Guild Wars 2

LB now! And if i might add: use the pvp lvl in matchmaking for gods love!

Eyia Hellhide.7320:

Maybe there is still a way to pull the community out of the current state of apathy. Maybe it’s not too late.

A working leader board is the first step. Competitive community without working leader board is a joke. Please stop with the experiments. Take an already proved as working system. We all know where from.

You have all the data needed, make a class based leaderboards. Not big enough player base? Don’t bother with top 1K, make it top 100, 50 or even 10. The player base will increase when people start to compete for the spots.

Don’t tell me again this will tolerate the elitism. It is already here at it’s worst. Because it’s elitism based on assumptions, not on real data. Can’t you understand this? Now give us real leaderboards. Not “soon”, not “when ready”, now.

+1

Eyia Hellhide.7320:

Maybe there is still a way to pull the community out of the current state of apathy. Maybe it’s not too late.

A working leader board is the first step. Competitive community without working leader board is a joke. Please stop with the experiments. Take an already proved as working system. We all know where from.

You have all the data needed, make a class based leaderboards. Not big enough player base? Don’t bother with top 1K, make it top 100, 50 or even 10. The player base will increase when people start to compete for the spots.

Don’t tell me again this will tolerate the elitism. It is already here at it’s worst. Because it’s elitism based on assumptions, not on real data. Can’t you understand this? Now give us real leaderboards. Not “soon”, not “when ready”, now.

+1

I would actually like to see maybe 2 leaderboards, i feel there should be some monthly leader board for the grinders. They enjoy grinding games and they should be rewarded for that and i also feel there should be a leaderboard for the elite players of the game.

LIke they should have to play at least 10 games a week to be on the board and those who have the highest skill level on 1 board and a board for the players with the most points from the current lb

Eyia Hellhide.7320:

Maybe there is still a way to pull the community out of the current state of apathy. Maybe it’s not too late.

A working leader board is the first step. Competitive community without working leader board is a joke. Please stop with the experiments. Take an already proved as working system. We all know where from.

You have all the data needed, make a class based leaderboards. Not big enough player base? Don’t bother with top 1K, make it top 100, 50 or even 10. The player base will increase when people start to compete for the spots.

Don’t tell me again this will tolerate the elitism. It is already here at it’s worst. Because it’s elitism based on assumptions, not on real data. Can’t you understand this? Now give us real leaderboards. Not “soon”, not “when ready”, now.

+1, you wanna test, you open a test server.

I’m afraid they can’t understand the emergency of the situation. I thought they’ll understand when their best shoutcaster left because of this. They didn’t.

Working on the expansion is not an excuse. It will happen somewhere in the vague future, and the players are already tired of getting their daily dose of frustration. And they leave. Why you’d try to attract new players, if you can’t keep your old ones?

It won’t be a expansion of the player base, it’ll be an attempt for resuscitation.

Previous LB (pre 16Dec) was so much better as it more reflected skill on top rather than amount of games you’ve farmed, as we have now.

Bring back MMR LB!

While your at it, make it so you can view it in game. Its annoying to come to the site just to view the LB while I’m playing. Plus, most casual people probably don’t even know there is a LB, and could make more people wanting to play and get better to be on the LB, thus increase the player base of pvp.

Man, the various acronyms really throw me off sometimes. “MMR based longbow? What, so that Point Blank Shot and Pin Down can attack your opponent’s rank directly or what?”

Edit: It’s Matchmaking Ranking based Leaderboard, by the by.

Should just make a sort by MMR tab, and add class filter. Even if rewards are tied to games played over MMR, at least ranked would be competitive.

They need to change the way a displayed MMR would work. A small queue size makes MMR too volatile, so the weight of a loss or a win vs huge MMR spreads during low queue time should be nerfed. Eg; if a 2200 player keeps queuing into 1200 players at 3am and looses, this shouldn’t tank his MMR -and it shouldn’t boost the 1200 players MMR. When population is too low to provide an ideal match, the system should acknowledge any match it can come up with as equal, and adjust MMR as if everyone in the match were facing their own MMR, then divide any gain/loss in rating by 2.

To fix MMR volatility, you have to set up conditionals in your algorithm to deal with what’s making it volatile.

Agree they need to do this, and soon.

81% of people want this, this poll’s been up for 3 months already:
http://strawpoll.me/3638200/r

PvP community is competitive by default. And we need objective and visual evidence of our improvement. We need to know where we are and where are we going.

Right now, we are like runners in a big marathon. We took part, because we like to run and like to compete, but after the start we realize we have no idea where is the final and if there is a final at all, or you just run until you give up.

So we run and run, we got tired but keep running, it’s a mess around us and we don’t know if we’re in the avangard or in the tail, we just run for the sake of running.

We see the same faces around, looks like these people run with the same temp like us, so we start talking to them, it’s always better to run in good company. We become friends and running is even more fun now, but we also hope at some point someone will tell us if we perform good or bad in this marathon.

But the staff of the marathon are just sitting around the road, with big grins on their faces, giving us bottles of water and blocks of chocolate and shouting “go guys, you’re doing great, keep running and don’t worry EVERYONE ARE WINNERS!”

ketchup.2783:

I guess the only way to know how good you are is to simply do community run tournaments. Any stats within the game is not a good indicator.

You guess wrong. And nobody is talking about the stats within the game.

If they let me couple of endpoints on their API, I would do it myself.

EDIT:
hmm, one way to do it is to apply to editorship, might do that (http://wiki.guildwars2.com/wiki/Guild_Wars_2_Wiki:Requests_for_API_editorship)

at the very least, i’d like to see MMR be visible from now on. i as a player have no idea if the matchmaking system is matching me against people it thinks are my match, because i don’t even have a clue how good or bad the system thinks i am. especially when i spend a whole day fighting in 4v5s, or against teams that queue against me queuing solo.

IMO, the leaderboard should state the MMR of every person on it.

ALSO, we should be able to sort the leaderboard by other stats as well, (this has been discussed ad nauseum, so my apologies for beating a dead horse, so to speak) such as all the stats listed per match when pressing “B” for the scoreboard.

ALSO, we should be able to select a specific player, to go to another page dedicated to that player, to be able to see total stats, average stats, etc, for matches played.

ALSO, stats pages and leaderboards should be available during the off season.

Evan Lesh.3295:

Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.

How long does it take to settle? Could the leaderboards simply filter out players who don’t yet have X games? (Where X is 50 or 100 in my mind? More than enough time for folks to settle, I would think?)

Evan Lesh.3295:

Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Similar to my suggestion above, could the LB just filter out players who don’t have Y games in the past 30 days? (Where Y could be 10, for example, I wouldn’t think it has to be large.)

You could absolutely still create another system, but I think there would be value in letting players see their raw Glicko number for tracking and comparison.

Evan Lesh.3295:

[…]
Thoughts?

I think you are right, a 100% MMR based ladder is not the best choice, but it is way better, than what we currently have.

What is your though on a system like the arena-rating in WoW? Accumulating points based on your MMR, lower jumps, after ~20-50 games basically MMR based.

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

You guys had/have a good system in Guild Wars 1 GvG, i don’t understand the problem…

Evan Lesh.3295:

No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Why not? Lack of tech or something?

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

Also in general.

Why not have two MMR-based systems then. A less volatile MMR/ELO/whatever system that doesn’t fluctuate as rapidly, and is visible to everyone and published on the leaderboard. Then along side that one, the GLICKO MMR (and the one used for matchmaking), which should also be published on the side, that shows perceived/potential MMR. Eventually if both systems are working properly, they should both come very close to each other in showing how each player performs in relation to other players.

Frankly though, any MMR system is better than what is in place now.

Evan Lesh.3295:

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Honestly? Even with those ‘minor’ problems, they were still more accurate than the one we’re having now with points, which can basically be farmed, despite the losses.
Even with those setbacks, it felt more right than the one we have now (point system)

Despite a top player being AFK for 1 week, he still deserves a high spot, imo.

I remember when every top player, performing weekly in ESL, was also in top 100 Leaderboard (Team Arena Leaderboard). Right now, the pointsystem is not really attracting them to get into top 100 again. Lots of top players right now arent even in top 1000, because they also feel it’s not how it should be (following their comments and/or chat ingame).
Eventually, this brought non-experienced players with loads of time on their hands and AFK turret engi (back then) to give them a free ticket to place nr1 (and actually dare to call their team noobs for losing the match, while he was afk’ing close whole game!!!!)

Lots of lower ranked team are facing top teams exactly due that problem; top players ranking arent as high as it should be because they dont play ranked as much. The only ‘competitive’ they got are our Community Run Tournaments and/or WTS. They aren’t facing other top teams as much as they should, farming low/beginners/premades while having no practice theirselves.

Nothing can be done about it except if you adjust the leaderboard back how it was or as you suggested; leaning more towards MMR Glicko system to give top players a goal again (maybe something special as rewards top 100; titles, gold, gems, etc) to stimulate them and bringing them back. (please no Llama’s :p) And maybe also make matchmaking more based on Leaderboard ranking, which should feel more balanced. Atleast for SoloQ’ers.

What if we brought back 2 Leaderboards?

• a Point-System Leaderboard like now for SoloQ
• a Glicko MMR Leaderboard like before for Team Arena/Guilds

I think after all, the matchmaking problem and the leaderboard problem is going hand in hand.. One is the source of the other and the other way around..

FYI, I would also strongly suggest to think about:

• Increasing PvP ranks cap (maybe PvP rank 200, just a number. Everyone is r80 now, it’s a slap for ppl who were r55 pre April 2014. It’s like bringing all PvE/WvW levels from 55 to 80 and be treated equally as ‘veterans’) This might sound elitism but isn’t the fact of reaching cap something to be proud of, anyway? PvP Rank 80 is too low to be called a PvP Veteran. Maybe new titles/achievements/rewards post-rank 80?
• Open Ranked Queue at a certain PvP rank, such as 30 or 40. It doesn’t take long to hit 30-40 now (maybe a week, if not less) but alteast people SoloQ’ing won’t have the burden as much to try to carry starters (which shouldn’t be matched together anyway, lol)

Evan Lesh.3295:

timmyf.1490:

How long does it take to settle? Could the leaderboards simply filter out players who don’t yet have X games? (Where X is 50 or 100 in my mind? More than enough time for folks to settle, I would think?)

Glicko settles pretty quickly, the problem is that it doesn’t always stay settled. If you leave the game for a while and come back, your next games could throw your rating all over the place again. A strict games played counter wouldn’t solve everything.

timmyf.1490:

Similar to my suggestion above, could the LB just filter out players who don’t have Y games in the past 30 days? (Where Y could be 10, for example, I wouldn’t think it has to be large.)

We could. Do we want people coming and going from the leader board, but still having the same ‘skill’ level behind the scenes? Will people still just front load their games and afk? Not sure if a system we can actually decay is strictly better or not. I’m just a programmer :P

Doesn’t the second suggestion solve the first? Meaning: " If you leave the game for a while and come back, your next games could throw your rating all over the place again" is solved by “the LB just filter out players who don’t have Y games in the past 30 days?”

You have an MMR of 1800 and leave. After 30 days, you’re removed from the LB. You come back and your MMR drops to 1200 after you get killed in the first game, up to 1600 in the second, blah blah. But until you play 10 games, you’re not re-added to the leaderboard, so it gives your MMR some time to re-settle.

Certainly this will lead to temporary blips, but it’s almost certainly an improvement over the current grind-a-thon.

Have 2 lists, one for the true MMR, and one for leader board rating.

After every match, do the following:

• leaderboard_rating += (true_MMR – leaderboard_rating) * 0.1f

This effectively means the volatility of the true MMR is guarded and can only get closer to the real MMR by 10% each game the player played. Leaderboard rating starts at 0, and is reset to 0 every month (season).

Pros:

• no more top of the leaderboard with less than 10 wins. If the const is kept low enough, the player must play at least ~20 games for the leaderboard rating to get close enough to the true MMR.
• can implement a true leaderboard decay, independent with the true MMR. The player then need to play more games to keep it closer to their MMR.
• no grind, except for about 20 first games of the season. After that, grinding doesn’t increase your leaderboard position much; it will just converge to the true MMR. If you play badly, and your true MMR drops, playing more losing game will make your ranking drops too.
• MMR based, encouraging the players to play more often, without grinding.

Cons:

• more code to write
• grinders are not gonna be happy

P.S:

• tips appreciated! If you find the idea helpful, maybe reward me some gems?

I think having a Leaderboard Rating and Matchmaking Rating that are related though different is fine, so long as we can see and sort by both on the Leaderboards. Meaning, if I really want to, I should be able to view MMR rankings directly by re-sorting the list.

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

Why couldn’t a system be implemented based on kills, deaths, direct stomps, wins losses, and point captures (or something similar). Each category would be assigned a number points out of a total and the higher percentage results in a higher ranking.

So four categories, K/D ratio 35 points, stomps 15 pts, W/L 35 points and point captures 15 points. Then we set the minimum number of games where you are unranked, then once you hit this threshold the system calculates your points and percentages and assigns you a rank based on the points and percentages of other players. I also think the top leaderboard could definitely be narrowed down to 100 and you become unranked say after 1000 or 2000 (depending on the player pool size) no matter how many games you’ve played.

Evan Lesh.3295:

Sunshine.5014:

Have 2 lists, one for the true MMR, and one for leader board rating.

After every match, do the following:

• leaderboard_rating += (true_MMR – leaderboard_rating) * 0.1f

This effectively means the volatility of the true MMR is guarded and can only get closer to the real MMR by 10% each game the player played. Leaderboard rating starts at 0, and is reset to 0 every month (season).

I personally really like this idea.
A goal we’ve had for PvP is to make the game very approachable and understandable. While I think this algorithm would solve some of the MMR leaderboard problems, it still uses hidden numbers (Glicko MMR) that people wouldn’t be able to really understand and calculate themselves. I think there are options that will probably give us the best of both sides, though.

Thanks!

So now the problem is hidden number. Let’s solve it the easy way too: unhide the number.

No, I’m not kidding, lol. Let’s make it so that only the signed in player can see their own MMR. Everyone can see everyone Leaderboard Rating. Only the signed in player can see the matches history, and how many Leaderboard Rating they gained or lose per match. No one will see how their MMR changed per match, only the latest MMR number.

Pros:

• No more bad effect of the hidden numbers. The wiki can say something like “Go to the leaderboard site, you can see your true MMR <insert some link to explanation>, and leaderboard rating. Here is the formula for them.” The whole scheme becomes much more easier to understand for players.
• Still keep the privacy of the player.
• Still keep both the MMR and LeaderboardRating separated (keeping all the pros above)

Cons:

• More code for the website. Maybe new APIs could be nice too

Let me know what you think

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

Why don’t you examine some more respected competitive games like LoL or DotA and see what systems they use to create a respected system, and then implement it? It’s not bad to copy things that work well.

Even WoW’s system for 3v3 is miles ahead of GW2s leaderboards in terms of accuracy. Yes, people sit rating in their system, but it’s not as bad as “whoever plays the most games in highest rated”. Also, their matchmaking has all of my characters sitting on almost exactly 50% win rates, something GW2 has always struggled with (some people with ridiculously high or low win rates like 35% or 80%, although its probably due to the smaller community).

I also don’t see why MMR leaderboards couldn’t be implemented with a small amount of MMR decay over time.

http://worldofwarplanes.com/community/players/1007198141-quiznosBear/

http://worldofwarplanes.com/community/leaderboard/faq/#

Like these guys^
They’ve really got a good, or at very least, motivating, scheme.
Also, tons of stats for us stat junkies.
Have daily, weekly, monthly, all time LB’s. Avoid volatility by adjusting games required for each board, and/or by requiring recent play.

If you let us sort by MMR, glory per battle, awards per battle, decaps, player kills, etc. It’ll really help. Because while I may be a low mmr, I can still brag about being #1 in xyz obscure stat.
At the end of the day, people like to brag about whatever they can, it’s why we compete. We also like to discount other players (to attempt to prove we’re “better”) by saying “oh, well creature kills doesn’t count!”

I mean, being nice is well… Nice.
But what’s best in life is crushing your enemies, seeing them driven before you, and hearing the lamentations of their women.

Also, there’s nothing explaining anything here…
https://leaderboards.guildwars2.com/en/na/pvp

Immediately a few faq’s come to mind.
What are these “points?”
Why do I show a record of 0-1 when I know I went 19-12 in ranked last week?

If someone clicks through to the essentially hidden LB’s they have to dig even further to understand what they mean/indicate.

If you want to truly enter eSports you’ll need more accessible boards.

Quiznos.4296:

Also, there’s nothing explaining anything here…
https://leaderboards.guildwars2.com/en/na/pvp

Immediately a few faq’s come to mind.
What are these “points?”
Why do I show a record of 0-1 when I know I went 19-12 in ranked last week?

If someone clicks through to the essentially hidden LB’s they have to dig even further to understand what they mean/indicate.

If you want to truly enter eSports you’ll need more accessible boards.

Theres info on the wiki.

I’m aware, hence “dig even further”
Is it that difficult to throw a link to the wiki page on the LB page?
And to have the wiki provide some actual helpful information…

For example, it doesn’t say the time range of the LB on the wiki page. Why am I showing 0-1? I’ve played plenty more than a single ranked game. Maybe it’s based on this month, or this week? Idk.

I’m relatively new to the PvP scene here. So please keep that in mind…
As compared to many other games the stats, ranks, and main site visibility are severely lacking. (Aside from the tournaments)

hmm… Wonder what my record is on kylo… Nope, no way of knowing, unless I write it down (6-2 btw excl. d/c)

What we need to do is scrap the current concept of the leaderboard and go with personal rating like WoW.

So 3 people could have a rating of 3103 like in the link below with a rank of 23, 24 and 25 respectively. We all know that someone with a 3103 rating is godlike in PvP (The leaderboard is serving its purpose).

http://eu.battle.net/wow/en/pvp/leaderboards/3v3
-
This way we don’t have to worry about decay or inactivity, we rectify position on the leaderboard based on wins and losses.
-
So if someone gets 1500 rating and stops playing, that wont stop someone else from playing more and possibly surpassing that number.
-
Also, we need seasons where the leaderboard resets.
-
p.s. we NEED to separate it by profession, so 1 person could have 3 professions at different spots on the leaderboard if they are that good.

e.g.

1. Jasher [Guardian Symbol]
2. Magic Toker [Thief Symbol]
3. Jasher [Mesmer Symbol]
4. Helseth [Mesmer Symbol]

Still don’t understand why you try to reinvent the wheel.

The system from other games are fine (like wow-arena, lol, dota, even hearthstone…)

Evan Lesh.3295:

Jasher.6580:

This way we don’t have to worry about decay or inactivity, we rectify position on the leaderboard based on wins and losses.

This is one thing that the current point system has going for it. We don’t need decay because players can pass each other. However, this gives the impression of grind being how to climb. The point system would have to be much less granular for wins to be a good tie breaker.

But as my post suggests, wins should only serve to rectify ties with personal rating. Just like it shows with the link in my last post for the people in rank 23, 24 and 25.

With the WoW system it’s IMPOSSIBLE for someone to play just 30 games and AFK with a rating of 3201 and have the number 1 spot. You simply can’t get that rating unless you play a certain number of games.

Their arena system would be perfect.

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Thoughts?

It doesn’t need to decay to prevent inactive people sitting at the top of leaderboard, if leaderboard will display only players who played during certain period of time (e.g. last three days).
You can also force minimum amount of matches to be played by participant in season to compete for season rewards. However I would recommend short leaderboard seasons (monthly?).
Also I don’t see a single reason to don’t decay MMR directly. MMR is tool to evaluate player skill, we could say that someone who doesn’t play gets rusty and his skill is going down over time while not playing.

Evan, any reply for my suggestion above?

Sunshine.5014:

So now the problem is hidden number. Let’s solve it the easy way too: unhide the number.

No, I’m not kidding, lol. Let’s make it so that only the signed in player can see their own MMR. Everyone can see everyone Leaderboard Rating. Only the signed in player can see the matches history, and how many Leaderboard Rating they gained or lose per match. No one will see how their MMR changed per match, only the latest MMR number.

Evan Lesh.3295:

Jasher.6580:

But as my post suggests, wins should only serve to rectify ties with personal rating. Just like it shows with the link in my last post for the people in rank 23, 24 and 25.

What I was getting at is that Glicko MMR is very granular, and would almost never need tie-breakers to differentiate players.

Okay fair enough. My question though is, what are the challenges of just copy pasting World of Warcraft’s current leaderboard system to Guild Wars 2?

Differentiating placements on different Professions and the whole nine yards… (seasons etc)?

Basically if you did this, a LOT of people would be happy.

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

This can be solved by having short seasons (1 month seasons) and setting a minimum number of games to play per season to appear in the leaderboards for that season, like 50 games.

So you reset the leaderboards every month but keep the mmr of players. The minimun number of games would prevent people from not playing and getting in the leaderboards and also to adjust their position accordingly.

And also reduce the volatility for not playing to avoid people from playing only the last week and boosting their mmr. If their skill level really decreased the mmr should adjust fast even without volatility, and it s very unlikely that someone improves without playing.

I don’t think it s good to have a high volatility for not playing in a month. I think that the skill level of someone doesnt change much from not playing in a month, even 2 months, because there are very few balance patches, and even if there s a balance patch, they just need to read the patch notes to see what changed.

………………………………………………………………..
……………………….
……………………………..
……….
…………………………………….
………….
(censored , promises have to be kept)

The current model = play more games = higher spot in the leaderboad is already here ….
They are using WoW model ….
PPl ATM ARE UNHAPPY WITH THIS MODEL ……

Ppl in WoW have to form TEAMS so they dont have to care about 50% win ratio ….
HERE THE SYSTEM TRIES TO AJUST EACH PLAYER W/L WHEN HE JOIN RANKED AS SOLO , SO HE CAN HAVE THAT 50% WIN RATIO ………

Ppl in WoW have to make new teams , with each different class .
If they want to troll their team8s and afk , if will EFFFECT THEIR TEAMS MMR

Here if 1 person that join solo in Ranked ,
and wants to troll their team8s and afk IT WILL EFFECT HIS MMR AND THE OTHER 4 UNKNOWn PPL MMR
= so having different MMR for different Classes , will mess the leaderboard quite a bit

(solution = if you want WoW model then Ranked que must not be alowed by solo players ….. you have to create a 5man team ……….)

Even a clueless like me , can understand the gist of it …………….

Evan Lesh.3295:

I still don’t think using MMR directly is a good idea for a leaderboard, but I fully understand why people would want it: It is perceived as a better indicator of skill most of the time.

Here is why it doesn’t work well:
Random jumps: Glicko is used to make the best match possible. It does this by adjusting quickly which can result in very large jumps up and down before settling.
No Decay: We can’t decay MMR directly, which causes the problem of people sitting inactive at the top of the leaderboard.

Our best bet is to create a system that lives side-by-side with the Glicko MMR. The original system does its thing and makes good matches while the new one can be a new algorithm, or a filtered glicko algorithm to give us more friendly numbers without negatively affecting matchmaking.

Thoughts?

Hi Evan,

I read the first paragraph and thought of the second paragraph, then read it.

I agree. The rating to determine matchmaking should have more swing factor than the rating of visual leaderboards.

I also think this difference in algorithmic treatment should be most noticeable in ‘volatile’ matchmaking rating such as for new or returning players, and should be less distinguishable in nature once a few (days) of time has been spent using the matchmaking system.

By separating leaderboard and matchmaking rating, you could also produce an algorithm that only factors in a win/loss if the player enters into the queue solo, creating a (solo) leaderboard of sorts. Just a thought.

I’m assuming gw2 will have seasons that do not impact ranked mmr when reset, and doubting it will have leagues. I know it won’t have qualifier points since I heard the infrastructure was built to outsource. If all of the competitive side is outsourced than ladders are being built like a seasonal reward track. Ultimately it doesn’t matter if it is a system that replicates mmr instead of reflecting mmr since it is separated from the qualifying and seeding of tournaments.