the ranked one.

Just the contrary, the system works in a total random scenario, when you are able to team up you hide your personal skill behind a "team skill", and no, the system wants to put you in a place where you win 50% of the matches not the 100%.

For making the system work propper to get acurate personal ratings they have to remove from the equation duos trios or whatever(possible) and also remove alt acounts from the league(hard to achieve )

[Vancho.8750]

@Exedore.6320 said:

@Vancho.8750 said:

@Exedore.6320 said:I wish they'd do less of that sometimes. The community is generally bad at knowing what they really want.

I always cringe when someone says "the community", you have many people with different desires you don't have "community".The elo/glicko doesn't work since it is based on single player numbers and we play 5v5 it is probably works for 2v2 when you pick your team mate, but for conquest is really terrible there.You're doing a great job of illustrating my point with your uninformed opinion.

Glicko/Elo may not be perfect, but it's been shown by basically every online game that it's still a reliable system and better than all others to date.

UGH sorry for my filthy uninformed peasantry beliefs that chess single player system doesn't work on random multiplayer game without taking into consideration anything besides winning.I want to play the fucking game not some broken bullshit outside the game.