Guild Wars 2

Instead of gaining and losing mmr based solely on wins and losses, we should gain and lose mmr based on the difference between our final scores and our win/loss.

If you lose by a little it shows the match was close. If it is a blowout then obviously someone’s mmr needs to change. It also incentivizes people to fight to the end and not just give up.

If there is a reason to not do this I’d love to know it.

Generally I agree. I don’t see any major issues here. Even if something like this may make it a bit harder to climb, it should make it harder to fall as well.

I’d say a flat +/- 5 Rating for win/lose. (Both could get + 5 on a tie). And then +/- an additional 1 Rating to each team for each 20 points they are off of the other team.

So if it ends 500:100, the winning team gets 25 points, losing team loses 25 points. If it ends 500:400, the winning team gets 10 points, losing team loses 10 points. And so on.

Maybe it shouldn’t be +/- 1 Rating at every 20 points, because that would make it possible to both gain and lose a lot in a single match (which could be a big issue for losing team if someone DC’s or rage quits). Maybe the losing team should lose 1 point per 30 and winning gains 1 per 20, or something like that (that may cause inflation, not sure).

this goes againts the logic of the sistem
blowout matches are caused to system fault to create even matches and rewarding extra points to a “preset” winner and extra punishing the “preset” looser will destroy the entire mmr rank
the logic system is that in S1 extra rewarding “preset” looser to win or to get close, and punishing preset winer for leting the rival to get closer, but S1 proves that system fails to determine it, and those fails can be exploited(specialy if it is any sort of “team” involved(duo,trio,quad or full team)

It shouldn’t give you extra MMR for winning by a lot, but it should definitely not lower your MMR by as much if you lose by only a few points opposed to losing by several hundred.

would really skew the ratings with DC’s and afkers

NeXeD.3042:

would really skew the ratings with DC’s and afkers

On paper, nothing would change if we still have DCers and AFKers. However, if the proposed system is implemented it would perhaps encourage players to try harder.

So for example if the score is 500-480:

Winning team +17
Losing team -5 (instead of -17)

OriOri.8724:

It shouldn’t give you extra MMR for winning by a lot, but it should definitely not lower your MMR by as much if you lose by only a few points opposed to losing by several hundred.

if mm works fine yes, but mm has proved that fails to create even matches extra punishing the “preset looser” will hurt more the system.

asuming a perfect mm in perfect conditions(all matches 50/50 or near win loss odds) it will be ok that looser gets a “reward” for geting 450+ points a normal loss if 350+ and an extra punish if 350- but the system is not perfect and the conditions are less perfect

Evan Lesh.3295:

Dariya.9380:

NeXeD.3042:

would really skew the ratings with DC’s and afkers

On paper, nothing would change if we still have DCers and AFKers. However, if the proposed system is implemented it would perhaps encourage players to try harder.

So for example if the score is 500-480:

Winning team +17
Losing team -5 (instead of -17)

Glicko would actually give +5 for the winning team (simplified example). So the winning team would also be encouraged to win by as much as possible.

To clarify on what he means in case people are confused. Basically glicko balances the rating gains and losses evenly. If I lose 15, another person gain 13-15 depending on their rating.

If I were to lose 3 instead, the enemy team would gain 3 rating for a close match.

The winning team is already encouraged to win by as much as possible. That’s one of the big things that turns people off PvP.

Spawn camping, calling out “gg ez” in map chat…. and people wonder why PvP has the reputation it does. It’s because of the “sportsmanship.”

Ithilwen.1529:

The winning team is already encouraged to win by as much as possible. That’s one of the big things that turns people off PvP.

Spawn camping, calling out “gg ez” in map chat…. and people wonder why PvP has the reputation it does. It’s because of the “sportsmanship.”

Not so much spawn camping but the contextual map chats is something Every competitive game endures. Games like LoL is a whole lot more enduring than gw2..

Saiyan.1704:

Ithilwen.1529:

The winning team is already encouraged to win by as much as possible. That’s one of the big things that turns people off PvP.

Spawn camping, calling out “gg ez” in map chat…. and people wonder why PvP has the reputation it does. It’s because of the “sportsmanship.”

Not so much spawn camping but the contextual map chats is something Every competitive game endures. Games like LoL is a whole lot more enduring than gw2..

Gw2 is super tame compared to other pvp mmos I have played.

Ultima online has been around since 97 and the competitiveness or its gw2 synonym “toxicity” is still very much prevalent to this day.

Here is an example of what I’m proposing.

Say under the current system a match results in an average MMR win-lose of +15/-15. The people on the team with higher MMR lose more per volatility or gain less per volatility—but the balance is overall roughly even.

Now lets look at what I proposed which is an MMR adjustment bonus for close and not so close matches. The bonus works by rewarding people who win by a lot or decreasing the loss of the people who lost by a small amount.

+15/-15 then is adjusted as follows:

+5 MMR for win >450 points
+3 MMR for win >350 points
+1 MMR for win >250 points
0 MMR adjust for win/loss 100-250 points
-1 MMR for win < 100 points
-3 MMR for win < 50 points
-5 MMR for win < 25 points

Reversed order for a loss perspective.

Matchup scenario 1: 504-434 (70 points: MMR adjust +/- 1)

Winners +14, Losers -14

Match up scenario 2: 500-325 (175 points: MMR adjust +/- 0)

Winners 15, Losers -15

Matchup scenario 3: 500-125 (375 points: MMR adjust +/- 3)

Winners 18, Losers -18

Matchup scenario 4: 500-225 (275 points: MMR adjust +/- 1)

Winners 16, Losers -16

Matchup scenario 5: 500-0 (500 points: MMR adjust +/- 5)

Winners 20, Losers -20

Actually its the matches that have a win and loss between 450 and 500 that I would like to see adjusted the most or even 400-500. These matches are good matches each team is really have a great time but a loss is a loss and it kills the drive on the part of the team that loses to strive for comebacks. For example if a game ends in 490 to 501, or in some cases I have had like 499-500, this should be a huge factor and not cause a reduction at all or really ver minimally to the losing team. This means the matchmaker did a great job and definitely both teams should be rewarded in some way. In fact you shouldnt just lower the amount of rating that the losing team gets but incentivise these matches greatly as in a bonus chest or something. Not a pips based thing because you always want people striving for that bonus chest in ranked or unranked. The lack of rating loss or at least a minimal one would make it more rewarding for teams to strive for this in ranked play.

saerni.2584:

Here is an example of what I’m proposing.

Say under the current system a match results in an average MMR win-lose of +15/-15. The people on the team with higher MMR lose more per volatility or gain less per volatility—but the balance is overall roughly even.

Now lets look at what I proposed which is an MMR adjustment bonus for close and not so close matches. The bonus works by rewarding people who win by a lot or decreasing the loss of the people who lost by a small amount.

+15/-15 then is adjusted as follows:

+5 MMR for win >450 points
+3 MMR for win >350 points
+1 MMR for win >250 points
0 MMR adjust for win/loss 100-250 points
-1 MMR for win < 100 points
-3 MMR for win < 50 points
-5 MMR for win < 25 points

Reversed order for a loss perspective.

Matchup scenario 1: 504-434 (70 points: MMR adjust +/- 1)

Winners +14, Losers -14

Match up scenario 2: 500-325 (175 points: MMR adjust +/- 0)

Winners 15, Losers -15

Matchup scenario 3: 500-125 (375 points: MMR adjust +/- 3)

Winners 18, Losers -18

Matchup scenario 4: 500-225 (275 points: MMR adjust +/- 1)

Winners 16, Losers -16

Matchup scenario 5: 500-0 (500 points: MMR adjust +/- 5)

Winners 20, Losers -20

I repeat for that is not clear for someone:
mmr is not a reward, its a skill measurement, losing a close one means to the system you have lose to a equal skilled team and you should loss mmr, losing a blowout match instead means you lose a game that are preset to be losed and you deserve a minor mmr loss. this thread is againts the system logic.
fighting to the end a losed match have to be rewarded, but not in mmr basis, giving the extra pips is a good move, but also they have to give some more reward route points to reward it not only during seasons

Actually, a close game reduces the disparity under my suggestion. The 500-485 game gives -5 to winner and +5 to loser. Per the above mmr adjustment it becomes +10/-10 instead of +15 and -15.

Winner still gains and loser still loses, just less so.

Not sure I agree losing 500 to 0 means you shouldn’t lose ranking. If we want accurate matches then 500-0 requires some substantial adjustments.

I agree, had already argued that delta match score should be considered in rating change. Would encourage people to fight till the end. But winners should also get a bonus for exceeding so that they don’t just go easy on the other team and let them stay within a hundread. A 500-499 game should not have the exact same mmr adjustment as a 500-10 game

Again, is not about if game is close or not, s1 what have the correct aproach, but unfortunately got xploited by teams via smmurfing acounts, sets a point objective for loser team and a limit to the points that enemy can achieve for winner team, and you are “rewarded” to get that or surpass it, you are “rewarded” for doing that the system expects or better that their calculations, not for a static point threshold

megilandil.7506:

Again, is not about if game is close or not, s1 what have the correct aproach, but unfortunately got xploited by teams via smmurfing acounts, sets a point objective for loser team and a limit to the points that enemy can achieve for winner team, and you are “rewarded” to get that or surpass it, you are “rewarded” for doing that the system expects or better that their calculations, not for a static point threshold

Just because something was not implemented right does not inherrently mean the idea was wrong. What is the logical flaw in arguing that a 499-500 match should not see the same change to MMR of the participants as a 10-500 match?

shion.2084:

megilandil.7506:

Again, is not about if game is close or not, s1 what have the correct aproach, but unfortunately got xploited by teams via smmurfing acounts, sets a point objective for loser team and a limit to the points that enemy can achieve for winner team, and you are “rewarded” to get that or surpass it, you are “rewarded” for doing that the system expects or better that their calculations, not for a static point threshold

Just because something was not implemented right does not inherrently mean the idea was wrong. What is the logical flaw in arguing that a 499-500 match should not see the same change to MMR of the participants as a 10-500 match?

the logic is that mmr is a implementation of glicko that is a variation of ELO and the mmr variations should be based on “strength” of rival(you win against a high mmr rival great mmr “reward” , you win against a equal skilled rival normal mmr reward, you win a less skilled rival low mmr reward, you lose agains a less skilled rival big mmr penalization …) no in achieving a static objective even if you are expected to do better, other thing is in case your team is the weakest rival implementing a consolation prize for surpasing their expected match outcome losing for more than X,“you are paired against a superior team you are suposed to lose for more than x points but you performed well and lose by minus than x points(you have wined the match in some abstract way), you get a personal victory against the odds and your mmr variation should be compensated in some way”

Your logic appears to be that you want mmr to be winner take all, where match winners and losers are adjusted based on expected outcome and whether they happened to win or lose. Correct me if I’m wrong but from what I could understand your entire premise rests solely on liking winner take all as a system.

My suggestion is that instead of winner take all we reward competitive play by adjusting rating gain by the finer grain of distinction between a 500-1 victory and a 500-499 victory. Players would still gain and lose relative to the expected outcome (mmr disparity between teams) but not solely based on that. Obviously, I disagree that winner take all produces accurate or fun results.

megilandil.7506:

shion.2084:

megilandil.7506:

Again, is not about if game is close or not, s1 what have the correct aproach, but unfortunately got xploited by teams via smmurfing acounts, sets a point objective for loser team and a limit to the points that enemy can achieve for winner team, and you are “rewarded” to get that or surpass it, you are “rewarded” for doing that the system expects or better that their calculations, not for a static point threshold

Just because something was not implemented right does not inherrently mean the idea was wrong. What is the logical flaw in arguing that a 499-500 match should not see the same change to MMR of the participants as a 10-500 match?

the logic is that mmr is a implementation of glicko that is a variation of ELO and the mmr variations should be based on “strength” of rival(you win against a high mmr rival great mmr “reward” , you win against a equal skilled rival normal mmr reward, you win a less skilled rival low mmr reward, you lose agains a less skilled rival big mmr penalization …) no in achieving a static objective even if you are expected to do better, other thing is in case your team is the weakest rival implementing a consolation prize for surpasing their expected match outcome losing for more than X,“you are paired against a superior team you are suposed to lose for more than x points but you performed well and lose by minus than x points(you have wined the match in some abstract way), you get a personal victory against the odds and your mmr variation should be compensated in some way”

If you are arguing that all or nothing can not be accomodated by any variation of Glicko, then it is not that my point is wrong, it is that the Relative Ranking we are using is based on something made for chess where close wins didn’t really factor in and so the system can’t handle it.

Even applying Glicko to the outcome of an individual whom participates in a team that changes with every match is fundamentally flawed as it was designed for individuals, and abstractions handle teams that stay more or less the same. When you shuffle the team every game it obviously looses validity.

So yes if your saying it couldn’t be MMR that is adjusted because MMR is literally a property of glicko, then I’m saying I was using MMR in the generic sense of a reltaive rating of an individuals capability and this relative rating should not then be an MMR (if MMR is so inflexible).

saerni.2584:

Your logic appears to be that you want mmr to be winner take all, where match winners and losers are adjusted based on expected outcome and whether they happened to win or lose. Correct me if I’m wrong but from what I could understand your entire premise rests solely on liking winner take all as a system.

My suggestion is that instead of winner take all we reward competitive play by adjusting rating gain by the finer grain of distinction between a 500-1 victory and a 500-499 victory. Players would still gain and lose relative to the expected outcome (mmr disparity between teams) but not solely based on that. Obviously, I disagree that winner take all produces accurate or fun results.

its not my logic, its sistem(glicko, ELO) logic

mmr is not a reward, is a skill calculation based on who wins or loses and more important against who

the system can be adjusted more to “fine grain” but your proposal not make it, adjusting the system to fine grain is S1 system (calculating a expected outcome from match and rewarding or penalizing in function of that). your system could “reward” the expected winer for losing but no the expected loser for a wide margin for shortening it cuz you are basing it in static results not in expected ones(an extremely unbalanced match can have a expected result of a blowout of 500-50 and the correct fine grain adjustement is “rewarding” the loser for geting more than 50 points and “penalizing” the winer for allowing it, the loser have win in some way achieving a better result than expected and winer had lose in some way getting a worse result than expected )

How about we fix matchmakimg first… have the system make even-sided matches then we can look upon the Elo, glicko calculator… if it even needs improving.

If it was profession based queueing, mmr calculator would have been more accurate. Just saying.

you want season 1 style but w/ mmr points not pips, yes +11111111111111111111111

Given that I can’t understand you very well I’ll try to respond as best I can under the circumstances.

You seem to think that I think of MMR as a reward. I do not make that assumption. In an ideal system it is assumed both sides have an equal chance to win. A blowout match indicates that something went wrong with that assumption. A close match indicates that the mmr was fairly indicative of an even fight. In the middle are marginal cases (100-250 point differences) where it is hard to say composition and not player skill was the reason for the victory or the loss.

Players want to be at their true mmr. That is where the motivation comes from. There is no “reward” except for closer matches and fewer blowouts. Players need the system to be more accurate to increase the close matches. It makes the game feel less arbitrary in the long run.

Last, I disagree that we should all just accept the current system. Just saying it’s winner take all, so we should accept it as designed, is circular logic. The system is this way so the system should be that way. That isn’t an argument for keeping it that way.

Evan Lesh.3295:

Dariya.9380:

NeXeD.3042:

would really skew the ratings with DC’s and afkers

On paper, nothing would change if we still have DCers and AFKers. However, if the proposed system is implemented it would perhaps encourage players to try harder.

So for example if the score is 500-480:

Winning team +17
Losing team -5 (instead of -17)

Glicko would actually give +5 for the winning team (simplified example). So the winning team would also be encouraged to win by as much as possible.

So if you need Glicko that much. Lets say we have something like /- 5 mmr points win/loose.
Than easily making it rising from +/- 1 (up to 100 game pts on loosers) moving to +\- 10 ( on 350 game pts on loosers) than to +\-1 pt (on 450 Game pts)
So win 500-48 would give 6 pts, as well as at close win, and nobody wants to be in middle, where is some movement needed if you want equal matches.
You might ask what is the reward on such a thing? Exactly the reward tracks or wins on rank rewards which would be pushing you by making that difference supporting the ranked druring season and will to win by big difference (making bonus points towards that chest for example), not let the people exploiting by letting them gain a few points(100) so that we get more mmr points(as such an exploit could be done by the higher ranking people). Because there is a risk if they let them win points till 350.

Ps.: Thx for ignoring my question on forum. But all this answers it you don’t know how to, so perhaps this suggestion would make some sense.