Support NeoGAF

[Hazaro]

Spoiler

For Hylian

More updates: http://www.playdota.com/forums/showthread.php?t=1398477
DB: http://dotabuff.com/players/86751819/matches

Main MMR: 5600(?)
Start: 2930 MMR -> Current 4650 MMR

Summary: Going from 2930 to 3506 took 24 games, a 24 win streak (24-0)
Summary: The road from 3506 to 4017 took 27 games, (24-3, 89% win rate)
Summary: 4017 - 4509 within 29 games (24-5, 83% win rate)
Overall: 72-8

I'm really starting to dislike this rating range because people are defeatist, egotistical, they think they're very good at the game and try to tutor everybody else (why do they all think they're smarter than the other people at their rating, sigh), they whine and cry from beginning, try to force their item choices and opinions onto others (which are wrong most of the time) and generally are their own worst enemy and the reason they sit there. Games are generally unenjoyable, everybody seems to be fed up with dota here and while people can get annoying at higher ratings I can assure you it's nothing close to this. People who random heroes are generally -1 in the team as they won't repick even something that they can't play (earth spirit from my slark game) and then will just proceed to feed and be useless all game long. People also tend to tunnel vision a lot and die before using their spells or in extremely dumb ways, laning and farming ability slowly starts to get decent, however game sense is still low.

Apparently 'git gud' still applies.

Google Doc with Q&A:
https://docs.google.com/document/d/1uU2uPdhsx6mmJ6dt0ORtGBapbXDEIaNiNLIBrUWkwWI/edit?pli=1

ChaQ's guide to mid:
http://chaqdota.wordpress.com/2013/09/05/chaqs-ultimate-guide-to-advanced-solo-middle/

 

[Haly]

I read both of those yesterday. Helped me put into perspective some of the flaws in my own play, and going to try to get to that next level of mid.

 

[Kard8p3]

I read both of those yesterday. Helped me put into perspective some of the flaws in my own play, and going to try to get to that next level of mid.

Hint : it's void.

 

[Heavy's Sandvich]

Did he play pure support?

All I can see are carries or players with high impact.

 

[Hazaro]

I read both of those yesterday. Helped me put into perspective some of the flaws in my own play, and going to try to get to that next level of mid.

Yeah when I remember and want to improve I read ChaQ and dotastrategy again
http://dotastrategies.tumblr.com/post/63110161888/dota2-early-game-middle-lane-gif-heavy

Did he play pure support?

All I can see are carries or players with high impact.

Read first page of Google Doc

 

[Heavy's Sandvich]

Ok so complaints that MMR is awful are justified then.

 

[Hazaro]

Ok so complaints that MMR is awful are justified then.

Good support play needs part of your team to follow up and recognize the same things that you do to capitalize on them in a coordinated fashion in solo queue.

It's easier just to play smart and farm. Good support is still useful, but just not as useful. As winning.

Spoiler

I'm studying TB

 

[SteveWinwood]

Ok so complaints that MMR is awful are justified then.

because...

 

[Madouu]

Many people seem to be under the impression that TA is a ''ganker'', whatever that means. Most of these people are probably peruvian

Is that you Thacker

 

[Heavy's Sandvich]

because...

He seems to admits that the MMR ranking doesn't recognise those that play well in their support role.

You should read the google doc.

Keep up.

 

[SteveWinwood]

He seems to admits that the MMR ranking doesn't recognise those that play well in their support role.

You should read the google doc.

Keep up.

bad players cant take advantage of what the supports give

and it puts all the bad players together

so

sounds like it works to me

I've read the google doc before bud. It's mostly from the joindota thread.

 

[Boken]

Ok so complaints that MMR is awful are justified then.

can you point to the exact part that allows you to draw the inference that supporting is inherently bad due to the MMR system?

 

[Hazaro]

He seems to admits that the MMR ranking doesn't recognise those that play well in their support role.

You should read the google doc.

Keep up.

You can support better than your other team support and play well, but game is often decided by carries. Your carry needs to understand what to do with the space, farm, safety, stacks, etc. Not dying and getting farm is way easier than playing support since you also need your team to take advantage of the things that you do, and they often already can't take advantage of what the enemy gives them so it's moot.

That's just the way it is unless you play more aggressive supports, not a fault of MMR system. It's just the game impact at that play level.

 

[Quesa]

You can support better than your other team support and play well, but game is often decided by carries. Your carry needs to understand what to do with the space, farm, safety, stacks, etc. Not dying and getting farm is way easier than playing support since you also need your team to take advantage of the things that you do, and they often already can't take advantage of what the enemy gives them so it's moot.

That's just the way it is unless you play more aggressive supports, not a fault of MMR system. It's just the game impact at that play level.

As a support player I don't if I should feel vindicated or bummed. Not that I'm secretly 6k mmr or whatever, but knowing that the heroes I'm most familiar with won't let me climb the mmr ladder without good teammates kinda bothers me.

 

[Heavy's Sandvich]

Firstly, you have your placement matches to ascertain where you belong.
It appears to be weighted in the tangible statistics one can gleen ie gpm/xpm/kdr. It certainly doesn't necessarily have a direct correlation between you winning and losing.

Some say that the placement matches just reflect hidden ranking you had already - so what was the point of the ranking matches then if it was to confirm what was already there?

Then you move onto the ranked matches themselves. As the google doc suggests, if you played your role in an exemplary fashion ie smoke ganks, warded, upgraded courier etc, the ranking system doesn't reward those plays.

It offers points for the win and subtracts points for the loss in the vast majority of games.

If you've afforded your carry every opportunity to do well and he fails to "carry", your input isn't recognised in the binary way points are handed out.

So if you're not going to be rewarded proportionately for your support role, you're going to have a lower amount of ranking points, which means you're going to be placed with people in that ranking, hence being placed with lower ranked players playing in the carry role.

I've performed my role as much as I should, the carry didn't. Yet we're both penalised for that.

So, after all that, clearly from the google doc you can easily infer the MMR is awful as it doesn't recognise anything bar impact play.

 

[Calidor]

Peruvian doto schooling that cool kid. Go ravage!

 

[Haly]

He seems to admits that the MMR ranking doesn't recognise those that play well in their support role.

Supports have less impact in games with low coordination, WHO KNEW?

So, after all that, clearly from the google doc you can easily infer the MMR is awful as it doesn't recognise anything bar impact play.

There is no way to quantifiably measure how much impact any "support" made in their game. You're saying the MMR system is awful because it isn't an omniscient intelligence that can impartially review and judge every single game.

 

[Sigismundo]

Friends, how do I get good at Shadow Fiend?

 

[Marauder Shields]

Friends, how do I get good at Shadow Fiend?

Get good at hitting razes?

In other news Revenge going to town on EG atm. And just as I say that EG win a big team fight.

 

[Haly]

Learn to last hit.

The Raze thing isn't as important, what Shadowfiend demands is consistent mechanics. Raze usage doesn't really come into it until you're gunning for wodota moments in 1v1 mid duels.

 

[Chris R]

EG Cumbacks!!!

 

[Heavy's Sandvich]

Supports have less impact in games with low coordination, WHO KNEW?

There is no way to quantifiably measure how much impact any "support" made in their game. You're saying the MMR system is awful because it isn't an omniscient intelligence that can impartially review and judge every single game.

Ah, hello false dilemma, I was wondering when you were going to turn up.

Either it's omniscient or it's what we have now.

Yeah mate.

 

[Haly]

Ah, hello false dilemma, I was wondering when you were going to turn up.

Either it's omniscient or it's what we have now.

Yeah mate.

What's your suggestion then? Wards bought? Couriers upgraded? Because let me tell you, you can buy all the wards the entire game and still fail as a support, and you shouldn't be rewarded for it just because you deigned to play the least desired role in pub games.

Any 5-6k dedicated support player worth their salt can do whatever Juice is doing anyway. It's not about which role you're "good" in. It's a matter of basic mechanics and understanding of the game.

 

[Marauder Shields]

The comeback is real!

 

[Chris R]

The comeback is real!

That wall!!!

 

[Marauder Shields]

That wall!!!

Yeah this DS has been making great plays all game.

 

[red13th]

those ridiculous DS walls! great game.

 

[Boken]

Firstly, you have your placement matches to ascertain where you belong.
It appears to be weighted in the tangible statistics one can gleen ie gpm/xpm/kdr. It certainly doesn't necessarily have a direct correlation between you winning and losing.

This is incredibly dumb, thats all i can say

 

[Hazaro]

This is incredibly dumb, thats all i can say

I think it's also compared to your hero performance. My MMR went from 3300 (1 game of 10 calibration when you could check) to 3500MMR on 8-2 (I think) and each game was a 5-6 KDA Earth Spirit win. IDK how true it is. Kind of want to take a new unranked account into ranked and see where it places me.

 

[Heavy's Sandvich]

This is incredibly dumb, thats all i can say

That's conjecture on my part.

I was 8-2 in my placement matches and was playing offlane Halo grunt.

 

[Heavy's Sandvich]

What's your suggestion then? Wards bought? Couriers upgraded? Because let me tell you, you can buy all the wards the entire game and still fail as a support, and you shouldn't be rewarded for it just because you deigned to play the least desired role in pub games.

Any 5-6k dedicated support player worth their salt can do whatever Juice is doing anyway. It's not about which role you're "good" in. It's a matter of basic mechanics and understanding of the game.

I mean the dreamleague is trying something with the fantasy stuff.

Maybe something similar in the game, somehow it recognises your success rate at last hitting (whichever role you play), attempts at stacking, successful last hitting with creep pulling, securing runes for teammates etc.

Obviously that comes down to weighting, just seems strange that everyone is lumped in together when you might have done a bang-up job.

 

[SteveWinwood]

just seems strange that everyone is lumped in together when you might have done a bang-up job.

its almost like

its a

wait for it

team game

 

[Satch]

my tinker winrate after 8 matches is 0.00% please help me rectify this it really hurts my feelings that i can love a hero so much and never win with it

 

[Haly]

I mean the dreamleague is trying something with the fantasy stuff.

I don't know what Dreamleague is doing, elaborate?

Maybe something similar in the game, somehow it recognises your success rate at last hitting (whichever role you play)

How does it determine your role is? How does it handle legitimate cases of dual mid vs cases where two people called mid and run into conflict? How does it handle adhoc trilanes with one or more traditional carries?

attempts at stacking

This is doable, and also abusable. It would basically be free MMR to go around stacking camps all day instead of doing what you need to do.

successful last hitting with creep pulling,

You don't always sit there last hitting the pulls, sometimes you have to go back in lane. An automated system cannot make a judgement call on this.

securing runes for teammates

Same here, there is just no way for a computer to say "in this game, CM helped SF secure his rune, but in that game, CM was standing around doing nothing".

Obviously that comes down to weighting, just seems strange that everyone is lumped in together when you might have done a bang-up job.

"Comes down to weighing" is just dismissive of the complexity of the problem you're putting forth. Many other games far simpler than DOTA2 use the same system, flaws and all. Do you really think it's because every company that ever tried to measure "player skill" is lazy? Or is it because the problem is really really hard, and time consuming, not only to pioneer (because there is no precedent in measuring player skill in games like DOTA2), but also to maintain? Ultimately, these weights you keep talking about cannot make judgment calls, and therefore are open to abuse, which will lead to its own host of problems and complaints.

my tinker winrate after 8 matches is 0.00% please help me rectify this it really hurts my feelings that i can love a hero so much and never win with it

Hard hero is hard. No good advice here. You can try to do the Laser/Rocket ganker Tinker instead of the early March farmer Tinker because the former is way easier.

 

[BHK3]

This is incredibly dumb, thats all i can say

I think it's also compared to your hero performance. My MMR went from 3300 (1 game of 10 calibration when you could check) to 3500MMR on 8-2 (I think) and each game was a 5-6 KDA Earth Spirit win. IDK how true it is. Kind of want to take a new unranked account into ranked and see where it places me.

That's conjecture on my part.

I was 8-2 in my placement matches and was playing offlane Halo grunt.

There was a thread on reddit about placement matches being the most important and it was found that super greedy carry players with very good stats like high gpm high xpm were placed almost 1K higher then a support player. I can't for the life of me find it.

 

[Granadier]

It seems to me that support players should focus more on their Party MMR than their Solo MMR since that would show somewhat realistically how well they work with teammates.

Spoiler

don't mind me I'm just a shitty unranked support player

 

[SteveWinwood]

I don't know what Dreamleague is doing, elaborate?

I believe he's talking about the dreamleague fantasy system which awards points in game for these things:

0.3 points per kill
-0.3 points per death
0.15 points per assist
0.003 points per last hit
0.002 points per gold per minute
0.002 points per XP per minute
0.07 points per seconds of enemy stuns
0.0004 points per allied healing done
1 point tower kill
1 point per Roshan kill

Also everyone is either a core or a support.

 

[shira]

http://imgur.com/a/PH2jk#1
Aui pocket alch strat #ti4

 

[Haly]

Seems alright, except I can confidently say that those metrics and weights were reverse engineered from their history of games so that Akke would have a similar amount of Fantasy Points as Puppey, and Mushi would be close to Dendi.

 

[SteveWinwood]

http://imgur.com/a/PH2jk#1
Aui pocket alch strat #ti4

That has to be majorly against dotabuff's tos.

also lol burning not number one for am

 

[Drkirby]

There was a thread on reddit about placement matches being the most important and it was found that super greedy carry players with very good stats like high gpm high xpm were placed I lmost 1K higher then a support player. I can't for the life of me find it.

It was anecdotal evidence with out any proof to back it up. I played support for all 10 of the placement games, and ended up 500 mmr higher then the people I normally stack with, and they are the ones who normally play carry/mid/offlane.

 

[Quesa]

its almost like

its a

wait for it

team game

Yeah its a team game, but what i think heavy is saying that it rewards good carries far more than it rewards good supports. Good carries like naga tb lycan etc dont have to play the team game, supports mostly do.

At the risk of enforcing the meta, I think the system should dock you less points if you play a hard support. Now, you ask "how does valve determine whats a support?" And this would be a case where the more wards, smokes/dusts/gems, courier, and stacks you do or whatever would contribute to you losing less points. You would not gain more points for winning however. I think if this were the case, theyd have to put a cd on all support items.

If people are worried that everyone will just buy wards once they realize they're about to lose, thats what the cd is for and youd have to buy like four sets for it to have a meaningful effect. Like 0 for one set, but like 7 for four. Just spitballin

 

[Heavy's Sandvich]

"Comes down to weighing" is just dismissive of the complexity of the problem you're putting forth. Many other games far simpler than DOTA2 use the same system, flaws and all. Do you really think it's because every company that ever tried to measure "player skill" is lazy? Or is it because the problem is really really hard, and time consuming, not only to pioneer (because there is no precedent in measuring player skill in games like DOTA2), but also to maintain? Ultimately, these weights you keep talking about cannot make judgment calls, and therefore are open to abuse, which will lead to its own host of problems and complaints.

That was never uttered from my mouth.

My point was that the MMR system doesn't appear to recognise the nuances that it purports to regulate = it is awful. It may be very hard but blanket win loss just doesn't appear to do enough for a team game.

Edit: added in a word.

 

[Haly]

My point was that the MMR system doesn't appear to recognise the nuances that it purports to regulate = it is awful.

No system in the world can, it remains to be seen whether such a system can be developed. A "not ideal but workable solution" is by no means "awful".

 

[Schweinehund]

Spoiler

For Hylian

More updates: http://www.playdota.com/forums/showthread.php?t=1398477
DB: http://dotabuff.com/players/86751819/matches

Main MMR: 5600(?)
Start: 2930 MMR -> Current 4650 MMR


Apparently 'git gud' still applies.

Google Doc with Q&A:
https://docs.google.com/document/d/1uU2uPdhsx6mmJ6dt0ORtGBapbXDEIaNiNLIBrUWkwWI/edit?pli=1

ChaQ's guide to mid:
http://chaqdota.wordpress.com/2013/09/05/chaqs-ultimate-guide-to-advanced-solo-middle/

Sounds about right.

Your hero choices get slimmer as you go up, at some point you'll have to play the most competitive heroes in order to boost your mmr, I think that break point is somewhere in the 4k range. I just don't find that fun, I'd rather jerk around with kunkka then tryhard on mirana or invoker for 50 games straight. I think at some point you have to ask where the time your investing into dota is going and if you're still having fun. If you really want to play competitive, I don't think mastering the most trending hero in solo ranked is the best way to go about it.

 

[neurosisxeno]

What's your suggestion then? Wards bought? Couriers upgraded? Because let me tell you, you can buy all the wards the entire game and still fail as a support, and you shouldn't be rewarded for it just because you deigned to play the least desired role in pub games.

Any 5-6k dedicated support player worth their salt can do whatever Juice is doing anyway. It's not about which role you're "good" in. It's a matter of basic mechanics and understanding of the game.

Do people not realize MMR has nothing to do with skill and everything to do with success? I have never looked at MMR as how effective I was at a given role, but rather my ability to win games. If you are a good enough support you do make a difference, and your team will win because of it--short of having straight-up feeders.

 

[Haly]

Heavy's complaint is that even though supports don't have as much game impact as carries, the matchmaking doesn't account for it. Which is kind of a legitimate complaint but misses the point of the experiment, which was to disprove "MMR Hell" and "I belong at a higher MMR than I am". What he wants is a system where a dedicated support player can climb the ladder as fast as a dedicated mid provided they're of equal "skill" for some nebulous definition of "skill".

The only way I can see this working at all is if you queue up for a specific role, like League's Team Builder or WoW's party finder, so the game will only compare you to other support players and your MMR will be in relation to the entire pool of support players rather than the pool of all players.

 

[LiQuid!]

CAN'T STOP FEEDING ANYMORE

There's a Pflax, REO Speedwagon parody here.

 

[BHK3]

It was anecdotal evidence with out any proof to back it up. I played support for all 10 of the placement games, and ended up 500 mmr higher then the people I normally stack with, and they are the ones who normally play carry/mid/offlane.

He did have evidence though, no more anecdotal then the guy playing on a low mmr account to higher mmr or the ranked mmr experiment this guy is doing

 

[Purple Cheeto]

Spoiler

For Hylian

More updates: http://www.playdota.com/forums/showthread.php?t=1398477
DB: http://dotabuff.com/players/86751819/matches

Main MMR: 5600(?)
Start: 2930 MMR -> Current 4650 MMR


Apparently 'git gud' still applies.

Google Doc with Q&A:
https://docs.google.com/document/d/1uU2uPdhsx6mmJ6dt0ORtGBapbXDEIaNiNLIBrUWkwWI/edit?pli=1

ChaQ's guide to mid:
http://chaqdota.wordpress.com/2013/09/05/chaqs-ultimate-guide-to-advanced-solo-middle/

I swear people did similar shit in LoL and proved that if you're good you'll rise to where you belong. Too much Dunning Kreuger's for people to accept this.

Edit: I think most of the people who did this in LoL played carry roles, but I'm pretty sure there are people that have risen up the ladder playing support. The problem is that Elo is a rating system design to gauge individual play ability in a 1v1 game. League and DotA2 are team based games so assigning an individual rating to each player is almost completely missing the point of the games. MMR and Elo in League are almost as dumb as things like W/L records for pitchers.