Jump to content
  • Sign Up

Change Rating gain/loss


nativity.3057

Recommended Posts

@hotte in space.2158 said:

The system would reduce the amount of afk players cause it would matter to gain more score to not lose so much rating.

Today I had a troll player in my team who was following me the whole match, but without fighting. He used very obscene names for his ranger pets btw.Eventually a bot ??? I dont know, but the next match he was in the opponent team and he was actively playing then.These kind of players dont care about loosing rating, and I think afk players in generally dont care for anything.

Apart from that, imagine you win 3 close matches, hard ones. And then you loose one very high, no matter for what reason. You would have won 0 points then. lolI would rather knit myself a bobble hat then

Well, at least you'd have some chance to affect the end rating of the match cause now you just auto lose -14 when you have an afk player and there's nothing you can do to reduce that loss.

Link to comment
Share on other sites

@nativity.3057 said:

@Tycura.1982 said:@"Ragnar.4257" said:So, with your proposal, players with win-rate greater than 50% will keep on gaining rating infinitely.

This means that rating just becomes a measure of "games played" rather than "skill".

A gold-3 who plays 500 games will have a higher rating than a legendary mAT winner who only plays 200 games.

You might as well just get rid of rating all together, and just have a "number of games played" leaderboard.

Do you want this?

Think things through.

At a certain point, the gold 3 player will reach a cap of winning 50% of the time. A gold 3 player with a higher rating will be playing against players who are consistently plat 3. They won't achieve a 50%+ winrate consistently, rather, they'll see their winrate going lower and lower. If we set ratings gained and lost equal to each other, at 50% winrate, that player will stop climbing.

You are leaving out so many issues though.

A gold 3 players will get paired with plat 3 players, true. It's not the gold 3 players who will take a huge rating hit though if he loses. It's the plat 3 players who will take that hit if he loses. As such, yes, the more unbalanced the population is (as in total players playing), the more your win rating will shift below 50-50 potentially. Yet your rating might not drop.

If you are losing more rating than you are gaining on average, no matter the win-rate, you are above your correct MMR rating (give or take some lose and win streaks).

@nativity.3057 said:

@"Sikieiki.3189" said:We already had the system you are proposing in season 1-4 and it got changed to the current systetm because people took the leaderboards as a joke grindfest.

But the top players were still on the top, and the average players were still at average rankings. Not to mention we have better rewards for participating in PvP.

No, your proposed change would keep the top players on top, and the average players would simply climb further than they should be according to the current system.

@nativity.3057 said:I do think that generally, there would be a shift in ratings for all players, pushing them up a division (G3->P1, P1->P2). The benchmark of being a decent player would just move up a division as well.

Exactly, your proposed change would simply improve the loot and rewards for average players beyond what is intended currently. If that is the goal, you can simply propose a change to rewards instead of the underlying competitive matchmaking.

Link to comment
Share on other sites

@Cyninja.2954 said:

You are leaving out so many issues though.

A gold 3 players will get paired with plat 3 players, true. It's not the gold 3 players who will take a huge rating hit though if he loses. It's the plat 3 players who will take that hit if he loses. As such, yes, the more unbalanced the population is (as in total players playing), the more your win rating will shift below 50-50 potentially. Yet your rating might not drop.

In the current system, if a Plat 3 player gets matched with a Gold 3 player and loses, they would then have to win at least 2, even 3 more games in a row to break even on the ratings lost. Equalizing the ratings gained/lost would make that one defeat less significant.

If you are losing more rating than you are gaining on average, no matter the win-rate, you are above your correct MMR rating (give or take some lose and win streaks).

What are you saying with this? If you mean a single win/loss, no that wouldn't make sense because the current system has a bottleneck in place in Plat 2+. If you are talking about throughout the whole season, then yes, this would stay the same in the proposed change.

No, your proposed change would keep the top players on top, and the average players would simply climb further than they should be according to the current system.

The average player would only climb if they play more PvP. So we would hope to see an increase in the queue population as well as improvement in gameplay that comes from playing more games.

Exactly, your proposed change would simply improve the loot and rewards for average players beyond what is intended currently. If that is the goal, you can simply propose a change to rewards instead of the underlying competitive matchmaking.

We already have a good reward system in place. Haphazardly increasing loot would just increase the number of gold farming bots.

Link to comment
Share on other sites

plat and up this is an issue, but not lower than that, i have 2 accounts one in plat and one in gold 2 where i like to play non meta builds. While in plat I basically only play the matches required to be in leaderboard and like this season, not getting decay. The gold 2 one with 55% wr i got almost 150 pnts abover from where i started

Link to comment
Share on other sites

@nativity.3057 said:

You are leaving out so many issues though.

A gold 3 players will get paired with plat 3 players, true. It's not the gold 3 players who will take a huge rating hit though if he loses. It's the plat 3 players who will take that hit if he loses. As such, yes, the more unbalanced the population is (as in total players playing), the more your win rating will shift below 50-50 potentially. Yet your rating might not drop.

In the current system, if a Plat 3 player gets matched with a Gold 3 player and loses, they would then have to win at least 2, even 3 more games in a row to break even on the ratings lost. Equalizing the ratings gained/lost would make that one defeat less significant.

You seem to misunderstand how big the skill gap is between a plat 3 and gold 3 players. Hint: it's massive.

@nativity.3057 said:

@Cyninja.2954 said:If you are losing more rating than you are gaining on average, no matter the win-rate, you are above your correct MMR rating (give or take some lose and win streaks).

What are you saying with this? If you mean a single win/loss, no that wouldn't make sense because the current system has a bottleneck in place in Plat 2+. If you are talking about throughout the whole season, then yes, this would stay the same in the proposed change.

No it would not. Not sure how you can even suggest this when you yourself are admitting your change would lead to across the board rating increase for everyone.

@nativity.3057 said:

@Cyninja.2954 said:No, your proposed change would keep the top players on top, and the average players would simply climb further than they should be according to the current system.

The average player would only climb if they play more PvP. So we would hope to see an increase in the queue population as well as improvement in gameplay that comes from playing more games.

No, the average player would climb higher in the current ranking comparative to the total rank possible. Unless you intend more ranks above plat and below legend.

@nativity.3057 said:

@Cyninja.2954 said:Exactly, your proposed change would simply improve the loot and rewards for average players beyond what is intended currently. If that is the goal, you can simply propose a change to rewards instead of the underlying competitive matchmaking.

We already have a good reward system in place. Haphazardly increasing loot would just increase the number of gold farming bots.

Then haphazardly increasing everyone's rating would be just as detrimental.

Link to comment
Share on other sites

@"Cyninja.2954" said:

What does the skill gap have to do with a player needing to win 2~3 more games after a loss? You don't seem to be reading what I am writing. I'm not misunderstanding anything, I know how big the skill gap is.In this situation, it's more likely a Plat 3 player will be on the same team as a Gold 3 player facing an enemy team made up of Plat 2 and Plat 1 players. Then the game becomes a toss-up between how hard the P3 player can carry and how fast the G3 player can throw.In the current system, if the P3 player loses, it becomes a "hope matchmaking doesn't do this again". In the proposed system, it'll become "sucks, but let's go next".

Did you not read the part after, where it says:

I do think that generally, there would be a shift in ratings for all players, pushing them up a division (G3->P1, P1->P2). The benchmark of being a decent player would just move up a division as well.

Again, there would be a shift, but it would not affect the actual # placement. Even if Gold 3 players move up to Plat 1, the current Top 250 in Plat 1 would move up to Plat 2. Just because players finally placed in Plat won't mean they'll get a title.The past Diamond division isn't necessary unless we see a massive influx in the population.

Then haphazardly increasing everyone's rating would be just as detrimental.

Providing an example would be the best way to explain your train of thought, instead of just repeating what I said and thinking that it would somehow make sense.

Link to comment
Share on other sites

@nativity.3057 said:

@"Cyninja.2954" said:

What does the skill gap have to do with a player needing to win 2~3 more games after a loss? You don't seem to be reading what I am writing. I'm not misunderstanding anything, I know how big the skill gap is.In this situation, it's more likely a Plat 3 player will be on the same team as a Gold 3 player facing an enemy team made up of Plat 2 and Plat 1 players. Then the game becomes a toss-up between how hard the P3 player can carry and how fast the G3 player can throw.In the current system, if the P3 player loses, it becomes a "hope matchmaking doesn't do this again". In the proposed system, it'll become "sucks, but let's go next".

Did you not read the part after, where it says:

I do think that generally, there would be a shift in ratings for all players, pushing them up a division (G3->P1, P1->P2).
The benchmark of being a decent player would just move up a division as well.

Again, there would be a shift, but it would not affect the actual # placement. Even if Gold 3 players move up to Plat 1, the current Top 250 in Plat 1 would move up to Plat 2. Just because players finally placed in Plat won't mean they'll get a title.The past Diamond division isn't necessary unless we see a massive influx in the population.

You do realize being in plat directly affects rewards, don't you?

@nativity.3057 said:

Then haphazardly increasing everyone's rating would be just as detrimental.

Providing an example would be the best way to explain your train of thought, instead of just repeating what I said and thinking that it would somehow make sense.

Plat players receive 2 extra pips per game. Legendary earn 4 pips per game on top. (ps. I did the majority of my 98 games this season in high gold and p1. I would love to have gotten those ~196 extra pips had I been in P1-P2 consistently. That's literally a free Byzantium chest on top for me.)

Also you do not account for what happens in the new system when the same P3 player plays against those P1 and P2 players. If there is no harsh point penalty in game, yes losing as P3 against weaker opponents costs a ton of points, you would not get such a small number of players at the top. The mere reason why only 1-5% of the pvp population (or which ever tiny number) is at the top is due to the current system in place. Flooding the higher ranks only so players can feel good about them selves is hardly a good reason for a competitive system. Hence why I said: unless you intend to increase the amount of divisions and in that process also adjust reward gain.

Link to comment
Share on other sites

What does the skill gap have to do with a player needing to win 2~3 more games after a loss? You don't seem to be reading what I am writing. I'm not misunderstanding anything, I know how big the skill gap is.In this situation, it's more likely a Plat 3 player will be on the same team as a Gold 3 player facing an enemy team made up of Plat 2 and Plat 1 players. Then the game becomes a toss-up between how hard the P3 player can carry and how fast the G3 player can throw.In the current system, if the P3 player loses, it becomes a "hope matchmaking doesn't do this again". In the proposed system, it'll become "sucks, but let's go next".

Did you not read the part after, where it says:

I do think that generally, there would be a shift in ratings for all players, pushing them up a division (G3->P1, P1->P2).
The benchmark of being a decent player would just move up a division as well.

Again, there would be a shift, but it would not affect the actual # placement. Even if Gold 3 players move up to Plat 1, the current Top 250 in Plat 1 would move up to Plat 2. Just because players finally placed in Plat won't mean they'll get a title.The past Diamond division isn't necessary unless we see a massive influx in the population.

You do realize being in plat directly affects rewards, don't you?

Then haphazardly increasing everyone's rating would be just as detrimental.

Providing an example would be the best way to explain your train of thought, instead of just repeating what I said and thinking that it would somehow make sense.

Plat players receive 2 extra pips per game. Legendary earn 4 pips per game on top. (ps. I did the majority of my 98 games this season in high gold and p1. I would love to have gotten those ~196 extra pips had I been in P1-P2 consistently. That's literally a free Byzantium chest on top for me.)

Also you do not account for what happens in the new system when the same P3 player plays against those P1 and P2 players. If there is no harsh point penalty in game, yes losing as P3 against weaker opponents costs a ton of points, you would not get such a small number of players at the top. The mere reason why only 1-5% of the pvp population (or which ever tiny number) is at the top is due to the current system in place. Flooding the higher ranks only so players can feel good about them selves is hardly a good reason for a competitive system. Hence why I said: unless you intend to increase the amount of divisions and in that process also adjust reward gain.

Yes an extra 2 pips per game, or in this case, an extra chest that gives 28.28 gold (gold + shard + dye) a season would really bring back PvP bots in full force.That's assuming bots would place in Plat (a gamble to have a 50+% winrate) when the bot class (Mirage) is getting nerfed next patch.It's entirely more efficient for these gold farmers to stick to PvE.

I don't want to keep repeating what I've said over and over again. Plat 1 + Plat 2 players won't magically start to rank higher than Plat 3 players. It would require them to consistently have a >50% winrate. Plat 3 players can achieve this easily, while P1/P2 would struggle while facing players of similar caliber. And you won't just be able to spam games, since the ratings gained/lost would decrease as the number of games played increases.

Link to comment
Share on other sites

What does the skill gap have to do with a player needing to win 2~3 more games after a loss? You don't seem to be reading what I am writing. I'm not misunderstanding anything, I know how big the skill gap is.In this situation, it's more likely a Plat 3 player will be on the same team as a Gold 3 player facing an enemy team made up of Plat 2 and Plat 1 players. Then the game becomes a toss-up between how hard the P3 player can carry and how fast the G3 player can throw.In the current system, if the P3 player loses, it becomes a "hope matchmaking doesn't do this again". In the proposed system, it'll become "sucks, but let's go next".

Did you not read the part after, where it says:

I do think that generally, there would be a shift in ratings for all players, pushing them up a division (G3->P1, P1->P2).
The benchmark of being a decent player would just move up a division as well.

Again, there would be a shift, but it would not affect the actual # placement. Even if Gold 3 players move up to Plat 1, the current Top 250 in Plat 1 would move up to Plat 2. Just because players finally placed in Plat won't mean they'll get a title.The past Diamond division isn't necessary unless we see a massive influx in the population.

You do realize being in plat directly affects rewards, don't you?

Then haphazardly increasing everyone's rating would be just as detrimental.

Providing an example would be the best way to explain your train of thought, instead of just repeating what I said and thinking that it would somehow make sense.

Plat players receive 2 extra pips per game. Legendary earn 4 pips per game on top. (ps. I did the majority of my 98 games this season in high gold and p1. I would love to have gotten those ~196 extra pips had I been in P1-P2 consistently. That's literally a free Byzantium chest on top for me.)

Also you do not account for what happens in the new system when the same P3 player plays against those P1 and P2 players. If there is no harsh point penalty in game, yes losing as P3 against weaker opponents costs a ton of points, you would not get such a small number of players at the top. The mere reason why only 1-5% of the pvp population (or which ever tiny number) is at the top is due to the current system in place. Flooding the higher ranks only so players can feel good about them selves is hardly a good reason for a competitive system. Hence why I said: unless you intend to increase the amount of divisions and in that process also adjust reward gain.

Yes an extra 2 pips per game, or in this case, an extra chest that gives 28.28 gold (gold + shard + dye) a season would really bring back PvP bots in full force.That's assuming bots would place in Plat (a gamble to have a 50+% winrate) when the bot class (Mirage) is getting nerfed next patch.It's entirely more efficient for these gold farmers to stick to PvE.

I don't want to keep repeating what I've said over and over again. Plat 1 + Plat 2 players won't magically start to rank higher than Plat 3 players. It would require them to consistently have a >50% winrate. Plat 3 players can achieve this easily, while P1/P2 would struggle while facing players of similar caliber. And you won't just be able to spam games, since the ratings gained/lost would decrease as the number of games played increases.

I never mentioned bots. Bots don't care about rank or placement.

I said your suggestion would affect rewards, while it would not have any positive effect on players besides making the average player rank up higher, until they are again at a sub 50-50 win rate. Even worse, given they lose less points now, they would have to hit a 40-60 win rate or below until they break even on points. That's great while you rank up, and just as sucky when you are then stuck. Or you would rank up indefinitely if you keep the points for loses lower consistently.

You are correct, plat 1+2 players would simply move to plat 3, gold 1-2 players would simply move to g3-p1. Then be stuck there again. If you can't see how a MMR system works (based on distributing points based on rating, and not individual win rate), there is nothing more I can tell you. Your suggestion literally takes out the MMR based factor which measures a player skill and distributes points accordingly to create a normal distribution. I still fail to see the net benefit from a competitive standpoint.

Also your original assumption is incorrect. A 50-50 win rate is NOT a 66% win rate, at least not in a normal gauss distribution where the most players are stacked in the middle. A 66% win rate is only needed once you hit p3 and above (approximately), while at the same time a sub 50-50 win rate is sufficient in lower brackets to climb. This is by simple fact that the majority of players you will face will be in the mid range (I'm assuming s3-g1 is the center of the gauss curve here, thought that might be to high. This is based off of 5 Divisions with adjusting for not being able to drop below bronze), thus requiring higher ranked players to win more against more often weaker players, while weak players need to win less against on average higher ranked players. That's working as intended.

You entire win-rate argument falls apart by that last simple fact. The win rate required is completely different between brackets/divisions. That's a mathematical given if we assume a normal distribution across the entire divisions.

Link to comment
Share on other sites

HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Link to comment
Share on other sites

I mean it’s perfectly fine for someone actually in high rating already, but like on 1 account this season I spammed ques and even when I was in low plat2 or high plat 1- which is actually pretty average, I would lose 150-200% per loss than I would gain making it honestly hard to climb while solo q and match spamming. So I agree with it in this regard, if I’m at a totally baseline ez rating even for top 250 players why should I have a tough time climbing just because I only have a 66% win rate. But for my account I try harded on I was always in top 10 solo Qing on non op builds so I don’t know what to say about that- rating seemed fair for top 10 but for ppl not there it’s totally hard to climb by urself

Link to comment
Share on other sites

@"Bazsi.2734" said:HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Just going to throw it out there that almost every other team-based competitive game already uses a combination of Performance, streaks, and rating to calculate rating gain/loss. Even low pop games like Paladins. The only instance i've ever seen where it was complained about was in early Overwatch where healers would almost always get more/lose less, but that was fixed. It's pretty much widely accepted everywhere.

Gw2 is probably one of the few adhering to the old RNG system of depending solely on the ratings of other people in your lobby. It's made worse by the fact that you can only queue Solo/Duo because that's at minimum 8 other random players who determine if you will waste your time for +3 or -18.

Imagine being plat and getting into a game with low golds and silvers, one proceeds to throw by sitting in spawn for whatever reason. If they're lower rating than their team, then they will lose less than the others; even if their team pours their heart out fighting 4v5. If their team still wins, you; the plat player, will have gained less for carrying them, whereas they gain a solid amount for doing absolutely nothing. I'd say both happen to pretty much everyone every season at least once.

Link to comment
Share on other sites

@Multicolorhipster.9751 said:

@"Bazsi.2734" said:HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Just going to throw it out there that almost every other team-based competitive game already uses a combination of Performance, streaks, and rating to calculate rating gain/loss. Even low pop games like Paladins. The only instance i've ever seen where it was complained about was in early Overwatch where healers would almost always get more/lose less, but that was fixed. It's pretty much widely accepted everywhere.

Gw2 is probably one of the few adhering to the old RNG system of depending solely on the ratings of other people in your lobby. It's made worse by the fact that you can only queue Solo/Duo because that's at minimum 8 other random players who determine if you will waste your time for +3 or -18.

Imagine being plat and getting into a game with low golds and silvers, one proceeds to throw by sitting in spawn for whatever reason. If they're lower rating than their team, then they will lose less than the others; even if their team pours their heart out fighting 4v5. If their team still wins, you; the plat player, will have gained less for carrying them, whereas they gain a solid amount for doing absolutely nothing. I'd say both happen to pretty much everyone every season at least once.

That makes sense, but is not what was suggested in the thread opening, nor discussed by topic creator.

I'm quite sure many players would welcome any system which better distributes rating increase or decrease while factoring in as many useful and necessary factors as possible.

What was suggested in this thread though, was a simple rating inflation via reduction of rating loss. That's certainly not the same as taking into account more factors, which can also lead to increased rating loss or decrease in rating gain, that same plat players would literally get nearly no rating if he won a game against gold opponents in a 5v4 situation favoring him, as mentioned in other games.

Link to comment
Share on other sites

Imo the leavers/disconnecters/afkers/trolls are the biggest problem. Also the problem with people contributing a different amount to the total team effort. I would like to revise the system I suggested above where I also was in favor of using the actual end score to determine the rating loss/gain.

Let's say we compare teams A and B and each team has an average rating. With that we can calculate a loss/gain as if it were just 2 players (like in chess). Now use a multiple of 5 and distribute this among the team according to their individual ratings. For example if team rating gain was 14 you'd distribute 70 and players above the team average would get more and players below would get less. Still ... losing would be a problem for higher ranked players if caused by trolls or bad lower ranked players playing below their performance. (Then again in cases of a win they could win more.) I don't know how this could be completely fixed. Especially at the top end of the ladder.

But let's say we use the actual score (in-match/victory points) ... then getting 450 to 500 could mean you have played very well. (Use this to modify the team average scores. The 14 I used in the example above.) Could lead to less loss then.

At first I said that with this score even a winning team could have to lose rating (if they were expected to win by a higher margin ... score 1700 vs score 9000 would be very bad to win with 500 to 450 only for the higher scored team). Let's say we cap this at 0. Winning gets you at least 0 (no rating loss even if played very badly and just barely won even if you were expected to win higher). Losing gets you max 0. (No gain even if 450 to 500 was pretty good while losing and you were expected to lose by a higher margin.) You could still avoid a rating loss which might be enough for most people.

As for afkers/disconnecters: They might be easy to detect/track.Problem is the people trolling staying online at keyboard. In the spawn area you could use the normal time (without quickness) to leave that area + let''s say 5 seconds. Apply a buff after that that marks the player excluded from rating.(Still will not help against trolls. I don't see a real solution there.)

You could use the factor time when calculating the average team rating (used to calculate the gain/loss). 1200 average and with 1 guy afk it would be 1000 and thet guy was afk for 5 minutes (needs the system to keep track of this) and the match runs for full 15 mins ... then (120010 + 12005)/15. They should be taken into account as a weaker player - when using the average team score (when comparing the 2 teams as if it were 2 individuals playing against each other).

Can't find any better modifications than this atm. Other than making systems to exclude trolls. Would rely on reports + review and later a change fo the gain/loss after it got reviewed. I can see this would take tons of human work on the side of ArenaNet. (As long as there is no replay system where players could be voted as judges and review matches later. I thinkg games like LoL have something like that?) You certainly can't just say "if x people report someone he automatically gets excluded from calculation". Would lead to abuse and people reporting on purpose.

Link to comment
Share on other sites

@Cyninja.2954 said:That makes sense, but is not what was suggested in the thread opening, nor discussed by topic creator.

I'm quite sure many players would welcome any system which better distributes rating increase or decrease while factoring in as many useful and necessary factors as possible.

What was suggested in this thread though, was a simple rating inflation via reduction of rating loss. That's certainly not the same as taking into account more factors, which can also lead to increased rating loss or decrease in rating gain, that same plat players would literally get nearly no rating if he won a game against gold opponents in a 5v4 situation favoring him, as mentioned in other games.

Gotcha, gotcha.

The OP seems a bit open-ended. I wouldn't vote for a straight inflation of rank if that's what was meant. Can't just straight-up reduce the amount from losing as a whole, that's is no bueno.

I mean, I voted in favor, but that was just for a change in general. I'd prefer something more modern like I say.If we're voting to keep the RNG, and just cut back the loss numbers, then i'd retract that vote in a heartbeat.

Link to comment
Share on other sites

@"Cyninja.2954" said:

I never mentioned bots. Bots don't care about rank or placement.

I did mention bots, because that is what increasing base rewards to where it's more efficient to farm PvP would do.

I said your suggestion would affect rewards, while it would not have any positive effect on players besides making the average player rank up higher, until they are again at a sub 50-50 win rate. Even worse, given they lose less points now, they would have to hit a 40-60 win rate or below until they break even on points. That's great while you rank up, and just as sucky when you are then stuck. Or you would rank up indefinitely if you keep the points for loses lower consistently.

I wouldn't say the small increase in pips earned by reaching Plat would be truly felt by players. The difference is small, making it seem inconsequential and overlooked. But it is still a reward. And why would they break even on points in 40-60 winrate when they would break even on points in 50-50. The whole idea is to keep wins/loss rating equal to each other, so at 50-50 you're stuck. The small variance in earned rating or lost rating depending on matchmaking would average out to be equivalent given enough games.

You are correct, plat 1+2 players would simply move to plat 3, gold 1-2 players would simply move to g3-p1. Then be stuck there again. If you can't see how a MMR system works (based on distributing points based on rating, and not individual win rate), there is nothing more I can tell you. Your suggestion literally takes out the MMR based factor which measures a player skill and distributes points accordingly to create a normal distribution. I still fail to see the net benefit from a competitive standpoint.

Also your original assumption is incorrect. A 50-50 win rate is NOT a 66% win rate, at least not in a normal gauss distribution where the most players are stacked in the middle. A 66% win rate is only needed once you hit p3 and above (approximately), while at the same time a sub 50-50 win rate is sufficient in lower brackets to climb. This is by simple fact that the majority of players you will face will be in the mid range (I'm assuming s3-g1 is the center of the gauss curve here, thought that might be to high. This is based off of 5 Divisions with adjusting for not being able to drop below bronze), thus requiring higher ranked players to win more against more often weaker players, while weak players need to win less against on average higher ranked players. That's working as intended.

You entire win-rate argument falls apart by that last simple fact. The win rate required is completely different between brackets/divisions. That's a mathematical given if we assume a normal distribution across the entire divisions.

The MMR system reflects a player's skill level. A player's skill level is also shown (as much as it can in a 5v5 game) in their win-loss ratio. The current system bottlenecks ratings earned at the higher percentile for the sake of "keeping a normal distribution".My original point is not an assumption; it reflects reality. Currently in the higher divisions, 2 wins worth of rating = 1 loss worth of rating.And if currently in the lower brackets a 50% winrate is only needed to climb, I don't see how the proposed change would make it easier to climb, given a 50% winrate would mean you are stuck.

Yes, the proposed change won't end up with a normal distribution across all divisions. In fact, what I predict is we will see normal distribution per division, and the population of players in each division will vary.To make it a bit clearer using examples:Gold/Silver/Bronze 2 would represent the average Gold/Silver/Bronze player. The 1/3 tiers would be less populated, but being 50% win rate in these divisions would mean you're not quite at the next level, but also not quite at the level below.And saying everyone would be inflated a division up was incorrect on my part. I do not know where the majority of the population would end up. I would need Anet's PvP data to make that assumption.

Regarding "win-streaks" and "loss-streaks", players with above 50% win rate will climb until they reach an "inflated" rating that does not match their skill level, and will start to lose more games than win until they achieve 50% win rate at a certain division. The opposite is true for those unlucky enough to go on long losing streaks.

@"Bazsi.2734" said:HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Yet somehow people have played 500 games, are at 50% win-rate, and are still Gold (and according to another poster, 50% winrate in Gold is still a net positive in rating). I don't see how spamming games would magically let a Gold player reach Legendary.

@Multicolorhipster.9751 said:

@"Bazsi.2734" said:HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Just going to throw it out there that almost every other team-based competitive game already uses a combination of Performance, streaks, and rating to calculate rating gain/loss. Even low pop games like Paladins. The only instance i've ever seen where it was complained about was in early Overwatch where healers would almost always get more/lose less, but that was fixed. It's pretty much widely accepted everywhere.

Gw2 is probably one of the few adhering to the old RNG system of depending solely on the ratings of other people in your lobby. It's made worse by the fact that you can only queue Solo/Duo because that's at minimum 8 other random players who determine if you will waste your time for +3 or -18.

Imagine being plat and getting into a game with low golds and silvers, one proceeds to throw by sitting in spawn for whatever reason. If they're lower rating than their team, then they will lose less than the others; even if their team pours their heart out fighting 4v5. If their team still wins, you; the plat player, will have gained less for carrying them, whereas they gain a solid amount for doing absolutely nothing. I'd say both happen to pretty much everyone every season at least once.

I mean, the proposed change would be dealing with this issue?Equalizing rating won/lost would soften the effect of having a mismatch or a troll. So for the plat player, instead of losing ~20 like in the current system, you would lose 10~13.

Link to comment
Share on other sites

@nativity.3057 said:The MMR system reflects a player's skill level. A player's skill level is also shown (as much as it can in a 5v5 game) in their win-loss ratio. The current system bottlenecks ratings earned at the higher percentile for the sake of "keeping a normal distribution".

This is untrue. Win rate is never a reflection of a players skill level, especially not compared to a MMR system. This is quite simply disproven by mere fact that the same player can have a different win rate in lower brackets versus higher. As such, normalizing win rate reweards by forcing an equal 50-50 win rate makes no sense for stagnation in rating.

Normal distribution will occur in every competitive system by simple fact that not every one can be the best. Your proposed change is not a competitive system.

@nativity.3057 said:My original point is not an assumption; it reflects reality. Currently in the higher divisions, 2 wins worth of rating = 1 loss worth of rating.

Yes, and that is okay because on average, those high ranked players play a lot more games against weaker players than against even skilled players. Did you miss that part? I even explained why this is the case.

Now other have brought up points as to how to make this loss be better reflected outside of a simple win-lose, like having points count, afk, etc. Which all makes sense. But simply going off of a win-loss ratio to balance is kitten. The math does not support this approach.

@nativity.3057 said:And if currently in the lower brackets a 50% winrate is only needed to climb, I don't see how the proposed change would make it easier to climb, given a 50% winrate would mean you are stuck.

Lower brackets need LESS than a 50-50 win rate to climb, up until the point where the player reaches the maximum player density per rating, which I assumed to be at s3-g1. Meaning at that point a 50-50 win rate will keep a player stable in rating. After that the ratio switches to require more wins versus losses by simple fact that the larger amount of players one faces will be weaker, or one is rewarded with more points/less loss against stronger.

@nativity.3057 said:Yes, the proposed change won't end up with a normal distribution across all divisions. In fact, what I predict is we will see normal distribution per division, and the population of players in each division will vary.

That is not a competitive system and I still fail to see the benefit here, in part due to something you have not yet addressed but which was hinted at by me and others. If you force a continued reduction in point loss, players would continually rise until:A.) you either introduce a realistic loss of points some time later, which then hits the player like a truckB.) each player with enough games reaches a far higher rating then their actual skill level, thus leading to such a horrendous win-loss ratio to compensate for the lacking loss penalty that it would be far worse. That would literally affect every one too

This would also not lead to normal distribution within divisions, for that you would have to introduce fixed points for each loss, which in turn would lead to a huge amount of players getting stacked at the end of each division, for all who can't match the next divisions fixed loss penalty, and very few players at the bottom of each division, since all who can match the loss penalty will eventually climb to the top of that division.

@nativity.3057 said:Yet somehow people have played 500 games, are at 50% win-rate, and are still Gold (and according to another poster, 50% winrate in Gold is still a net positive in rating). I don't see how spamming games would magically let a Gold player reach Legendary.

You do know that win-rate is across ALL those 500 games right? You are telling me, that a player who has played 500 games, has not improved AT ALL? You do realize that the most reasonable explanation here is: their win rate was below 50-50 when they started, and has been above 50-50 towards the end with the cumulative rating of all wins-losses being 50-50, landing them in gold.

Are you sure you want to work off of win rate, you seem unclear of how this works?

@nativity.3057 said:I mean, the proposed change would be dealing with this issue?Equalizing rating won/lost would soften the effect of having a mismatch or a troll. So for the plat player, instead of losing ~20 like in the current system, you would lose 10~13.

Trolls, bots, afk players , etc will happen no matter which system is in place. They do not make up the majority of reasons for wins/losses. Unbalancing a competitive system to compensate for factors OUTSIDE the scope of player skill will simply unbalance the system. Also there would be far better ways to deal with these situations which would not unbalance the system, like is the case if a player disconnects.

Also please don't assume that plat players face the same amount of trolls, bots and afk players as bronze, silver or gold ranked players. They do not, and it is noticeable if you move through the divisions.

Link to comment
Share on other sites

@nativity.3057 said:I mean, the proposed change would be dealing with this issue?Equalizing rating won/lost would soften the effect of having a mismatch or a troll. So for the plat player, instead of losing ~20 like in the current system, you would lose 10~13.

Rather than straight up decreasing the amount lost, the better solution would probably just be to give people more potential to earn and lose more rating. It can't just all be straight gains because that would make rating meaningless... well, more meaningless I mean.

The formula could be as simple as adding up 3 things and using that to determine rating gain:1) Would be rating in comparison to other players in your lobby, AKA how it is now. It's important so that when the top 2 players in the game inevitably Duo with eachother and start noob-stomping, they aren't just farming an insane amount of rating. It's a bit RNG, but that can be offset slightly by the other 2.

2) Would be streaks. How many wins/losses have you had in quick succession? Lots of wins = rating boosts, lots of losses = rating penalties. If you're able to chain together a bunch of wins/losses that's a pretty clear sign that you're either doing too well/too poorly in your division. If you keep someone in a division who is constantly winning/losing games back to back, you only force a needless grind on them until they reach a division that's actually fair to them like they probably would have reached regardless. I know some people just get lucky sometimes, and you get massive streaks even if you don't deserve them, but this is offset by the other 2.

3) Would be Personal performance in relation to other people in your lobby. Hard to calculate for a game like Gw2, but it wouldn't be impossible. Would be even better if we had a scoreboard that actually showed helpful and relevant information to conquest. If we had something representative for every role(Damage, Support, Side-noder, Roamer) then this could be used.If you sat in spawn and contributed nothing, you deserve to lose a ton of rating, or gain nothing if your team manages to carry you.If you put in 50-60% of everything on the scoreboard for your team, you deserve to lose less/gain more.The matchmaking in this game is pretty terrible, and sometimes people get easy games where they easily carry. That would be offset by the other 2.

Same thing used in most other ranked team-based arena games. Some games still only use the rating's of other players in relation to you like gw2 does now, but it's not anywhere near as representative of player skill since it's entirely RNG.

Just a straight up reduction to how much you lose from losing in general doesn't really help that at all.

Link to comment
Share on other sites

@nativity.3057 said:

@"Bazsi.2734" said:HELL NO. Ratings would inflate to insane levels, matchmaking would break. Ladder would become a "who above 50% winrate can grind the most" toplist, and not resemble actual skill levels. Just think about it, 120 games played with 90% winrate vs 400 games played with 65% winrate... with flat rating changes for wins/losses the second one comes out on top.The penality for winning matches way below your personal rating is there for a reason. If you want a linear climb, I suggest you google 'the worlds longest staircases' before you book your next vacation.

Yet somehow people have played 500 games, are at 50% win-rate, and are still Gold (and according to another poster, 50% winrate in Gold is still a net positive in rating). I don't see how spamming games would magically let a Gold player reach Legendary.

Thats completely irrelevant. This change obviously wouldn't affect mid-tier peope with exactly 50% winrate, and thats why I WASN'T TALKING ABOUT THEM.Second attempt at conveying the same point:120 games played, 90% winrate -> 108-12=96 more wins than losses. This puts you at 1400+96 X rating in a linear system.400 games played, 65% winrate -> 240-120=120 more wins than losses. This puts you at 1400+120 X rating. Regardless of what X is, if its consistently the same number, the second player gets to be higher on the leaderboard. Any system that lets you get ahead of better players through grinding is a bad one.

Also linear climbing would create such great gaps in rating that matchmaking would force people to wait 15-30 minutes before even cosidering pairing them. You cannot fix that without ruining matchmaking for the rest.

Link to comment
Share on other sites

@Jayden Reese.9542 said:I think there should be some changes. The one I can think of is I won this match like 500-480 when we were losing all match because we wiped them last minute at mid to come back. I gained 17 and their team had 2 duos we had none. Next match we got wrecked 500-129 no duos and guess how many points I lost yeah 17. Or ill go 5-5 over 10 matches and gain 13 13 13 13 13 and lose 13 14 13 14 14. So even going break even I lose rank overall. It feels you can't make any progress. I think you should lose less if vs a duo or vs 2 duos if you have none or if it's a closer match

I don't get why they allowed duos in the first place, it makes no sense

Link to comment
Share on other sites

Players with high w/l ratios do need these big discrepancies between rating gained and lost. Sind for example is on about 90W/20L on his accs, now figure out what would happen to his rating if he got almost as many points for a win as he loses for a loss. People on high W/L would then climb indefinitely and it would become a race about who grinds the most games.

Link to comment
Share on other sites

No surprise that top players don't want changes. The more you're on top, the less your gain are. Consequence: top players don't reach their real rating and they can have stupid W/L ratio cause of it.Like @bluri.2653 says, he'll get 3k rating or even more, so MM should force him to play with bronze team against a low plat team. Not fun for him and result 50% win ratio like pleb.

@bluri.2653 : sorry to name you, just use you as an exemple.

Link to comment
Share on other sites

the reasons I voted to change it aren't because I want to grind to legendary. I don't think most people want that. what i'm worried about is that:1 - reaching legendary or even high plat (leaderboards too) is nearly impossible for most ppl so they resort to cheating2 - loss is too high which encourages playing minimal games until end of season3 - streaks ruin gameplay and can either demoralize into log off or put the player in a tier they don't belong in which messes up the game for those other pplsI have no idea how things could be changed to address this, but imo something should be done. maybe if nothing else anet can put a dolyak in the middle of pvp lobby and ppl can pay 1 gold to slap the yak. good stress relief, and hey, don't judge; some ppl chop wood, some hit the bag, some slap yaks.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...