|
|
Are we assuming 5 man stacks where 4 friends are silver and you're the gold? If matchmaking is perfect, then you should be matched another 5 stack that's also 4 silvers + gold. If you're objectively better than them then you should win now than you lose and your entire team will rank up with you. However, assuming the silvers never learn/ improve, then their ranks are too high and will lose more than win.
If you're just duo queuing with a silver or 2, then it's much more random, but still remains that you are the only constant as the "best" player and should thus win more than lose. Otherwise you are at the correct mmr level. It just will take a lot more games to settle into it.
It's just like dota2 mmr, it works for solo queue/party of 2-3.
I think it's silly to say you can't carry the team, game knowledge and team leadership is enough to win. Direct the sheep and pull a win, if you can't do that then you're definitely around the mmr. Gold-silver is close enough I think where skill is not enough to differentiate. Stick a master player with silvers and they'll probably win a lot more till the team mmr accelerates too fast for the silvers to compete.
|
Bisutopia19299 Posts
On February 10 2015 10:19 xDaunt wrote:Show nested quote +On February 10 2015 10:15 Volband wrote:
A few pages ago you linked a couple power rankings, and both o them had Rehghar pretty low (lowest I think, actually), and some of you even tried to explain why he's terrible, but I've just seen him being picked twice in a row on Grubby's stream and on 2 different maps. Was that some miraculously rare even I just witnessed (like seeing 2 Poppys in a row in some master+ level LoL games), or...?
Also, is bloodlust never worth it on him over ancestral healing? I think Rehgar is badly undervalued in the NA scene. Some of the Euro teams really like him, like WANP.
Reghar cloak wolf spec is badass. It's like a feral druid from wow. I love the burst dps!
|
On February 10 2015 12:21 [SXG]Phantom wrote:I don't think they'll implement it. You can be a little sarcastic though if you press "Y" and send the enemy team some phrases at the right moment, i alway laugh or cheer for example when an enemy missed their skillshot  but i think thats as far as it'll go. I got a question on hero league and ranked, if you skill s to be in gold, but you play with a bunch of silvers as team mates, chances are you'll get matched vs silvers, and as we all know, one person can't really carry the game, so theres a posibility that even if you are gold, you are stuck on silver because you depend on your team? I guess my question is, how could they accurately determine the true MMR/skill of a single player, if he only is rensposable for 1/5 of his wins and loses?
I guess by how well you did on that game. Like K/D/R, EXP contributed, or mechanics in comparison with other people in the game. Just a theory, but this is a good way for them to determine how well youre doing individually. You can lose and your entire team can feed, but if your stats are comparable to the winners, then your MMR might still go up. Just a theory.
|
Naw, the best way to tell if you're doing well is if you won or not. What seems to be a lot of people's problem with Elo is that they're impatient and don't understand that your rating will normalize over time. Just because you think you performed well in 1 game doesn't mean you should gain rating, this would lead to a messy, nonsensical, and very useless measurement.
Since dota/lol/hots aren't 1v1, there will be a lot more variability. The so-called Elo hell that "exists" only in the sense you've hit a skill cap and can no longer fundamentally win games above 50% off your skill alone. Here, people have a problem because they then perceive falsely that their rating should be X, and if they were playing with X rated players they would maintain 50% winrate.
With that said, I do believe Dota did some kind of calibration based off of in game stats, such as hero damage which led to a lot of people during calibration to spam zeus ults. I assume these calculations were patched out, though I assume some other metric exists. Don't know for a fact though.
|
Mexico2170 Posts
So here's the deal: i don't think game knowlegde can alone win/carry a game.
For example, i've had games where people have thanked that i take the leadership, on one game we had 2 IAs at the start and one was saying "we should jsut sit on the base wait for it to end blablabla" until i told him to shut up and play. Then i started to try and guide the team, and using the pings on the bots/ai etc. The one who said that we should sit on the base played well, and one of the people who had left came back. At the end, we won the match, having 2 IAs for at least the first 10 minutes, and one IA the rest of the game, and the person who told us to just sit on the base apologized and everything...but what if he hadn't taken it that way? What if he had gone afk/left too? or started feeding? or the team or team comp simply wasn't as good? We would have lost, and there was nothing even a master league player could have done about it.
I guess my problem is that Hero league, or any rank will never be able to display your true skill level because for better or worse, your team has a lot of influence on your wins/loses (that is, if the MMr only takes into account wins/loses), and thats why i can't wait for team league, to at least know how good my team is compared to other teams, and not clusterfuck that is hero leage right now.
|
The distinction is recognizing that your MMR is a function of all the games you have played and that your true skill is what your MMR oscillates around. That game where you have two AFKs and lose the game because you got screwed may cause your MMR to drop, but over time you will only see this as a momentary decline in MMR. The results of one game are meaningless. MMR is about the collection of all the games you have played.
|
On February 10 2015 13:49 sushiko wrote:
Naw, the best way to tell if you're doing well is if you won or not. What seems to be a lot of people's problem with Elo is that they're impatient and don't understand that your rating will normalize over time. Just because you think you performed well in 1 game doesn't mean you should gain rating, this would lead to a messy, nonsensical, and very useless measurement.
Since dota/lol/hots aren't 1v1, there will be a lot more variability. The so-called Elo hell that "exists" only in the sense you've hit a skill cap and can no longer fundamentally win games above 50% off your skill alone. Here, people have a problem because they then perceive falsely that their rating should be X, and if they were playing with X rated players they would maintain 50% winrate.
With that said, I do believe Dota did some kind of calibration based off of in game stats, such as hero damage which led to a lot of people during calibration to spam zeus ults. I assume these calculations were patched out, though I assume some other metric exists. Don't know for a fact though.
This is entirely incorrect. Hotslogs MMR rankings are highly predictive of how good players, and, using hotslogs, you can see exactly who you are being matched with. It is painfully obvious that Blizzard's matchmaker is throwing low rank players with high rank players against team of mid-ranked players and calling it "good." As just one example, I was matched with a guy today who had 3 reported games on hotslogs for hero league. I'll be generous and assume that there were a bunch of unreported games (like 20-30 of them) and that he was actually rank 40. Why the hell would Blizzard match someone who is rank 4 with someone who is rank 40?
|
Comon man, game knowledge will win you more games than not, especially at lower levels. Stop thinking in terms of single games. Ratings are constantly changing over time and not fixed after one game, right? Shit happens, but all that matters is that your winrate is >50%. The fact that you managed to scrap a win is great, and honestly is the real game in this genre. I mentioned this some time ago in a thread, but dota/lol is all about stomps. Most games are stomps, either for a win or a loss. And it a lot of those cases, the MMR spread is pretty equal, so something clearly happened to lead the game to a one sided match. The stomping wins will equalize with the stomping losses (unless you're vastly underrated or vastly overrated), and really the true test of whether you're improving or not is winning lost games, or closing out really close games.
Don't be impatient about ratings, it's really pointless. Actively try and improve your game via reading up, talking strats, or playing constantly, but concerning yourself over whether you're at the right mmr is only gonna lead to ladder anxiety and stress, which then means you're losing points.
If you've ever played dota/lol as a 5man stack, then expect the same. It's gonna be hard and a vastly different game.
@xDaunt: I know hots is having issues with mmr, I was refering to elo in general. And actually, I think there is something weird with Elo going on with hero level being calculated in. I think we had this discussion before, but there does seem to be more discrepancy in skill level when playing a level 1 hero + rank 40.
edit: Just reread your post and saw you mentioned hero league. I have no idea how or if they incorporation hero levels into calculation, just can't see it. Clearly though the pool of players is still to low for accurate matchmaking, on top of partied groups and such.
|
On February 10 2015 14:03 Blitzkrieg0 wrote:
The distinction is recognizing that your MMR is a function of all the games you have played and that your true skill is what your MMR oscillates around. That game where you have two AFKs and lose the game because you got screwed may cause your MMR to drop, but over time you will only see this as a momentary decline in MMR. The results of one game are meaningless. MMR is about the collection of all the games you have played.
Very well put.
I also think that people in general need to stop taking MMR so literally. It's ok if you believe that 4k MMR makes you a professional player (or being 2k MMR makes you nowhere near that), but there's more to gaming and improving as a concept, than a fucking number. I mean come on..
|
On February 10 2015 14:07 xDaunt wrote:Show nested quote +On February 10 2015 13:49 sushiko wrote:
Naw, the best way to tell if you're doing well is if you won or not. What seems to be a lot of people's problem with Elo is that they're impatient and don't understand that your rating will normalize over time. Just because you think you performed well in 1 game doesn't mean you should gain rating, this would lead to a messy, nonsensical, and very useless measurement.
Since dota/lol/hots aren't 1v1, there will be a lot more variability. The so-called Elo hell that "exists" only in the sense you've hit a skill cap and can no longer fundamentally win games above 50% off your skill alone. Here, people have a problem because they then perceive falsely that their rating should be X, and if they were playing with X rated players they would maintain 50% winrate.
With that said, I do believe Dota did some kind of calibration based off of in game stats, such as hero damage which led to a lot of people during calibration to spam zeus ults. I assume these calculations were patched out, though I assume some other metric exists. Don't know for a fact though. This is entirely incorrect. Hotslogs MMR rankings are highly predictive of how good players, and, using hotslogs, you can see exactly who you are being matched with. It is painfully obvious that Blizzard's matchmaker is throwing low rank players with high rank players against team of mid-ranked players and calling it "good." As just one example, I was matched with a guy today who had 3 reported games on hotslogs for hero league. I'll be generous and assume that there were a bunch of unreported games (like 20-30 of them) and that he was actually rank 40 . Why the hell would Blizzard match someone who is rank 4 with someone who is rank 40?
Because there are not enough players. I'm rank 1 and I often play with ranks 40+. My queue time is ALWAYS 380s if I tag solo, if I cancel before that and do a new search I never find a game. After 380s I will find a game with most of the time people who are not even a bit close to my level (they are around 3K or less) and face enemies who are "decent" (around 3500 or more). I almost never win these games (for obvious reasons).
The only solution is to queue with 2+ people and then you start meeting decent premades/having decent games.
|
Just checking my recent hero league games now that you guys brought it up and I was actually matched with 1.6k MMR Nova when everyone else in the match was 2800-3300.
So their team had overall MMR of 14862 while we had only 13479, Blizzard might have their own hidden MMR's but I could tell that Nova didn't have a clue of what he was doing while in the match and should not have been matched with me, this MMR just confirms it. I guess there's always the chance that he queued with some higher rated friend but shouldnt that put them into like 2k bracket match instead of 3k bracket.
|
Could anyone please tell me what "CC" means? As in
At 10 she either gets a great disengage / CC skill or she gets a good heal with escape and reposition potential.
|
On February 11 2015 00:20 phagga wrote:Could anyone please tell me what "CC" means? As in Show nested quote +At 10 she either gets a great disengage / CC skill or she gets a good heal with escape and reposition potential.
Crowd Control, i.e. stuns, silence, knockback, root, slow, etc.
|
|
|
No more evenom for arthas!!!!!!!!
|
Patch notes have dropped.
So it looks like the core is now invulnerable until all keeps are destroyed? I'm a little confused by what it means that the "core is invulnerable until the keep and fort in the corresponding lane destroyed."
Rewind is now a level 20 talent -- good change.
Rewind is now on Anub, Arthas, BW, Nova, Malfurion, Muradin, Murky, Tassadar, Tyrande, Rehgar, Zeratul (ROFL)
New level 20 talent: Hero Blades -- 20% increased AD and basic attacks slow enemies by 20% for 1 second, on Illidan, Kerrigan, Sonya, Thrall, Tyrael, and Zeratul.
New level 20 talent: Nexus Blades -- Attack speed and attack ranged increased by 20% for Jaina, Hammer, Raynor, Tychus, Tyrande, and Valla.
New level 20 talent: Hardened shield --- active ability that reduces all incoming damage by 75% for 4 seconds, with a 60 second CD; available for Anub, Arthas, Chen, ETC, Gazlowe, Muradin, Sonya, Stitches, and Tyrael.
Fury of the Storm: Every 5 seconds, the next basic attack will deal an extra 200 damage to primary target and an additional 500 damage to each nearby minion and merc; available for Gazlowe, Nazeebo, Hammer, Vikings, Zagara
Gathering power: reduced to 5% initial power with max stacks at 15% (Damn, this sucks for Zeratul).
Regen master was buffed to give flat 4 hps upon learning talent (Yay, I'm sure I'll still be the only player who uses this)
Abathur's evolve monstrosity ult was buffed.
Anub was tweaked, hard to tell if it's a net buff or nerf.
Arthas got nerfed finally. Envenom is gone. Frozen wastes is now level 4. Eternal hunter is now level 1. Rune tape is nerfed from 4% to 3%. Legion of ghouls was also nerfed significantly at level 10, but buffed at 20. Looks like Blizzard wants Arthas players to make some hard choices.
BW nerfed. No more bolt of the storm. Z cooldown increased to 45 seconds. Pixie dust was also nerfed badly from 2 charges to 1 (ouch).
Chen was tweaked. Hard to tell, but I think his overall damage (and definitely burst damage) is lower, but his survivability has been increased.
Looks like Diablo won't be as much of a fat cow to use anymore. He's been tweaked to allow better combos. His attack damage is also buffed a lot.
Illidan lost executioner (not a big deal).
Looks like Kerrigan was buffed.
Raynor got an interesting talent overall. Searing attacks is gone and his damage seems less overall, but he's been given a bunch of extra utility, including a new and improved bribe. Hyperion has been changed significantly. Looks like it's a big yamato strike now at level 10, then the lasers come at level 20.
Hammer's mines were nerfed. Only the middle mine applies a knockback.
Looks like Sonya was buffed. Her skills do more damage, cost more fury, but she also generates more fury.
Stitches hook only received a CD nerf from 14 to 16 seconds (LAME). Gorge had its damage cut in half. He'll still be a top tank. C'mon, Blizzard, you know better than this.
Thrall lost battle momentum. Motherfucking OUCH.
Odin just got raped. Odin's health is now tied to Tychus's at the time of activation, and Tychus dies if Odin does (no more double health bar). Odin also has no more CC reduction, but it does more damage.
Tyrael lost cast aside (why?). Tyrael also got a big buff to his shield with angelic absorption being doubled in effectiveness and send to level 13. On the downside, his smite damage bonus was cut from 30% to 25%. I don't feel like Tyrael does too much damage, so I don't understand this change.
Zeratul -- executioner is gone. Rending cleave is now level 16. Void slash is now level 7 (this is very interesting). Void prison pauses merc camp captures -- ie no one can cap when VP is on it. I think a lot of people are not going to like the changes to Zeratul overall. His burst is definitely lower (though only the gathering power change really affects my play style).
It will take some time to chew on all of this.
|
Haha time for AoE wombo meta friends.
I'm gonna be having a blast
|
I hope they just forgot to add Falstad under Fury of the Storm.
|
Cannot wait to try Raynor  . As for Arthas, this will be very fun to not have a cookie cutter build finally.
|
Hmm bw nerfed pretty hard, time to buy uther who is untouched :o Looks like stitches will still be the top tank pick since arthas and muradin were nerfed. Maybe chen will make a comeback but I doubt it since he lost the most op thing he had : the spammable 1sec root at lvl 7. Time to play some motherfucking rockstar ETC !
|
|
|
|
|
|
EPT
