Lost 2 placement matches in a row to same player deliberately throwing

[Bandlero.6312]

> @"ReaverKane.7598" said:

>

> > @"phokus.8934" said:

> > Your rating is directly related to last seasons rating. Placement doesn't mean a whole lot and there's not a lot of volatility with them.

> Actually, no, according to @"Ben Phongluangtham.1065" the matchmaker "soft-resets" everyone's rating to a average score, i think around low gold, iirc, that varies a bit depending on your previous standing, but it doesn't influence your standing as much as, say in LoL, and other games.

 

I was trying to find the actual post to link, but can't seem to find it - maybe somewhere in this thread (https://en-forum.guildwars2.com/discussion/19539/pvp-discussion-matchmaking-and-leagues/p3). He mentions that "average" rating is 1200. But yea, previous seasons' ratings do add a little to your current starting rating, but its very minimal like 50 or less I think Ben said. I can't remember exact specifics or find the post.

 

[SlippyCheeze.5483]

> @"Bandlero.6312" said:

> > @"SlippyCheeze.5483" said:

> >

> > That would be the MMR system having an accurate assessment of your MMR. It never *stops* changing, it just bounces around the actual value, a little up, a little down, delivering as close to balanced matches as possible.

> >

> > So, that isn't exactly "a tier purgatory", but rather, "your skill level purgatory"; even if you believe you should be rated higher, the MMR doesn't agree, and I'm inclined to trust the unemotional machine over your personal impressions, I gotta say.

> >

> Of course totally ignoring that your rating is not actually based upon your actual solo, individual skill rating, but upon the performance rating of sets of compiled teams of "like skilled" players, drawn together "at random." Your rating is actually a measurement of the performance of the teams you've been on. An improper assumption to make is that, because you've been on every team, you are the (sole) cause of your rating.

 

Eh, the bias from random grouping is in your favor: if your skill is higher than your MMR, you will on average be placed on a team with four average and one good player, against five average players. It's a small advantage, but a real one, and will lead to -- yes -- your individual skill causing your MMR to rise, along with the average players who are placed with you.

 

Is GLEIKO perfect for this? Nope. It'll take longer than would be desirable to correct if you get extremely unlucky, but on average that'll still be around ten to, worst case, fifteen games before you are in the right region for your MMR.

 

I appreciate that you sincerely believe that the MMR is inaccurate, and you are being given a lower rating than you deserve through flaws in the system, rather than it being an accurate reflection of your skill, but the math doesn't favor that claim.

[Rufo.3716]

I don't even know why they call them placement matches. It's not as if everybody starts off at the same point and gets placed from there.

[ReaverKane.7598]

> @"Rufo.3716" said:

> I don't even know why they call them placement matches. It's not as if everybody starts off at the same point and gets placed from there.

 

Actually they kinda do, read above...

 

And even if they didn't (like in all other games), you're still being placed in a league, according to your past MMR and your outcome of the 10 placement matches.

[phokus.8934]

> @"ReaverKane.7598" said:

> If only you could avoid this somehow, like, i don't know, bringing teams back...

> Here's how you handle toxicity in GW2: You don't...

> You hide in a corner and hope it goes away! Or do what i did and stop forcing yourself through that terrible experience. There's simply no tools to prevent people from spoiling your game, because Arena Net cares only the bare minimum to make an appearance that it cares about PvP.

> In other games you can accurately report people, and see reports actioned, you can block people and not be placed with them on your team, **you can make your own team by inviting friends or random players**.

> All this allows you to be able to enjoy the game despite the inherent Toxicity of PvP modes.

> In GW2, Arena Net is still convinced they have the "best community in the world", which is true, for the most part, in PvE, but PvP? I play League of Legends **a lot**, and that game is infamous for it's toxic community, and yet i enjoy myself much more there, and it's incredibly rare for me to get upset from my team's behaviour, while until i quit GW2's PvP, it rarely came a day that someone wasn't throwing matches left and right, or just complaining about inane stuff. Or simply being frustrated because the game has no balance.

>

> > @"phokus.8934" said:

> > Your rating is directly related to last seasons rating. Placement doesn't mean a whole lot and there's not a lot of volatility with them.

> Actually, no, according to @"Ben Phongluangtham.1065" the matchmaker "soft-resets" everyone's rating to a average score, i think around low gold, iirc, that varies a bit depending on your previous standing, but it doesn't influence your standing as much as, say in LoL, and other games.

Yeah that's the thing, placements don't really mean much in this game. When I finish in Plat last season and start in G3 after placements (going 7/3) and winning two games after granting me 28 points per win, placements are kind of useless.

 

 

[AegisFLCL.7623]

I'll say it again for all you people who think this system can calculate a single players MMR from 1/10 player pool, rating can not, will not, ever be accurate. It is a horrible estimate based on random factors; tilted players, dc's, toxic players, low player pop @ certain rating levels, MMR mismatch within teams due to low pop @ high ratings, you being only 1 of 5 players on your team (2 if you're below plat), class swapping, class stacking, win traders.... do I have to go on?

 

The closest way to estimate a players skill using GW2's modified Glicko system would be based on team queues of 2v2, 3v3, or 5v5.

[SweetPotato.7456]

> @"Hoodie.1045" said:

> > @"Sinful.2165" said:

> > Why anet? Just. Why.

> >

> > Why would you give a loss in a placement match when a player sits in a corner and does absolutely nothing?

> >

> > IMO upgrade your algorithm to give the person throwing the match all 5 losses and let everyone else try again on a more level playing field. *frustrated*

>

> You can always block the player who was throwing on purpose. That way you'll know when they're playing PvP and you can avoid playing with them. People who throw matches suck and blocking them to know when they're online playing PvP is the only way to fix the problem of win traders aside from reporting them for match manipulation.

 

that is no a good solution at all. why do we need to avoide these trolls, they should be the one not allow to play

[Ithilwen.1529]

The advice I was given was: Wait a week-or-so to play placements. That gives time for things to settle down.

[FyzE.3472]

> @"Shadowzerk.4715" said:

> i just lost 5 match in a a row, cos we got 1~2 ppl dont know how to play and wanna to get their "dailies" done...... anet need to remove dailies from "ranked" match :/

 

Who the f does dailies in ranked? IF that's the only goal, just do them in unranked! You can even play a troll build there! I don't understand... Feel bad for you man..

[Bigpapasmurf.5623]

They are quick to remove posts that tell ppl to get good, but not so much that actually affect their paying player's in-game experience and those breaking their rules (despite those who have proof via screenshots and video)

[SlippyCheeze.5483]

> @"AegisFLCL.7623" said:

> I'll say it again for all you people who think this system can calculate a single players MMR from 1/10 player pool, rating can not, will not, ever be accurate. It is a horrible estimate based on random factors; tilted players, dc's, toxic players, low player pop @ certain rating levels, MMR mismatch within teams due to low pop @ high ratings, you being only 1 of 5 players on your team (2 if you're below plat), class swapping, class stacking, win traders.... do I have to go on?

 

...and yet, after a little while, it tends to settle out into a fairly small range of movement for all the players involved, as if it were circling around a specific value. How could that happen, if it is so dominated by these random confounding factors?

 

> The closest way to estimate a players skill using GW2's modified Glicko system would be based on team queues of 2v2, 3v3, or 5v5.

 

You mean, estimating the specific team composition skill, and only for 5v5, with premades on *both* sides, right? Because otherwise those exact same confounding factors remain present: 2 people queue together, plus three random, bam. 5 person premade vs random team, bam. etc.

[AegisFLCL.7623]

> @"SlippyCheeze.5483" said:

> > @"AegisFLCL.7623" said:

> > I'll say it again for all you people who think this system can calculate a single players MMR from 1/10 player pool, rating can not, will not, ever be accurate. It is a horrible estimate based on random factors; tilted players, dc's, toxic players, low player pop @ certain rating levels, MMR mismatch within teams due to low pop @ high ratings, you being only 1 of 5 players on your team (2 if you're below plat), class swapping, class stacking, win traders.... do I have to go on?

>

> ...and yet, after a little while, it tends to settle out into a fairly small range of movement for all the players involved, as if it were circling around a specific value. How could that happen, if it is so dominated by these random confounding factors?

>

> > The closest way to estimate a players skill using GW2's modified Glicko system would be based on team queues of 2v2, 3v3, or 5v5.

>

> You mean, estimating the specific team composition skill, and only for 5v5, with premades on *both* sides, right? Because otherwise those exact same confounding factors remain present: 2 people queue together, plus three random, bam. 5 person premade vs random team, bam. etc.

 

In terms of queuing together, you misunderstood my second statement. I mean only 4 players in the map for 2v2, only 6 for a 3v3, and so on. Only full teams are allowed to queue in those ranked categories. Of course it is not perfect but it is far more accurate than what we have now as you eliminate the majority of random factors due to queuing with unknown players.

 

By the nature of the game and it's mechanics, the rating will never be accurate. Team queue only for ranked would be simply be the best option in terms of accuracy. We all know the player population wouldn't be able to support this model however, thus we are forever at the mercy of RNG.

[Sinful.2165]

I agree that team queues would represent a much more accurate rating of individual skill rating. Nevermind that your comp might be hard countered by another enemy comp, you all have the option of changing classes during the ready phase. It would always come down to each player's skill level comprised as part of a team.

 

Bring back team queues IMO! ;) but then that would encourage community and discourage the immensely toxic environment that sPvP currently is. No, that doesn't make sense at all.. *cough* *cough* *hint* *hint* @"Ben Phongluangtham.1065"

[SlippyCheeze.5483]

> @"AegisFLCL.7623" said:

> > @"SlippyCheeze.5483" said:

> > > @"AegisFLCL.7623" said:

> > > I'll say it again for all you people who think this system can calculate a single players MMR from 1/10 player pool, rating can not, will not, ever be accurate. It is a horrible estimate based on random factors; tilted players, dc's, toxic players, low player pop @ certain rating levels, MMR mismatch within teams due to low pop @ high ratings, you being only 1 of 5 players on your team (2 if you're below plat), class swapping, class stacking, win traders.... do I have to go on?

> >

> > ...and yet, after a little while, it tends to settle out into a fairly small range of movement for all the players involved, as if it were circling around a specific value. How could that happen, if it is so dominated by these random confounding factors?

> >

> > > The closest way to estimate a players skill using GW2's modified Glicko system would be based on team queues of 2v2, 3v3, or 5v5.

> >

> > You mean, estimating the specific team composition skill, and only for 5v5, with premades on *both* sides, right? Because otherwise those exact same confounding factors remain present: 2 people queue together, plus three random, bam. 5 person premade vs random team, bam. etc.

>

> In terms of queuing together, you misunderstood my second statement. I mean only 4 players in the map for 2v2, only 6 for a 3v3, and so on. Only full teams are allowed to queue in those ranked categories. Of course it is not perfect but it is far more accurate than what we have now as you eliminate the majority of random factors due to queuing with unknown players.

 

Oh. Yes, that would eliminate the problem, though it would also bring in more challenges around balancing all vs all in terms of specs: the less people in the team, the less chance that you will cover any gap in any individual spec. I don't have the skill to know where that trade-off stops being worth it.

 

> By the nature of the game and it's mechanics, the rating will never be accurate. Team queue only for ranked would be simply be the best option in terms of accuracy. We all know the player population wouldn't be able to support this model however, thus we are forever at the mercy of RNG.

 

It is true, MMR remains an estimate, and there will always be some degree of randomness in it, even in the most controlled of settings. (Because a Comcast outage in CA could drop someone from your team, leading to a loss, ouchies, and that is ... well, random enough, I guess.)

 

I'm two hundred percent behind the idea of a better system than GLICKO-2 being (a) possible, and (b) desirable. I'm honestly disappointed every time I go poke around for this stuff and find that neither academia or engineering have delivered anything that seems to be an improvement on the state of the art -- which is either GLICKO-2 or Microsoft's TrueSkill, a question which currently has no clear winner.

 

I'm even open to the idea that while all the evidence people cite here about MMR problems seems to show that GLICKO-2 is doing what it does, and basically working correctly, there is a systematic problem with GW2 PvP. I just want, personally, a higher standard of proof than we have so far.

 

If you are really concerned, and it seems you are, one way to combat this would be to use the GW2 API which can give you the stats for your last ten matches, and use that to see if you can identify a systematic discrepancy. If, over a few hundred matches, there are significantly more "unexpected loss" or significantly less "unexpected win" results than their counterparts, you have a very good case for asking ANet why this is happening.

 

Do that across more than one person, and you have an even better case for their being systematic bias, and getting it eliminated. ANet (and I) can ignore opinions if things "look right", but nobody can argue that someone showing solid statistics, with a robust measure of how they determined them, and a mechanism to replicate the results, isn't on solid ground.

 

(I know, it's hard work to do that. I suspect someone with a bit of enthusiasm and JavaScript could get it running in a Google Sheets / Scripts environment pretty quickly, and capture the raw data though...)

[AegisFLCL.7623]

> @"SlippyCheeze.5483" said:

> > @"AegisFLCL.7623" said:

> > > @"SlippyCheeze.5483" said:

> > > > @"AegisFLCL.7623" said:

> > > > I'll say it again for all you people who think this system can calculate a single players MMR from 1/10 player pool, rating can not, will not, ever be accurate. It is a horrible estimate based on random factors; tilted players, dc's, toxic players, low player pop @ certain rating levels, MMR mismatch within teams due to low pop @ high ratings, you being only 1 of 5 players on your team (2 if you're below plat), class swapping, class stacking, win traders.... do I have to go on?

> > >

> > > ...and yet, after a little while, it tends to settle out into a fairly small range of movement for all the players involved, as if it were circling around a specific value. How could that happen, if it is so dominated by these random confounding factors?

> > >

> > > > The closest way to estimate a players skill using GW2's modified Glicko system would be based on team queues of 2v2, 3v3, or 5v5.

> > >

> > > You mean, estimating the specific team composition skill, and only for 5v5, with premades on *both* sides, right? Because otherwise those exact same confounding factors remain present: 2 people queue together, plus three random, bam. 5 person premade vs random team, bam. etc.

> >

> > In terms of queuing together, you misunderstood my second statement. I mean only 4 players in the map for 2v2, only 6 for a 3v3, and so on. Only full teams are allowed to queue in those ranked categories. Of course it is not perfect but it is far more accurate than what we have now as you eliminate the majority of random factors due to queuing with unknown players.

>

> Oh. Yes, that would eliminate the problem, though it would also bring in more challenges around balancing all vs all in terms of specs: the less people in the team, the less chance that you will cover any gap in any individual spec. I don't have the skill to know where that trade-off stops being worth it.

>

> > By the nature of the game and it's mechanics, the rating will never be accurate. Team queue only for ranked would be simply be the best option in terms of accuracy. We all know the player population wouldn't be able to support this model however, thus we are forever at the mercy of RNG.

>

> It is true, MMR remains an estimate, and there will always be some degree of randomness in it, even in the most controlled of settings. (Because a Comcast outage in CA could drop someone from your team, leading to a loss, ouchies, and that is ... well, random enough, I guess.)

>

> I'm two hundred percent behind the idea of a better system than GLICKO-2 being (a) possible, and (b) desirable. I'm honestly disappointed every time I go poke around for this stuff and find that neither academia or engineering have delivered anything that seems to be an improvement on the state of the art -- which is either GLICKO-2 or Microsoft's TrueSkill, a question which currently has no clear winner.

>

> I'm even open to the idea that while all the evidence people cite here about MMR problems seems to show that GLICKO-2 is doing what it does, and basically working correctly, there is a systematic problem with GW2 PvP. I just want, personally, a higher standard of proof than we have so far.

>

> If you are really concerned, and it seems you are, one way to combat this would be to use the GW2 API which can give you the stats for your last ten matches, and use that to see if you can identify a systematic discrepancy. If, over a few hundred matches, there are significantly more "unexpected loss" or significantly less "unexpected win" results than their counterparts, you have a very good case for asking ANet why this is happening.

>

> Do that across more than one person, and you have an even better case for their being systematic bias, and getting it eliminated. ANet (and I) can ignore opinions if things "look right", but nobody can argue that someone showing solid statistics, with a robust measure of how they determined them, and a mechanism to replicate the results, isn't on solid ground.

>

> (I know, it's hard work to do that. I suspect someone with a bit of enthusiasm and JavaScript could get it running in a Google Sheets / Scripts environment pretty quickly, and capture the raw data though...)

 

Until ANET decides to be entirely transparent, and release public data on complete mmr history for players, it is nearly impossible to prove right or wrong. Call me a conspiracy theorist, but after several discussions with ANET input on the topic or similar topics, they know there are problems but a) don't think it's worth investing the time and or b) don't realize their design choices are deliberately hurting the way the Glicko system "should" work.

 

ANET has proven time and time again they have a poor track record of consistency for any game content in GW2, which directly relates to their poor internal management. I have zero faith they will do anything beneficial for PvP and or WvW at this point in the games age. Improving PvP, with it's ever dwindling population and toxic environment, is 100% not worth it from a business standpoint, let alone having a small team of 3-5 people to balance the entire game and address PvP concerns.

 

Don't get me wrong, I love the game and genuinely agree with the majority of yours and other players discussions on the PvP system, the chances of improvements are just slim to none. Ranked is a mess and I primarily see it as a means for pve, pvp, and wvw players to farm gold through pips with very little concern for the ranked systems actual purpose.

[omgdracula.6345]

> @"Hoodie.1045" said:

> > @"Sinful.2165" said:

> > Why anet? Just. Why.

> >

> > Why would you give a loss in a placement match when a player sits in a corner and does absolutely nothing?

> >

> > IMO upgrade your algorithm to give the person throwing the match all 5 losses and let everyone else try again on a more level playing field. *frustrated*

>

> You can always block the player who was throwing on purpose. That way you'll know when they're playing PvP and you can avoid playing with them. People who throw matches suck and blocking them to know when they're online playing PvP is the only way to fix the problem of win traders aside from reporting them for match manipulation.

 

That is just giving people who throw all the power. So OP is supposed to hop on and go? Welp Jerky McJerkface is playing so can't play pvp today! Cool! That is just not a good solution for anyone.

[Sinful.2165]

I did not really agree that there was a small PvP population but this season has been very enlightening. I consistently see the same people in unranked all day. In ranked I also run in to the same small pool of players from match to match even climbing all the way from gold 1 to plat 1 it is still the same names.

 

Being top 250 in upper gold seems like a joke right now, I won't even enable my PvP badge while gold. People are getting placed near plat 1 straight out of the gate. I get that it's only a few days in to the season but this is currently one of the best options available for both MMORPGs and PvP in general. GW2 doesn't have much in terms of competition at the moment. Where the heck is everyone?

 

Fix your PvP anet, pretty pretty please?

[Megametzler.5729]

The season is out for like four days... relax, mate. ;)

 

It is true, of course, you will probably recognise some people over time. That does not trouble me though.

[SlippyCheeze.5483]

> @"AegisFLCL.7623" said:

> > @"SlippyCheeze.5483" said:

> > If you are really concerned, and it seems you are, one way to combat this would be to use the GW2 API which can give you the stats for your last ten matches, and use that to see if you can identify a systematic discrepancy. If, over a few hundred matches, there are significantly more "unexpected loss" or significantly less "unexpected win" results than their counterparts, you have a very good case for asking ANet why this is happening.

> >

> > Do that across more than one person, and you have an even better case for their being systematic bias, and getting it eliminated. ANet (and I) can ignore opinions if things "look right", but nobody can argue that someone showing solid statistics, with a robust measure of how they determined them, and a mechanism to replicate the results, isn't on solid ground.

> >

> > (I know, it's hard work to do that. I suspect someone with a bit of enthusiasm and JavaScript could get it running in a Google Sheets / Scripts environment pretty quickly, and capture the raw data though...)

>

> Until ANET decides to be entirely transparent, and release public data on complete mmr history for players, it is nearly impossible to prove right or wrong. Call me a conspiracy theorist, but after several discussions with ANET input on the topic or similar topics, they know there are problems but a) don't think it's worth investing the time and or b) don't realize their design choices are deliberately hurting the way the Glicko system "should" work.

 

That's why I suggested the polling of the API; you can collect and store the data yourself. I don't think you can run through ten complete PvP matches in the five minutes the API caches data for, so you shouldn't miss out on anything. ;)

 

> Don't get me wrong, I love the game and genuinely agree with the majority of yours and other players discussions on the PvP system, the chances of improvements are just slim to none. Ranked is a mess and I primarily see it as a means for pve, pvp, and wvw players to farm gold through pips with very little concern for the ranked systems actual purpose.

 

*nod* I, also, would love to see problems resolved. We don't necessarily agree on what they are, but I think your efforts to improve things are absolutely valid, reasonable, and worth celebrating. If you are right, and I'm wrong, I'll be just as happy when things get fixed. :)

[Thais.4639]

> @"Bandlero.6312" said:

> tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

 

Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

 

Last season I started late and only had one or two placement matches with blatant manipulators. So far this season every single one has had match manipulators (sometimes my team, sometimes enemy). It sucks nine ways from Sunday, because my rank and my skill are only tenuously connected. At this point I'm only playing ranked for rewards - I can have more fun and get better matches in Unranked usually.

[SlippyCheeze.5483]

> @"Thais.4639" said:

> > @"Bandlero.6312" said:

> > tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

>

> Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

 

That is true, but ... isn't that exactly what **should** be happening? If you are winning exactly half the time, you are playing balanced matches. (or, at least, a balance of matches where you win or lose equally often.)

 

This sounds like a complaint that an effective MMR system is effective, to me, and I can't tell why you think that is an issue?

 

[Thais.4639]

> @"SlippyCheeze.5483" said:

> > @"Thais.4639" said:

> > > @"Bandlero.6312" said:

> > > tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

> >

> > Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

>

> That is true, but ... isn't that exactly what **should** be happening? If you are winning exactly half the time, you are playing balanced matches. (or, at least, a balance of matches where you win or lose equally often.)

>

> This sounds like a complaint that an effective MMR system is effective, to me, and I can't tell why you think that is an issue?

>

 

Re-read my comment please. The point is that placement matches are by far the most important matches you play, because they set you up for the seasonal MMR, and so the consequences of placement being gamed are far higher than meets the eye.

[SlippyCheeze.5483]

> @"Thais.4639" said:

> > @"SlippyCheeze.5483" said:

> > > @"Thais.4639" said:

> > > > @"Bandlero.6312" said:

> > > > tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

> > >

> > > Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

> >

> > That is true, but ... isn't that exactly what **should** be happening? If you are winning exactly half the time, you are playing balanced matches. (or, at least, a balance of matches where you win or lose equally often.)

> >

> > This sounds like a complaint that an effective MMR system is effective, to me, and I can't tell why you think that is an issue?

>

> Re-read my comment please. The point is that placement matches are by far the most important matches you play, because they set you up for the seasonal MMR, and so the consequences of placement being gamed are far higher than meets the eye.

 

I have done so, and I don't see anything in there the second time that contradicts what I saw the first time: a working MMR system will keep your rank stable with a 50 percent winrate. This is correct and intended behaviour.

 

To the comment above yours, which you were responding to, two players with the same win/loss counts, but different ranks is perfectly reasonable: a "1400 MMR" player and a "1700 MMR" player *should* stay equally distant if they both have the exact same win/loss count. (and expected win/loss count, and confidence, of course)

 

You should see significant change in MMR when it doesn't reflect skill. Absolute win/loss count is irrelevant, as the MMR system in GW2 (and, honestly, everywhere) is in no way a measure of the count of victories or losses, but rather, of player skill. An estimate, sure, one that can be wrong, absolutely, but none of that makes "placement matches" in any way significant.

 

A "1400 MMR" player who, through happenstance, places at 1700 when they exit will not maintain a 20:12 win:loss ratio, as their skill will contribute to the team losing, and their MMR will consequently drop until they reach their level. If they **do** maintain that ratio, vs a someone placed at 1400, then they have demonstrated the skill to maintain that level.

 

I'm certainly open to you offering additional information that might change this situation, but I do not see anything in your comment that changes this on a second reading.

[Thais.4639]

> @"SlippyCheeze.5483" said:

> > @"Thais.4639" said:

> > > @"SlippyCheeze.5483" said:

> > > > @"Thais.4639" said:

> > > > > @"Bandlero.6312" said:

> > > > > tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

> > > >

> > > > Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

> > >

> > > That is true, but ... isn't that exactly what **should** be happening? If you are winning exactly half the time, you are playing balanced matches. (or, at least, a balance of matches where you win or lose equally often.)

> > >

> > > This sounds like a complaint that an effective MMR system is effective, to me, and I can't tell why you think that is an issue?

> >

> > Re-read my comment please. The point is that placement matches are by far the most important matches you play, because they set you up for the seasonal MMR, and so the consequences of placement being gamed are far higher than meets the eye.

>

> I have done so, and I don't see anything in there the second time that contradicts what I saw the first time: a working MMR system will keep your rank stable with a 50 percent winrate. This is correct and intended behaviour.

>

> To the comment above yours, which you were responding to, two players with the same win/loss counts, but different ranks is perfectly reasonable: a "1400 MMR" player and a "1700 MMR" player *should* stay equally distant if they both have the exact same win/loss count. (and expected win/loss count, and confidence, of course)

>

> You should see significant change in MMR when it doesn't reflect skill. Absolute win/loss count is irrelevant, as the MMR system in GW2 (and, honestly, everywhere) is in no way a measure of the count of victories or losses, but rather, of player skill. An estimate, sure, one that can be wrong, absolutely, but none of that makes "placement matches" in any way significant.

>

> A "1400 MMR" player who, through happenstance, places at 1700 when they exit will not maintain a 20:12 win:loss ratio, as their skill will contribute to the team losing, and their MMR will consequently drop until they reach their level. If they **do** maintain that ratio, vs a someone placed at 1400, then they have demonstrated the skill to maintain that level.

>

> I'm certainly open to you offering additional information that might change this situation, but I do not see anything in your comment that changes this on a second reading.

 

It seems to me your source of confusion is thinking we're talking about the same thing. We aren't. We are talking about two different things.

[SlippyCheeze.5483]

> @"Thais.4639" said:

> > @"SlippyCheeze.5483" said:

> > > @"Thais.4639" said:

> > > > @"SlippyCheeze.5483" said:

> > > > > @"Thais.4639" said:

> > > > > > @"Bandlero.6312" said:

> > > > > > tldr; Your placement matches matter the most. In-fact, your placement matches might be the only matches that matter. That's why in the leaderboards you can see people with rating 1400 at 20 wins and 10 losses and people with rating 1700 with 20 wins and 10 losses. What's the difference? The rating 1700 players won more placement matches. This is why placement matches have become so rife with manipulation and trading.

> > > > >

> > > > > Every ranked online game which uses MMR that I'm aware of does this with placement. Your placement matches are by FAR the most important ones you will play if you care about your rank - in fact with a strict 50% winrate after placement, your rank should barely fluctuate all season. (if you don't care about rank, then the farm aspect of Ranked PvP suggests that you want to drive your rank down as far as possible to make the pip farm easier).

> > > >

> > > > That is true, but ... isn't that exactly what **should** be happening? If you are winning exactly half the time, you are playing balanced matches. (or, at least, a balance of matches where you win or lose equally often.)

> > > >

> > > > This sounds like a complaint that an effective MMR system is effective, to me, and I can't tell why you think that is an issue?

> > >

> > > Re-read my comment please. The point is that placement matches are by far the most important matches you play, because they set you up for the seasonal MMR, and so the consequences of placement being gamed are far higher than meets the eye.

> >

> > I have done so, and I don't see anything in there the second time that contradicts what I saw the first time: a working MMR system will keep your rank stable with a 50 percent winrate. This is correct and intended behaviour.

> >

> > To the comment above yours, which you were responding to, two players with the same win/loss counts, but different ranks is perfectly reasonable: a "1400 MMR" player and a "1700 MMR" player *should* stay equally distant if they both have the exact same win/loss count. (and expected win/loss count, and confidence, of course)

> >

> > You should see significant change in MMR when it doesn't reflect skill. Absolute win/loss count is irrelevant, as the MMR system in GW2 (and, honestly, everywhere) is in no way a measure of the count of victories or losses, but rather, of player skill. An estimate, sure, one that can be wrong, absolutely, but none of that makes "placement matches" in any way significant.

> >

> > A "1400 MMR" player who, through happenstance, places at 1700 when they exit will not maintain a 20:12 win:loss ratio, as their skill will contribute to the team losing, and their MMR will consequently drop until they reach their level. If they **do** maintain that ratio, vs a someone placed at 1400, then they have demonstrated the skill to maintain that level.

> >

> > I'm certainly open to you offering additional information that might change this situation, but I do not see anything in your comment that changes this on a second reading.

>

> It seems to me your source of confusion is thinking we're talking about the same thing. We aren't. We are talking about two different things.

 

That would certainly do it. I'm sorry to have misunderstood your point. I'm fine if you don't want to, but I'd be interested if you could try restating it in a different way, so perhaps I can actually understand what you are saying? Obviously, up to you, and I hold no grudge if you don't want to invest that time.