(edited by crunchyraisin.6054)
New leaderboards are a good thing.
1.) It attempts to balance player performance and player time spent. We must agree that in order to earn a top spot a player should have to be both active and successful.
Yes, it attempts. But when your leaderboard rank is based 95% on how active you are, and how successful you are is weighted at 5%, I can’t really see the attempt at balance.
There’s one guy on the NA leaderboards who’s rated twice as high as the next guy. He yolo queues every hour of every day, and doesn’t even go one win for every loss.
Top players who nobody can even take a game off them – literally have no chance to be rated well unless they queue every hour of every day to match the people no-lifing the solo queue.
Twitch.tv/chaithh
New Twitter: @chaithhh
(edited by Chaith.8256)
Read what you’re arguing with before you argue with it.
My suggestions include possible solutions to the problems that you reiterated.
What would you think of including a points per day cap alongside incoorporating winning percentage as a factor?
(edited by crunchyraisin.6054)
Factoring in MMR somehow might be workable (though we are already doing that by changing the number of points you win or lose), but adding anything more might awkward since we don’t expose that information.
Using the number of wins, or the win/loss ratio, is asking for problems. The number of wins vs loss does not reflect your skill because it’s not from a large random sampling of matches, and does not include critical information, i.e. 1) who you played against, and 2) how close of a victory it was.
As far as people’s worries about it being number of games played that means you’re at the top, try to keep a few things in mind.
- Seasons will be short.
- A 500-50 record is more impressive (and more statistically significant) than 10-1 record.
- Eventually, we’ll show you the match prediction information and tell you how you need to perform to get X number of points.
The ladder points system already gives players what they are trying to find by looking at wins vs losses, but is also more accurate because it works around the non-random sample by carrying with it the information that is missing from wins/losses. (See 1 & 2 above.)
I’ve looked at the top people, they’re playing games against opponents that are challenging, and they are consistently winning or coming close to winning.
I think part of the issue is that since 1 & 2 are not immediately apparent (we don’t even show it in-game yet), we humans prone to assume it doesn’t exist, is trivial, or is random.
Isle of Janthir: Flux, Latch, Aegir
What puzzles me is that if top players are winning 80% of their matches, then their MMR is not high enough. Shouldn’t they typically be winning 60-65% if they are really good?
Silentshoes (Thief), Wind of the Woods (condi ranger)
What puzzles me is that if top players are winning 80% of their matches, then their MMR is not high enough. Shouldn’t they typically be winning 60-65% if they are really good?
Yes and no. With ideal matchmaking, and a sufficient # of players available at any time, everyone should be close to 50%. With random matchmaking, the best players should definitely have a higher win %.
With real world matchmaking it’s somewhere between the two for any number of reasons, so it’s not uncommon to have people with higher than 50% averages.
Isle of Janthir: Flux, Latch, Aegir
Factoring in MMR somehow might be workable (though we are already doing that by changing the number of points you win or lose), but adding anything more might awkward since we don’t expose that information.
Using the number of wins, or the win/loss ratio, is asking for problems. The number of wins vs loss does not reflect your skill because it’s not from a large random sampling of matches, and does not include critical information, i.e. 1) who you played against, and 2) how close of a victory it was.
As far as people’s worries about it being number of games played that means you’re at the top, try to keep a few things in mind.
- Seasons will be short.
- A 500-50 record is more impressive (and more statistically significant) than 10-1 record.
- Eventually, we’ll show you the match prediction information and tell you how you need to perform to get X number of points.
The ladder points system already gives players what they are trying to find by looking at wins vs losses, but is also more accurate because it works around the non-random sample by carrying with it the information that is missing from wins/losses. (See 1 & 2 above.)
I’ve looked at the top people, they’re playing games against opponents that are challenging, and they are consistently winning or coming close to winning.
I think part of the issue is that since 1 & 2 are not immediately apparent (we don’t even show it in-game yet), we humans prone to assume it doesn’t exist, is trivial, or is random.
I understand that all wins are not equal. You’ve done a nice job of explaining that, but here’s my evidence for why win percentage does matter.
Average players that play a ton, like the one Caithe just mentioned, should not be on the top. He’s a 50/50 win loss player in old solo q with about 4500 wins and 4500 losses.
Top teams are missing out on points because they aren’t being matched up against equal opponents (because there isn’t one playing at the time).
Top players tend to play on teams. With this matchmaking system teams tend to win a high percentage.
If there aren’t better players playing, win percentage goes up. If you’re one of the best players at the time you should be rewarded. The theory of matchmaking causing everyone to have close to 50% win percentages doesn’t work unless there’s always competitive teams and players for you to fight at that time of day. Theres no roof to that ceiling anymore…if that makes sense.
(edited by crunchyraisin.6054)
What puzzles me is that if top players are winning 80% of their matches, then their MMR is not high enough. Shouldn’t they typically be winning 60-65% if they are really good?
Yes and no. With ideal matchmaking, and a sufficient # of players available at any time, everyone should be close to 50%. With random matchmaking, the best players should definitely have a higher win %.
With real world matchmaking it’s somewhere between the two for any number of reasons, so it’s not uncommon to have people with higher than 50% averages.
Brackets sound best to me. Even just 3, Bronze Silver and Gold.
Silentshoes (Thief), Wind of the Woods (condi ranger)
Top teams are missing out on points because they aren’t being matched up against equal opponents (because there isn’t one playing at the time).
I agree with this, and it’s something to worry about and must be included in any reasoning about how the ladder works.
It may end up balancing itself out because easy wins take less time. That’s negated right now by longer queue times, but the build on 16th should include a change that will keep queue times shorter by dynamically scaling min-potentials.
Top players tend to play on teams. With this matchmaking system teams tend to win a high percentage.
Ladder points factor in rating, rank, and roster size when predicting odds of victory. I don’t see how that’s a bad thing. If I went against Abjured or TCG and won (hah!), I should get more points than if they beat me.
Maybe the issue is people believe we are calculating odds too generously? That we’ll be giving out +3 points left and right?
Isle of Janthir: Flux, Latch, Aegir
(edited by Justin ODell.9517)
Honestly I think the issue is very high expectations. I do not envy you.
This reminds me of college football in that way. So difficult to compare teams with different records.
seems to me like pvp got the ol casual catering patch. #1 player on leaderboards is a “bad” player that most people think is a bot because of how mindless it runs around and doesnt help the team. #1 player also has a negative win rate .. yet is #1 .. seems flawed.
seems to me like pvp got the ol casual catering patch. #1 player on leaderboards is a “bad” player that most people think is a bot because of how mindless it runs around and doesnt help the team. #1 player also has a negative win rate .. yet is #1 .. seems flawed.
This logic is flawed. 47% win rate is within the range you’d expect to see if matchmaking is working well.
I’ve looked at their game history… most losses, when they lose, are narrow losses. Maye they’re in some matchmaking sweet spot where they are getting frequent good quality matches. It could also be simply that they’ve been taking ranked serious from day 1, where everyone else hasn’t.
-shrug-
Isle of Janthir: Flux, Latch, Aegir
Top teams/players go out of their way to not queue vs each other so why does it matter if they are at the top of the leaderboards?
I’m a pretty casual PvPer, just hit rank 40, and got matched against Ostrich Eggs and Caed tonight in Unranked. I was so excited, I spent all my time yammering to my guild and forgeting to switch my weapons over from the Courtyard cluster-kitten I had just participated in.
One of my worst performances in a PvP match ever.
I had an evening of pretty lopsided matches playing with my guild. We have widely varying levels of experience, so we might be an interesting case to look at, Justin.
www.getunicorned.com / northernshiverpeaks.org
Here is my problem justin. Leaderboards are supposed to show the best players. Currently there are 4 teams that basically win 99% of their matches vs others. These teams consist of 20 players who I honestly would all rank in the top 50 and ideally should hold the top 20 spots.
Abjured absurd radioactive and the dankening.
There are basically these four teams in top tier PVP. All these teams play at prime time with their full roster.
Abjured – undefeated
Absurd- undefeated
Radioactive- 30-x. All losses to abjured or absurd
Dankening – hasn’t team qd in new system
I understand that these players represent an extremely small portion of the community but the fact remains they are the best.
The leaderboard should reflect this. Honestly I think you guys need a “pro” leaderboard (maybe this is already coming with seasons) where players on top tier teams are shown…. Or it could even be guild based. Regardless the current system just isn’t working for this population of players
I agree a .5 ratio is ideal… But when there isn’t adequate teams to create this, you guys need to swoop In and fix it. ( make bots that can beat us??) currently in your system people are getting yolo points for losing to “better” players in solo q…
(edited by Aaron.2413)
This is because the ladder is unique.
In a bronze, silver etc. league configuration, where top 50/100 players fight for “master league” probably this problem wouldn’t happen because there is already a “skill” filter and basically you need only points to win your division.
After that, at the beginning, how to put people in every division is another problem