Guild Wars 2

here are the estimated probabilities for tomorrow’s matchups. as before, the “na1” and “eu1” files have probabilities of getting a particular server as an opponent, while the “na2” and “eu2” files have probabilities of getting specific pairs of servers as opponents.

-ken

Rigel.3092:

Anet might want to seriously consider getting its collective act together when it comes to WvW or they will definitely see a large number of hardcore WvW’ers (paying customers, as well) head to ESO.

it’s way too late now. the kinds of fixes GW2 WvW needs would take months to implement, even if they pulled everyone off of everything else to work on it.

TESO is only a month away. a lot of people are going to try it out and there’s little that anet can do at this point to stop it. the best plan might be to go ahead and let people give TESO a try, then after a few months when the novelty has worn off, try to get them to come back to GW2 with some kind of event or promotion.

-ken

I do agree that an “enduring community” is important, but unfortunately WvW in its current state maintains the higher-tier communities by bleeding players from the lower tiers. on my server, there is no community left to draw from. the current system that sacrificed my community to keep someone else’s strong, is a failure for the people left behind.

why does this happen? because coverage is king. as a practical matter, the only way a lower-population server can remain competitive versus a higher one is by working harder. too often, it really is “work” harder rather than the preferred “play” harder.

eventually, working harder leads to burnout, causing players to either quit the game or transfer to higher-population servers where they can play more and work less. this is understandable, and I cannot blame any player for finding a way to make their own gameplay less work and more fun. but it only makes the problem worse for the players who remain, and thus the death spiral begins.

EotM is fun in large part because it solves the population problem. in regular WvW there is nothing I can do as an individual to fix the population problem except to transfer up. I cannot conjure up more players for my side, nor is there anything I can do to reduce the size of an enemy zerg.

-ken

in this case, it really was a loss of numbers as well as experience. after earlier departures of some well-known guilds, [ZoD] made a concerted effort to recruit all our militia into one guild. that effort was largely successful, and when they left they took almost our entire militia with them.

the number of players remaining at this point is very small.

-ken

DevonCarver.5370:

I’d say there is a pretty strong sentiment against the idea of collapsing all of WvW into 3 colors, rather than the current world set up. Which I agree with. I think world pride and association is an important part of the way that WvW works currently.

world pride is all well and good on servers where “WvW works currently”. but there are servers where it’s not working, and if you’re on one of those servers your choices are to limit yourself to EotM or transfer to another server. neither of these options is compatible with “world pride”.

further, while EotM is a fun way to pass the time, it’s not particularly fulfilling if the gameplay you’re after is the gameplay represented by the WvW scoreboard. again, if you’re on one of those servers where WvW isn’t working, your options are to stay on a sinking ship (giving up any kind of fun gameplay for the sake of world pride), or transferring to another server. the latter option is the only reasonable one to expect (it’s not reasonable to ask a player to put up with a bad WvW experience simply due to pride) and to add insult to injury, such players are doubly penalized: to escape a situation not of their own making, they must give up their world pride, and they must pay you, Arenanet, to relocate to another server.

while I don’t have a problem with discouraging server-hopping by charging gems to transfer, I do not see any reason why I should have to pay $22 to fix a gameplay problem I didn’t create.

and yes, I know I can use gold instead. at current exchange rates that’s about 107gold, which is an enormous sum considering that I’ve been playing pretty consistently (20+ hours per week) since early access and in that time I’ve managed to accumulate 258 gold.

anyway, I’m getting off track. my point is, “WvW works currently” does not apply on every server. I know that the servers that have the most WvW players and the most competition are the logical place to focus your attention in terms of bang for the buck, but I’d argue that the servers where WvW isn’t working are the ones that need your attention the most, because if you can’t do that you will never escape the stacking problem and all that comes with it.

-ken

Chris.3290:

Lastly, it would remove the contribution of those in a dominant block towards the final outcome. If (hypothetically) a server pulls out all the stops during NA to tick +600, it means something in the current system. With this proposed change, it would make those heroic efforts meaningless.

I disagree that this would remove their contribution, it would simply limit their contribution to the 8-hour segment that they’re on for.

what the current system allows is for one group of players to push PPT to +600, so that later players can accumulate points that they didn’t actually work themselves to achieve. let’s suppose that your server has a big NA presence, and only a fair night presence, and you’re playing a server that has little NA presence, but equal night presence. the current system allows your NA to build a big lead while they’re on, then hand that lead off to your night crew. despite your opponent having equal numbers at night, they start at a disadvantage because maintaining a lead is much easier to gaining it to begin with.

this has been a problem on NA servers for a long time — whatever server has SEA or EMEA coverage does better because when everyone in NA logs on, the servers without SEA/EMEA coverage need to spend hours retaking and re-upgrading everything, which is doubly hard because the SEA/EMEA players had lots of time to upgrade things making them extra hard to retake.

timezone disparity puts players in equal time zones on unequal footing. the time zones that inherit a better position from players to the east have the advantage.

but I do agree that 8 hours seems very short, considering the effort required to upgrade something. there have been times I spent 4-6 hours getting a waypoint built in garrison. why on earth would I go to all that effort only to have it vanish 2-4 hours later? upgrades either need to build a lot faster, or matches need to run longer (say, 7 matches of 24 hours each).

-ken

really, you need to have 2 pins. one for the melee to stack on, and another for the ranged to stack on.

-ken

the risk that new games pose to existing ones isn’t that they might be better, it’s simply that they might divert attention.

a player who takes a break from GW2 for any reason, is a player who is a high risk to never return. it doesn’t matter if ESO is terrible, if people stop playing to try it out, there is a good chance they won’t come back.

this is especially a problem for GW2, which is incredibly casual-friendly. there’s not much of a grind, and if you take a break for a while you won’t find yourself hopelessly outgeared when you return. as a result, there’s no urgency to come back. so you can comfortably put off your return to GW2 for another week, week after week, knowing that it won’t make much difference.

at this point, the vast majority of the players in my wvw guild are no longer playing this game. many of them quit because GW2 wasn’t the game they hoped it would be, but many of them simply took a break for outside reasons. of those, almost none of them will return because they have no good reason to bother, and those few that do will come back, see how few people we have and say “meh, this game is dead”.

-ken

to see your enemies driven before you, etc.

-ken

Cantur Soulfyre.5409:

kaltastic.3146:

I’d love to see ‘Player doesn’t drop loot bags when killed’ added as well’…

This would be awesome. GREAT idea.

this sounds good at first, but when your opponents realize that they’re not getting bags, they are going to figure out that it must be due to outnumbered, and they’re going to call their friends to come sweep your map.

telling your opponents which map your own players aren’t on, isn’t a good idea. this is why players lobbied so long to have the outmanned buff made invisible to your opponents.

-ken

if you constantly have Outnumbered then it sounds like the server you’re playing against is exactly the one you want to be on. they have a lot of people on at exactly the times of day you’re on.

you don’t really need a server that has 24/7 coverage, just coverage when you want to play. I’m not sure that any server has 24/7 coverage, although JQ and BG may come close.

-ken

they are allocated by color. whatever color you are on your “home” map, that’s what color you’ll be assigned to in EotM. so if you’re green this week, and BlackGate is also green, then you might find yourself playing side-by-side with BlackGate players. or with anybody else who’s also green this week, for that matter.

next week, if your servers are different colors, you might find yourself playing against them.

-ken

yes, but not during the first couple months of ESO’s release. I’ll be, umm, “busy” then.

-ken

for me, Outnumbered tells me two things:

1. an enemy zerg is on this map: roamers be careful, defenders man your siege
2. deaths will be free (no repair costs)

death is still something to avoid, of course. there’s no reason to give the enemy rewards if you don’t have to (not to mention the inconvenience of losing your stacks).

the other features of Outnumbered are meaningless.

-ken

here are the estimated probabilities for tomorrow’s matchups. as before, the “na1” and “eu1” files have probabilities of getting a particular server as an opponent, while the “na2” and “eu2” files have probabilities of getting specific pairs of servers as opponents.

-ken

it has been my experience in online gaming that for this kind of problem (one that affects specific individuals but is not a game-wide issue), if it hasn’t been fixed in this amount of time, it never will be.

of course, “past performance is no guarantee of future results”. maybe Arenanet will surprise us. I’m not saying you should give up hope, but it might be wise not to hold your breath.

-ken

if both of the other servers want “the most win with the least work” surely you can think of something your server can do to help make that happen — either help them get more win, or help them do less work.

pick the opponent who thinks they might be able to beat the other without your help, then offer to ensure that they can beat the other. do I have to spell it out for you?

-ken

if the other two are so evenly matched, why don’t you pick one of them and offer to help them come in first place every single week? then the two of you can gang up on the other one.

-ken

double teaming means that WvW is working as intended.

what did your server do to make the other servers think that you are so strong, they must work together to beat you? whatever it was, keep it up.

-ken

I can say from personal experience that 100% map completion is achievable. the wvw objectives are, of course, the easy ones to get. it’s the bazillion pve ones that are painful.

the trick is to take it slow, only doing pve areas when nothing is happening in wvw. for hearts, it can help make it tolerable to limit yourself to only one heart per day.

also, for the tricky spots that you may not be able to do solo (like some of the skill points in Cursed Shore) leave yourself logged off near them, and make sure you never exit wvw by taking the lion’s arch portal — instead, go back to character select and log on again — this will put you back in pve where you left off, making it easy to see if that pve objective you need has been opened up by other players yet.

also, PVT gear (aka Invaders, aka Soldiers) is easy to get in wvw and once you’ve outfitted yourself in a full set of it a lot of those pve objectives become a lot easier to solo. and a good wvw build (especially one with stability) can help you run in, get a pve objective, and get out without dying to trash npc mobs or being interrupted.

-ken

karma training and farming are fun for a lot of people. which is why so many of them do it.

if you’re a roamer or you like smaller scale organized fights, the OP’s statement that EotM is more rewarding probably does not apply to you. this is probably for the best — if you can find a way to get the zergs off of the regular WvW maps the quality of open field fights on the regular maps should go up.

I do not see a downside here.

-ken

perhaps they’re trying to manage WvW the same way they manage PvE.

in PvE, they have the Living Story arcs to keep PvE from getting stale. when the game was new people ran Arah all the time. now it’s rare to see except maybe on the larger servers; people are doing the new content instead.

maybe that’s what EotM is — a way to keep WvW fresh and interesting. first we had ‘plain’ WvW which was fun for a while, but people started getting tired of it and asking “what’s the point?” then they introduced league play so suddenly WvW mattered again, at least for some of the servers, and after that was over a lot of players figured out that PPT only mattered during league play, and outside of that you should only play for fun. except that due to population imbalance, WvW isn’t nearly as much fun as it should be.

so, what to do while we wait for the next season? EotM. good individual rewards, and the color-merging helps address the population imbalance.

the intended new meta may very well be: EotM for fun, and ‘legacy’ WvW during league play when PPT matters.

-ken

here are the estimated probabilities for tomorrow’s matchups. as before, the “na1” and “eu1” files have probabilities of getting a particular server as an opponent, while the “na2” and “eu2” files have probabilities of getting specific pairs of servers as opponents.

-ken

just to see, I ran the numbers for BG using the updated matchup randomization scheme. for tomorrow’s match, BG has an 18.3% chance of getting green, an 81.7% chance of getting blue, and a 0.0% chance of getting red.

-ken

it’s my understanding that, for the first 3-5 minutes after capture (I don’t remember which, but I think it’s either 3 minutes or 5 minutes), an objective can only be claimed by the guild that had most of the players present for the capture. after that period, if the objective remains unclaimed anybody can claim it.

I don’t know the precise rules of how this works. if a zerg rolls through where 30 are from one guild, and 10 are from a mix of other guilds, the guild with 30 should be able to claim it right away (a clear majority). but if there are 20 from one guild, and 10 each from 3 other guilds (i.e. a plurality but not a majority) I have no idea what happens.

-ken

here are the estimated probabilities for the next matchup. the “na1” and “eu1” files contain the probabilities of getting any particular server as one of your opponents. the “na2” and “eu2” files contain probabilities of getting particular pairs of opponents.

-ken

for comparison, I have a Core i5 3570K and a Radeon HD 7950, and I typically get around 60 fps (with vsync enabled), dropping to 14 fps or so during really big fights. I don’t remember for sure whether I’m playing at max settings, but I’m pretty sure nothing is set any lower than “high”.

-ken

there’s some randomness involved. your rating and ranking make one color more likely than the others, but you have a chance to get a different color. how much of a chance depends on the ratings of the other servers.

-ken

sorry, late again. NA numbers only since EU matches have already started.

-ken

it’s working about as well as it can work, unless you like being in blowout matches where you either crush your opponent or get crushed.

at least with the current system, matches are typically close, and in cases where they’re not it’s usually because it’s not possible to make the matchup better for 1 server without making it much worse for another (that is, there are some servers stuck in bad matchups, but they are the “least bad” in the sense that the only alternatives are to give another server an even worse matchup).

and I agree that a better scoring system is your best bet for real improvements at this point.

-ken

I am under the impression that guesting doesn’t work for WvW, only for PvE.

-ken

yes, Vabbi is in a bad situation. they have a 94.2% chance to get red, a 5.8% chance to get blue, and it’s basically mathematically impossible for them to ever get green unless the ratings change dramatically.

-ken

the ratings are actually pretty accurate as they are. yes, that means match stagnation for a lot of servers (mostly in NA), but the alternative is blowouts which isn’t much better.

the only problem is NA tier 1, where the ratings gap has gotten so wide that there’s no longer any randomness at all. ordinarily this would be something to worry about, but during season play those servers will get an opportunity to play against lower tiers and we’ll find out then whether their ratings need any adjustment downwards. and during non-season play, none of the other servers want to play against them anyway.

-ken

it is somewhat random, but not completely. instead of the chances of YB getting green / blue / red being 33.3% / 33.3% / 33.3%, it was more like 23.6% / 21.0% / 55.4%. also, the probabilities change every week (they’ve probably remained pretty close to these numbers, but I didn’t check).

it’s odd to see that YB actually has a better chance of getting green than blue (or at least, they did this week) but sometimes that’s just how the numbers work out.

-ken

I’d change the scoring so that the point value for objectives goes up the longer you hold them. or possibly based on how many upgrades are built.

the reward for taking and holding territory should far exceed the reward for sweeping a map with a huge zerg, then leaving everything undefended until your next sweep.

-ken

yes, for some servers (especially in NA) the pair-wise numbers aren’t too bad, but in many cases the number of distinct possibilities is so high that the numbers get hard to read.

for example, if you look up the numbers for AR or GoM (or Gunnar’s Hold in EU) you will see many possibilities, all with fairly low probabilities. for servers like that the single-opponent probabilities are more useful because it’s easier to see in one glance which servers you’re more likely to face.

this problem used to be a lot worse. the original matchup randomization system had a lot more randomness, which dramatically increased the number of possible outcomes. I’ve attached a file of pairwise probabilities from last June for comparison. at that time, there was so much randomness that these files were not especially useful.

nowadays, with the randomization very much reduced, it probably makes sense to start including these numbers again.

-ken

I don’t think you can calculate it accurately using the numbers I posted. the problem is, the individual probabilities aren’t independent. so you can’t simply multiply the probabilities of x and y together to get the probability of x AND y.

but I can do another simulation run to produce pairwise probabilities instead, which will tell you the likelihood of getting any particular pair of opponents. I usually don’t post the pairwise probabilities because most people are less interested in them, but I don’t have a problem doing it if there’s a specific reason.

likewise, I can also calculate color-sensitive probabilities (pairwise probabilities but also showing the likelihood of getting a particular color combination), but I don’t usually post them because they are usually less useful.

anyway, here are the pairwise estimated probabilities.

-ken

EU file fixed now, thanks for letting me know.

-ken

here are the estimated probabilities for tomorrow’s matchups.

-ken

edit: EU attachment fixed, not sure what went wrong there

colors are randomized each week as a side-effect of the matchup randomization system, but matchup randomization depends on relative server ratings, which means that your chances of getting a particular color may change depending on your own server’s rating as well as the ratings of other servers.

based on the current scores, I’ve calculated estimated probabilities for next week’s matchup. for the next match (i.e. the day after tomorrow), YB has a 24.1% chance to get green, a 19.9% chance to get blue, and a 56.0% chance to get red.

if you play WvW a lot, map completion shouldn’t take too long — objectives don’t typically stay the same color all week, the trick is to be on the map when it changes to your own color. and being in WvW a lot makes it more likely for this to happen.

if you can’t do that, then you can rely on the randomization instead, although that may take longer depending on whether you’re lucky or unlucky with the dice rolls. in that case it’s better to start working on it very early, rather than leaving it until the end.

-ken

sorry for the delay. can’t believe I had to go to page 6 to find this thread.

here’s the numbers for NA, didn’t calculate EU since they’ve already started.

-ken

1v1 puts my stacks at risk.

-ken

here are the estimated matchup probabilities for tomorrow. for those who haven’t seen these posts before (I used to post these in the matchups forum, before it went away), these numbers are based on estimated ratings based on current scores from the current matches. the probabilities are calculated by simulating a large number of random matchups (I typically use 100 million) and counting how often each outcome occurs (this is a statistical technique called “Monte Carlo” analysis).

the “na1” file contains numbers for NA servers and the “eu1” file contains numbers for EU servers. the “1” means that the probabilities shown are for getting 1 opponent (and since matches always have 3 servers that means that probabilities for each server always add up to 200%, because you always get 2 opponents). I can also calculate pairwise probabilities (the chances of getting a specific pair of opponents) and color-sensitive 3-way probabilities (the chances of getting a specific matchup with a specific color assignment) but I don’t usually post them because they are rarely useful.

some comments about this week’s numbers:

just like last week, BG JQ and SoR are mathematically guaranteed to play each other — SoR and TC are too far apart in rating for it to be possible for TC to play in tier 1, or SoR to play in tier 2 (or lower).

DB and BP are almost guaranteed to play each other, and there’s a better than 4 in 5 chance that EB will fill the third spot in their matchup.

in EU, Vabbi and WSR are guaranteed to play each other, with either Blacktide or FoW filling the third spot. Blacktide is about twice as likely as FoW.

-ken

Dayra.7405:

Ken, will you post your probabilities now in this thread?

I’ll post numbers somewhere here, but I’m not sure yet whether this exact thread will be the one. the OP may have had something different in mind for this thread so maybe starting a new one would be better.

-ken

incidentally, it looks like Desolation has a 27.6% chance of getting red next match. chance of blue is 29.7% and chance of green is 42.7%.

this is based on the currently expected ratings for the servers at the start of the next match — as the ratings change each week, the probabilities change as well.

-ken

matches used to be made strictly in order by rating. anet would sort the servers by rating, then take them in groups of three. the top server in the group gets green, the middle one gets blue and the bottom one gets red.

once server ratings settled down, the matches became very stagnant with servers getting the same opponents (and colors) week after week. there was also a problem with ratings in higher tiers becoming artificially inflated, which made the stagnation problem worse.

to address this, anet added some randomness to the ratings when determining matches. originally the amount of randomness was quite high, which allowed servers to play against much higher or lower servers, and as a side effect of this the color assigmnents were for the most part random.

recently, anet has reduced the randomness so you won’t play servers ranked too far above or below you as often. as a side effect, the amount of randomness in colors has been reduced as well. if 2 servers having ratings that are very close together, the chance for them to switch colors is quite high. but if servers have ratings far apart, then the higher-rated server is much more likely to get green than red, and the lower-rated server is much more likely to get red than green.

if you are interested in the actual math, anet has posted about this in the past. see this post for example, which describes the original random implementation:
https://www.guildwars2.com/en/news/big-changes-coming-to-wvw-matchups/

-ken

Thursday already! feels like a Monday to me (first day back at work).

here are the estimated probabilities for the next match.

-ken

sorry for the delay. the holidays have my schedule all messed up; I didn’t even realize it was Friday today until just a little while ago.

I’m only posting NA numbers because I already missed the EU reset.

-ken

I should clarify — there is still some possibility of color randomization within NA tier 1. but the choice of servers for tier 1 is locked in as BG, JQ and SoR. so my “no randomness at all” statement wasn’t actually true.

-ken

yeah, I guess we’ll have to start a new thread when the matchups forum is gone.

here are the numbers for the next match. wow, it looks like the rating gap between NA #3 and #4 is now large enough that NA tier 1 has no randomness at all — it’s guaranteed to be BG vs JQ vs SoR. there’s still a possibility for some randomness for most of the rest of us, though.

-ken