Guild Wars 2

So I guess we need to discuss match-ups. The previous system did not work as tiers became stagnant. I can already see the problem in this one. Some servers are jumping two whole tiers up or down creating some extremely unbalanced matches. While I still think it just might be better than the previous use of glicko, these unbalanced matches are quite bad also. I would like to propose a different way to determine matches.

First we need to discuss the principles of good matches:

• Balanced matches: are good and should be kept unchanged
• Mobility: if a match is unbalanced, then servers should be able to change tiers faster
• Stability point: it’s important to note that Anet talks a lot about tiers reaching stability. While this concept is good, sometimes guild movements or increased participation can break the current stability point. There should be means to break that stability when a change of forces occurs.
  So these basic and simple principles are what we should be aiming for.

Now, I’ll discuss three match-up systems and discuss what I see as their problems. The three match-up systems I’ll talk about are the previous system of using glicko (pure glicko), randomized around glicko (random glicko) and winner up loser down (WULD) which is also quite popular in these forums. Then I’ll talk about my proposal which uses glicko with extra means for mobility (mobility glicko).

Pure Glicko
Not much to say as this has been beaten to death. While it keeps balanced matches, it does not provide mobility. It also never reached a stability point as many matches were still quite unbalanced.

Random Glicko
While this system fixes the mobility issues in the pure glicko system. I don’t think it can achieve either keeping balanced matches or reaching a stability point. First, it has no means to keep balanced matches in place. I guess having a balanced match will be completely up to luck. Unless, of course, the deviation in the glicko formula goes to zero with time and the ratings distance themselves. I’d like to note that I don’t really think it will. I actually think the only way for the deviation to go down is if the ratings come closer together. And even if we do reach some stability point, there is nothing to break this stability in place if any server’s activity increases at some point.

WULD
This does have a few advantages on the mobility aspect and breaking stability than pure glicko. But the main problem of this system is that there is no way to keep balanced matches. No balanced match is played twice in a row. So at least 50% of the time, matches will probably not be balanced.

Mobility Glicko
This is the system I propose for match-ups. It’s based on the glicko score but it should provide mobility whenever any server is unbalanced in it’s tier. So we keep servers on their glicko rating like in the pure glicko but we can momentarily move a server to a different tier if it’s not balanced in that tier. To define this extra movement, I need to define when a server overpowers another. When server s1 and s2 play in a week, we say that s1 overpowers s2 if s2’s score is less than 80% of s1’s score, i.e., if score(s2) < 0.8 * score(s1).

The matches that need to be broken are the matches in which one server overpowers both other servers (much stronger) or the matches in which a server is overpowered by both other servers (much weaker). One way to increase mobility is to move up a server that is much stronger than the others or to move down a server that is much weaker than the others. Moving a server up means moving it to the higher tier dropping the server with the lowest glicko score in the tier it’s entering. This will break the glicko order just for one week. If in the next week its score went up enough for it to stay in the higher tier, then it stays. Otherwise, it will go back to wherever it should be based on it’s glicko score. We can define moving down similarly by taking the server with the highest glicko score in the tier below to move up.

Although this is an improvement over the WULD approach to keep balanced tiers in place once they get there, I would actually recommend a server moves up only if it’s much stronger or much weaker for two weeks in a row to avoid moving servers in an odd vacation week.

Ok guys, what do you think? I hope I put this up for discussion in a nice way.

Anet is attempting to do what you state here with the new system.

The issue is that there is no real starting point except the old glicko rank. Give this new system a few weeks to flesh out before we start long discussions on the matter. Basing arguments on a single data point is silly.

Doesn’t matter what system you use. The only way to get balanced matches is to get the player base to spread evenly among all the servers. Make no mistake the player base is as much to blame as Anet. From the first week with HoD, JQ, and every stacked server since then. Even now servers which are showing FULL status are trying to shove players in sideways.

Anet deserves some blame. You shouldn’t be able to transfer to full servers at all or it should cost twice as much. Bottom tier servers should have free or at least drastically discounted. There should be rewards for fighting as the underdog. Right now everything from the scoring system to class design punishes you for trying to fight outnumbered.

However, its the players that have the final decision. There really isn’t anything anet can do short of forcibly moving people that will make things more competitive. No one forced the player base to stack on a few servers. Look at all the most recent guild transfer posts. Are we trying to funnel players to servers that need them? Nope it’s all JQ, BG, DB. Servers that have full populations.

Anet can alter the transfer system to funnel players to servers that actually need the population, but it is up to the players to actually spread out. No system you come up with will result in balanced matches because the player base as a whole does not seem to want balanced matches.

Draygo.9473:

Anet is attempting to do what you state here with the new system.

The issue is that there is no real starting point except the old glicko rank. Give this new system a few weeks to flesh out before we start long discussions on the matter. Basing arguments on a single data point is silly.

I know Anet is attempting to do make people happy. But this discussion is very important for WvW matches right now.

There are a few peculiarities of WvW that make it so a proper match-up system is critical, all related to the fact that a WvW match lasts a full week. The main issue that every week a sever is mismatched, it is not fun to do WvW with all players in 3 servers for a full 7 days. This is quite a lot if you ask me. For example, in NA this week, I can only see good matches in tiers 1, 4 and 6. That means 5 tiers are unbalanced and 15 servers are filled with unhappy players. Isn´t that a bit much? The previous system of pure glico had more balanced matches than this and more happy players.

Another reason to make this discussion important is that there are two ways to analyse the outcome of a match-up system, in practice or in theory. In practice is to observe what happens in a few weeks, as you mentioned. The second is to stop and think a little on theoretical aspects that we find important for a match-up system and check if a system satisfies those requirements. This is what I´m trying to do. The problem with the practical way is that it takes many matches to analyse the match-up system and those are more weeks leaving the players unhappy. So we have a week of bad match-ups now. Say we have another bad match-up next week. Then on the week after. When will we start discussing this? Hopefully sooner than later.

Yshyr.8709:

Doesn’t matter what system you use. The only way to get balanced matches is to get the player base to spread evenly among all the servers. Make no mistake the player base is as much to blame as Anet. From the first week with HoD, JQ, and every stacked server since then. Even now servers which are showing FULL status are trying to shove players in sideways.

Anet deserves some blame. You shouldn’t be able to transfer to full servers at all or it should cost twice as much. Bottom tier servers should have free or at least drastically discounted. There should be rewards for fighting as the underdog. Right now everything from the scoring system to class design punishes you for trying to fight outnumbered.

However, its the players that have the final decision. There really isn’t anything anet can do short of forcibly moving people that will make things more competitive. No one forced the player base to stack on a few servers. Look at all the most recent guild transfer posts. Are we trying to funnel players to servers that need them? Nope it’s all JQ, BG, DB. Servers that have full populations.

Anet can alter the transfer system to funnel players to servers that actually need the population, but it is up to the players to actually spread out. No system you come up with will result in balanced matches because the player base as a whole does not seem to want balanced matches.

Sorry but I don´t believe that´s the only way. Low population against low population is fun in the sense of small scale fights. High population vs high population is also fun for high level strategy on the battlefield. They can coexist and both being fun.