|
Please keep the QQ to a minimum if you do not like this update. We are happy to hear your reasoning for not liking a ranked system, but no "OMG VOLVO WHY" posts. |
On January 10 2014 19:35 nojok wrote:Show nested quote +On January 10 2014 18:39 Velr wrote:
Some of my nicest laning phases i had with complete strangers and 0 communication aside from pinging… If you hit the jackpot and get a lanemate(s) that has about the same idea of how to play a lane there is normally not much communication needed… Same goes for Whisp + X. This, it's really important to have the same way to approach the lane. And indeed wisp + X can work really well, I instant random my hero every game, I hit tiny and one guy did the same and got wisp. He did not do the farm bitching, I understand it's fucking boring to tether to someone who is just farming, but he pinged nicely and let me a good 5 seconds before any relocate (including channeling time) so he got me very fat (785 gpm in a 30 minutes game).
Well, this is completely different to play ursa-wisp with a friend for countless games. Also there is a higher chance that wisp sucks to play and relocate into 5 man etc.. or never relocate all.
|
On January 10 2014 19:41 Laserist wrote:Show nested quote +On January 10 2014 19:35 nojok wrote:On January 10 2014 18:39 Velr wrote:
Some of my nicest laning phases i had with complete strangers and 0 communication aside from pinging… If you hit the jackpot and get a lanemate(s) that has about the same idea of how to play a lane there is normally not much communication needed… Same goes for Whisp + X. This, it's really important to have the same way to approach the lane. And indeed wisp + X can work really well, I instant random my hero every game, I hit tiny and one guy did the same and got wisp. He did not do the farm bitching, I understand it's fucking boring to tether to someone who is just farming, but he pinged nicely and let me a good 5 seconds before any relocate (including channeling time) so he got me very fat (785 gpm in a 30 minutes game). Well, this is completely different to play ursa-wisp with a friend for countless games. Also there is a higher chance that wisp sucks to play and relocate into 5 man etc.. or never relocate all.
That's the same for every hard to play hero, chances are higher people are not good with it. And wisp is very hard to play.
|
Anyone else following Juice's experiment?
http://www.playdota.com/forums/showthread.php?t=1398477
After a 22 game winstreak he went from 2.9k to 3.4k . If things keep going at this rate I think we can agree that:
1. Recalibration just doesn't happen or is not working as intended.
2. Juice doesn't seem to think personal performance count for shit outside of calibration.
I think that makes sense in a way. Uncertainty increases when you have a suprising outcome, but if the system can give you a 50/50 matchup then no outcome is "suprising". So yes it would take Juice unreasonably long to get back to 5.5k :/
|
On January 12 2014 10:52 TrainSamurai wrote:Anyone else following Juice's experiment? http://www.playdota.com/forums/showthread.php?t=1398477After a 22 game winstreak he went from 2.9k to 3.4k . If things keep going at this rate I think we can agree that: 1. Recalibration just doesn't happen or is not working as intended. 2. Juice doesn't seem to think personal performance count for shit outside of calibration. I think that makes sense in a way. Uncertainty increases when you have a suprising outcome, but if the system can give you a 50/50 matchup then no outcome is "suprising". So yes it would take Juice unreasonably long to get back to 5.5k :/
The system shouldn't need to change that drastically so it doesn't. A 3k player simply doesn't magically become a 5.5k player overnight. Perhaps valve will mess with how uncertainty works in the future to solve this issue, but it'd theoretically only be useful for boosters so why bother.
I'd also point out that a 49.9:51.1 match with the 49.9 side winning would increase the uncertainty (although probably by a very tiny amount). The game makes as close to a 50:50 outcome as possible, but the odds of it being that exactly are infinitely small. A record of what his uncertainly was after each game would be very interesting as well.
|
On January 12 2014 11:15 Blitzkrieg0 wrote:Show nested quote +On January 12 2014 10:52 TrainSamurai wrote:Anyone else following Juice's experiment? http://www.playdota.com/forums/showthread.php?t=1398477After a 22 game winstreak he went from 2.9k to 3.4k . If things keep going at this rate I think we can agree that: 1. Recalibration just doesn't happen or is not working as intended. 2. Juice doesn't seem to think personal performance count for shit outside of calibration. I think that makes sense in a way. Uncertainty increases when you have a suprising outcome, but if the system can give you a 50/50 matchup then no outcome is "suprising". So yes it would take Juice unreasonably long to get back to 5.5k :/ The system shouldn't need to change that drastically so it doesn't. A 3k player simply doesn't magically become a 5.5k player overnight. Perhaps valve will mess with how uncertainty works in the future to solve this issue, but it'd theoretically only be useful for boosters so why bother. I'd also point out that a 49.9:51.1 match with the 49.9 side winning would increase the uncertainty (although probably by a very tiny amount). The game makes as close to a 50:50 outcome as possible, but the odds of it being that exactly are infinitely small. A record of what his uncertainly was after each game would be very interesting as well.
To be honest, having read TSR's alg, I don't quite get it. Uncertainty is defined as decreasing with every outcome. To stop it decreasing to 0, it is increased slightly at the beginning of each match. It is this factor that probably has to be tweaked. You'd probably need to look at the streak and adjust it accordingly. If you won 10 in a row, system should become concerned about not placing you correctly, and bump your uncertainty every match accordingly. Until you lose.
Then again, you're completely right that people don't magically improve 2.5k rating points. It's almost entirely irrelevant outside of initial calibration (which follows its own rules)
e: I'm assuming dota has TSR-like alg, which is reasonable I feel.
|
slowing grinding to my true rating yolo
5.5k here i am
|
I do think the initial calibration is probably a bit too important if that's the only time they keep the uncertainty high.
There's (anecdotally) been a lot of people on significant win and loss streaks. A lot more, I feel, than there were pre-ranked. I can't help but think a lot of people probably didn't get correctly reevaluated in placement, and so are now grinding up or down to where they're supposed to be.
|
United States47024 Posts
On January 12 2014 11:15 Blitzkrieg0 wrote:
The system shouldn't need to change that drastically so it doesn't. A 3k player simply doesn't magically become a 5.5k player overnight. Perhaps valve will mess with how uncertainty works in the future to solve this issue, but it'd theoretically only be useful for boosters so why bother.
I'd actually disagree with that. While large increases in skill don't happen quickly, the reverse can happen in the case of a player who stops playing for a very long time and is hugely *over-represented* by his MMR when he comes back. Unless decay happens when a player is inactive, but Valve has not mentioned that this is the case, and even then a large uncertainty change is needed to re-calibrate his skill level because not everyone loses their touch at an equal rate when they're away from the game for a long time.
On January 12 2014 11:46 Belisarius wrote:
I do think the initial calibration is probably a bit too important if that's the only time they keep the uncertainty high.
There's (anecdotally) been a lot of people on significant win and loss streaks. A lot more, I feel, than there were pre-ranked. I can't help but think a lot of people probably didn't get correctly reevaluated in placement, and so are now grinding up or down to where they're supposed to be.
Uncertainty still was't that high in placements to be honest. Even if you started at your appropriate MMR and went on a 10-game loss streak, you're only 500 MMR below, which really is not all that much. People just make it out to be much more than it is.
|
Drastic skill increase could happen though. You just stop playing ranked and play "normals" long enough. So it should be taken into consideration.
Or you could play ranked of party forever and then go back to solo or vice versa.
|
On January 12 2014 10:52 TrainSamurai wrote:Anyone else following Juice's experiment? http://www.playdota.com/forums/showthread.php?t=1398477After a 22 game winstreak he went from 2.9k to 3.4k . If things keep going at this rate I think we can agree that: 1. Recalibration just doesn't happen or is not working as intended. 2. Juice doesn't seem to think personal performance count for shit outside of calibration. I think that makes sense in a way. Uncertainty increases when you have a suprising outcome, but if the system can give you a 50/50 matchup then no outcome is "suprising". So yes it would take Juice unreasonably long to get back to 5.5k :/
I think he proved it for the mid role, but my question/concern comes from other roles. Can you take EGM or LaNm, give them a 2.5k account and have them climb it to 5.5k just playing the 4 or 5 pos? I think that wouldn't be possible. Or it'd take a very long time.
Mid has the biggest impact in the game, so of course he'll climb faster.. I suspect his little project will start to dramatically slow down around 4k because teams are usually a bit more "aware" of how much teamwork effects the game, and you can't just yolo 1v5 like in the 3k's. We'll see.
|
United States47024 Posts
On January 12 2014 13:58 SnowfaLL wrote:
I think he proved it for the mid role, but my question/concern comes from other roles. Can you take EGM or LaNm, give them a 2.5k account and have them climb it to 5.5k just playing the 4 or 5 pos? I think that wouldn't be possible. Or it'd take a very long time.
If you gave EGM or Lanm a 2.5k account and told them they had to grind to 5.5k, they'd probably do it on mid heroes anyway.
Supports being weaker at carrying ladder games is only natural within how pub games work, and this is not just true in DotA, but in the genre as a whole. But I don't see how this is a problem.
On January 12 2014 13:58 SnowfaLL wrote:
Mid has the biggest impact in the game, so of course he'll climb faster.. I suspect his little project will start to dramatically slow down around 4k because teams are usually a bit more "aware" of how much teamwork effects the game, and you can't just yolo 1v5 like in the 3k's. We'll see.
But his teammates become better as well. He's unable to get such an enormous advantage, but the amount of advantage you'd need to actually secure wins will be far less.
He won't keep up such an enormous winstreak, but I would expect him to climb at a comfortable rate until he's ~500 rating away from his real rating.
|
its just unfortunate that if you enjoy/want to play support, but you also want to climb ladder.. You can't really do both.
I will say at 4200, at least 70% of my games are decent games; even having a few base races and generally pretty even, but I just find it so hard to compete vs 4-5 man stacks when im a solo support. Climbing wont be attainable unless I get extremely lucky with games or switch to mid/carry
|
United States47024 Posts
You can, it's just slower.
As long as you're better than everyone else playing support in your MMR, you will continue to raise rating in the long run.
|
There is a guy at 5k+ who only plays treant protector. And that is like the most passive support in the game. Sing has faced him multiple times on his stream.
Also D3xtrs wisp account still has the 2nd highest winrate in the game http://dotabuff.com/players/86736893.
paS wins a high percentage of his pubs playing support http://dotabuff.com/players/31078647
And then there is my friend (just to show its possible to do it as a nonpro) who climbed pretty high as a support player http://dotabuff.com/players/29735591
You can obviously climb as a support..
|
bla bla blah blah
just play better your rating will rise
|
On January 12 2014 21:21 DrPandaPhD wrote:There is a guy at 5k+ who only plays treant protector. And that is like the most passive support in the game. Sing has faced him multiple times on his stream. Also D3xtrs wisp account still has the 2nd highest winrate in the game http://dotabuff.com/players/86736893. paS wins a high percentage of his pubs playing support http://dotabuff.com/players/31078647And then there is my friend (just to show its possible to do it as a nonpro) who climbed pretty high as a support player http://dotabuff.com/players/29735591You can obviously climb as a support..
I would wager that TP guy would be higher rated if he'd play more impactful heroes, same for all the people you listed.
I don't know about your friend but both paS and D3xter have the majority of their games played with friends. If you're gonna be supporting in a stack it's gonna be much easier to climb that way.
MMR is basically the representation of how much impact you have in a game, and the way support is that impact is mostly relevant in the early phases of the game(not always and it depends on hero ofc, but generally speaking). While the mid role will have the opportunity of being relevant throughout the whole game and having more "space" to dictate the flow of the game.
What I'd like to say is that the mid role specifically, will grant you a slightly higher MMR than other roles IF&when you get it on the level of your other roles. Same principle applies for other roles(say offlaner>support).
Think of it this way, say we get a real ladder some day and see people's rankings. Which roles do you think the top100 people on the ladder would mostly play in a solo-queue pub environment?
|
brown ur lower mmr go mid dont pick support ffs
|
On January 12 2014 21:21 DrPandaPhD wrote:There is a guy at 5k+ who only plays treant protector. And that is like the most passive support in the game. Sing has faced him multiple times on his stream. Also D3xtrs wisp account still has the 2nd highest winrate in the game http://dotabuff.com/players/86736893. paS wins a high percentage of his pubs playing support http://dotabuff.com/players/31078647And then there is my friend (just to show its possible to do it as a nonpro) who climbed pretty high as a support player http://dotabuff.com/players/29735591You can obviously climb as a support..
Dont mix winrates in with mmr.
The problem is not that there arent many supports in a specific mmr range. The "maybe" problem is that clinbing the mmr ladder solo is way slower as a support, then as a carry or a mid. Yango "the walkin/talkin dotadatabase" even aknowledged that in this thread if i am correct. Most carries in lower mmr ranges cant even convert more farm into more lasthits even without having an enemy in the lane etcetc. The uncertainly of having "a fool of a took" in your lane is pretty damn high in lower mmr ranges. The supportrole i need to play feels different then what i see in high mmr ranges or on twitch. After raging about these experiences for a few weeks, i accepted it and am changing my heropicks slowly because of it.
Now, i try not to pick my favorite crystal maiden, windrunner or ogre magi anymore, but heros like venomancer or sjaman who to a certain extent can garantuee a tier1 tower early. Our team could snowball from that easy gained goldadvantage, even "those fools of tooks". Also omniknight in low mmr range seems potent, because his ultimate is simply amazing against rightclick heros who i see alot in my mmr range or help "fools of tooks" survive in battles. Ofcourse high-mmr-people must think that i am insane but /care, a man's gotta do what a man's gotta do to climb the social-mmr-statusbar. So in general, i can say that the introduction of mmr is slowly changing my heropicks as a supportplayer, to maximize my early game-impact in almost garantueed ounces o' gold.
|
I'm having harder and harder time reading through all of this. It's like people didn't care much about getting better and were just looking to be pleased or stimulated easily and quickly--
|
On January 12 2014 12:02 TheYango wrote:Show nested quote +On January 12 2014 11:46 Belisarius wrote:
I do think the initial calibration is probably a bit too important if that's the only time they keep the uncertainty high.
There's (anecdotally) been a lot of people on significant win and loss streaks. A lot more, I feel, than there were pre-ranked. I can't help but think a lot of people probably didn't get correctly reevaluated in placement, and so are now grinding up or down to where they're supposed to be. Uncertainty still was't that high in placements to be honest. Even if you started at your appropriate MMR and went on a 10-game loss streak, you're only 500 MMR below, which really is not all that much. People just make it out to be much more than it is.
Climbing 500 mmr even with a decent winrate is about ~200 games. That's a fair amount I'd say, especially if you have other obligations such as work.
|
|
|
|
|
|
EPT
