Guild Wars 2

Hi there, this has been on my mind for quite a while now seeing as I am an other unhappy camper about mmr being based off wins/losses because as you all know, each player represents 20% of their team more or less comms/leadership/smarts that a player can exude during the game.

But lets face it, most games are probably pugs and you have a couple groups of organised people who play with comms. Nice to see that they decided to add a 1-2 queue to solve the discrepencies there but lets cut to the chase.

I am mainly a soloQ guy and so are many of you. And the matchmaking/mmr reward/rating as it is doesn’t feel satisfactory towards individual skill level. If based solely on organized groups with comms it might be suitable but it doesn’t adress the needs of all player types.

That said, we need to identify the characteristics of play that are worth something in PvP, that is to say, what makes individual players good be counted as good and bad be counted as bad. There is already a ton of metrics in the game and probably only a couple more should be created to harness a system that would be adequatly representative.

I can’t say I have all the answers right now, but a couple ideas and I encourage you to discuss on what attitudes/actions/happenings in a game is noteworthy of qualities of the “good player”.

To me, here are a couple qualities of a “good player”

• Kills
• Rezzes/rallies of nearby teammates.
• Stopping stomps
• Successful disengages from battle and healing back
• less Time away from the fray
• Time alive in combat vs 1, 2, etc
• Time defending capture point
• Time assaulting capture point
• Kills/Death Ratios
• Time to kill once engaged (for successful +1 scenarios)
• High performance vs class type average

Those are the ones I can think for now.

There are also some bad player hebaviour

• Time to death once engaged (like suiciding on a point)
• Time spent @ spawn
• Idle time away from capture points (debatable)
• Low performance vs class type average

I’m sure there are more.

I think a system based on something like this could account for individual player skill. if you look into all that info…you could correlate a player’s participation to a kill to a couple factors, total damage dealt, and if damage dealt vs death of foe is closely tied in time (for +1’s and opportunity chase and finish).

That’s all I have for now but I wanted to get a discussion going and perhaps agree on weight factors/ratios and etc.

Food for Thought

Addition

Roles attached to amulets types: Offense (power/Condi), Support

Play styles

Bunker: Successfully defending point (kill or getting other player to disengage).

Support: Heals on other team members, boons on other team members, Rezzes, debuffs on enemy team, cc on enemy team leading to down/kill within relative short time.

Offense:

• Power: Damage dealt, dps (for bursts/aoe pressure)

• Condi: # of conditions applied, total Condition damage

Measuring individual skill would theoretically be the best but accurately measuring contribution is nearly impossible; for instance the player who decaps far then holds off 3 people in the middle of nowhere while their team 3caps ends up with a very low score but contributed to 40+ point difference for their team.

Also player will invariably find the most efficient way to farm points and sacrifice trying to win in favor of gaining points be that going full tank and only reviving or camping a single point or what have you.

That’s not to say someone shouldn’t try to add individual skill to MMR but it’s not something that can be considered lightly and from a development perspective something that difficult, while guaranteeing universal acclaim would require time and resources that don’t exist.

This list is mostly DPS role-oriented. A bunker, while extremely useful, would have a low kill-death ratio unless the enemy team was constantly streaming in 1 by 1 and getting killed by the bunker. In fact it’s possible they might have more deaths than kills due to being zerged a few times. To compensate for that I think time spent in/near an owned node without the presence of other team members (shouldn’t reward double/triple-bunkering) should be counted as positive behavior. Generally assaulting enemy-held nodes shouldn’t be rewarded unless successful, there’s plenty of idiots who get farmed mid who should be punished for their failure, not rewarded.

More stats like healing to allies/condis cleaned would be needed for support classes. Stats like kills, deaths, rezzes and such should be weighted according to the amulet chosen, since that’s the single strongest universal indicator of what role the character is primarily playing.

Oldyoung.6109:

Measuring individual skill would theoretically be the best but accurately measuring contribution is nearly impossible; for instance the player who decaps far then holds off 3 people in the middle of nowhere while their team 3caps ends up with a very low score but contributed to 40+ point difference for their team.

Also player will invariably find the most efficient way to farm points and sacrifice trying to win in favor of gaining points be that going full tank and only reviving or camping a single point or what have you.

That’s not to say someone shouldn’t try to add individual skill to MMR but it’s not something that can be considered lightly and from a development perspective something that difficult, while guaranteeing universal acclaim would require time and resources that don’t exist.

People trying to farm points shouldn’t be a big concern so long as win/loss ratio continues to be a strong identifying factor in MMR. If they work their way up, they’ll fall back down once they get beyond their skill level.

And I’m not really sure about the resources part. Anet remains a profitable company, feel free to provide evidence that they can’t do it if you have it. We’re doing a lot of the heavy lifting for them in this thread and on this forum in general.

Huskyboy.1053:

Oldyoung.6109:

Measuring individual skill would theoretically be the best but accurately measuring contribution is nearly impossible; for instance the player who decaps far then holds off 3 people in the middle of nowhere while their team 3caps ends up with a very low score but contributed to 40+ point difference for their team.

Also player will invariably find the most efficient way to farm points and sacrifice trying to win in favor of gaining points be that going full tank and only reviving or camping a single point or what have you.

That’s not to say someone shouldn’t try to add individual skill to MMR but it’s not something that can be considered lightly and from a development perspective something that difficult, while guaranteeing universal acclaim would require time and resources that don’t exist.

People trying to farm points shouldn’t be a big concern so long as win/loss ratio continues to be a strong identifying factor in MMR. If they work their way up, they’ll fall back down once they get beyond their skill level.

And I’m not really sure about the resources part. Anet remains a profitable company, feel free to provide evidence that they can’t do it if you have it. We’re doing a lot of the heavy lifting for them in this thread and on this forum in general.

resources refers more to putting devs to work on a particular project since they have alot of specialized skills that could be used elsewhere than on the actual cost

@Huskboy

So for bunker you’d say:

Staying alive on point, holding his own against 1 without dieing, ideally holding vs 2 or more. Avoid getting held point capured by enemies and staying alive as long as possible until help arrives.

So here you’d have time in combat without dieing, could have time in combat while outnumbered and time successfully defending that could count towards “good quality” for a bunker role.

I also agree that some “types” of play should be associated with amulets. Support amulets to reward support actions/activities, and so on.

Lets say we take success of individual skills as the only measure of a player’s MMR. Wouldn’t a lower skilled player be less able to get successful “good behaviours” factors to weigh in his side and that this would be self correcting as well though on a more player vs/situation/player basis?

Lets say you have a low MMR game, 9 low MMR and 1 high MMR. What are the odds of the high MMR player to get high returns from his “good behaviour” in that match?

how much success would the lower mmr players get? Those on his team, those on the other team?.

I don’t necessarily think that this should completely replace a pure win/loss system…I think it should defenitly be a part of it though. If it could be self levelling by it’s self then by all means…

Yeah I think you’re right about most of these, like I said win/loss should continue to be a factor as a buffer against variables that matter that aren’t being taken into account/aren’t being weighted properly.

Ok before being able to put up any kind of rating system for anything like this we’d need to be able to come up with some “equivalences”.

So for example, if 1 kill is worth 10 points, how much time in combat without dieing would be equal to that?

say if I kill on average 10 foes per match and spend of my average 10 minute match 5 minutes in combat, and that I die twice in the game, are my combat minutes worth more than my kills alone? are they equal or worth more?

Kills can be the result of long fought battles 1v1 many vs many or just the result of an easy opportunity. They weigh favorably in the match outcome for your team taking out an opponent for the 10s until they get to respawn and the travel time until they get back in the action, so probably closer to 20s advantage for your team.

It seems pretty clear that if I hold out 2 people in combat my team benefits clearly for every second above “x” seconds (with “x” being no greater than 20 for that would be relatively equal to a kill as for as team advantage goes). Then you’d have to look @ average holdout times in 1vx scenarios and rate higher/lower your performance.

My point with the above example is that we can break down different actions/events into parts that we can agree on as being worthwhile or the reverse and grant individual ratings. Sure this probably requires a little more than a system based solely on win/loss, but it’s so much more representative and would sell ESPN so much more.

You could even return individual scores and they would actually mean something. That is, that unless they mean something, the system would need to be tuned until they do for the vast majority.

From Flaw in placement Matches:

Cynz.9437:

Bardly.2571:

I don’t understand how this was supposed to be more indicative of your personal ability. You already have the framework for “Top Kills, Top Heal, Top Defense”. . .just use those.

Ideally, the people who are at the TOP of their games, will continually move up. . .

It doesn’t need to be that complicated.

Because top stats don’t necessary say much. For example as guard or ele you can kill necro minions 24/7 and get insane dmg numbers because of it but not actually kill anyone, as thief you do lowest dmg in game but the dmg you do actually scores kills.
The game is also not as great at tracking buffs like tranq or constant decaps.
What if enemy is trying to kill your Lord and you will lose match if you don’t defend it – you won’t get great stats for it but you probably saved the match for your team.

What if someone is training your necro and you peel for them and then run off to decap – you also won’t get great stats despite saving the fight and the match.

My friend is dying and i drop stealth for him to save him – it won’t show up on stats.

Those stats are very unforgiving for roamers from my experience so far. Most common i get ironically are rezzes for teammates (lol sr) and kills when i am running like murder machine around. My friend plays engi – he usually ends up with 4 stats every match. We do our jobs, just in different ways but stats don’t reflect it.

Couple things he mentions in here can be represented by stats noted above, in rezzes, time to kill (quick kills), and so on. All he says is relevant and there is a way to reflect it in a scoring system once we have a working list.

A rating system has to ultimately be driven by winning and losing. It’s by far the best measure of skill over time.

However, many other games have supplemental bonuses for high performing team members which use things like this. Adding a few points to a win or cushioning a loss for a “carry” team member would be okay.

well the new stats show are already good. I defended a point for about 2 minutes yesterday. First in 1:1 and then the engi got a +1 which finaly dowened me after another 30 seconds. I held the point till the end and got the defender. But it´s not much points. I would say this contributed a lot.

I like the new stats shown, however they have no impact on your mmr rating which will determine the kind of people you will be paired up with. Especially early in season, rng counts for so much that your initial rating is pure luck with whom you got paired up, noobs or other good players and your starting point is the same.

The grind from there is not representative of your actual skill level but rather where luck would have you and it’s a grind to get to your actual mmr score to reflect your skill level. Both good and bad, because it’s based exclusively on wins/losses.

All can agree that it’s what you do as an individual player contributing to giving your team an advantage that sets you appart as being good or bad player.

And yes I do believe that some players should deserve high mmr skills because they make great decisions over being a great 1v1 player, be it decaps, confusing multiple foes, defending against multiple foes, anything that would give a 1 man advantage for some time is good imo if the amount of time surpasses any penalty you would incur your team for dieing doing that practice.