^but we got the proof from the first rime, and the second, third, fourth....
ah jing...i remember him from the 2011 garena clan war days. starting item 4 slipper drow/4 mantle invoker = autolose his lane.
i think he improved quite a bit since coming to dota 2.
i think all the noobs are just happy they get thanked for letting their account get boosted 2k+ mmr where they normally would have to pay hundreds for
"I would still like some one who is a 2900-3300 MMR try and start at a 1800 and see what it would be like. I still feel like its much harder for a middling player at roughly 3k carry their 1800 team to consistent wins. "
lmao
^ i think that is actually true
I was in a game yesterday when my party mates were ranked 1.5k against 2.8~3.2k players. It was pretty much a breakeven at top 2v2 (dazzle+wk vs sk+pa). Thats more than a 1k difference but the laning phase was entirely breakeven except for maybe 1 kill. They lack the ability to capitalize on shit.
If you're wondering the MMR distribution it was (unranked but we got out the MMR's early into the game)
4.8k+4k+1.5k+1.6k+1.5k versus a 5 players between 2.8~3.2k.
yea but he means if a 3k player hopped on a 1800 account
anyways the team with the 2 4k players should win well over 70% of the time, just have the 1k players pick retard heroes (lich/cm/wraith king) and you're set
But thats the thing, 2v2 lane;
3k+3k versus 1.5k+1.5k
And it was breakeven! If it was a 4.5k+4.5k versus 3k+3k, it would be a whole other story of feedfestivals.
how fucking stupid are people to think they can get from 2.2k to 4.5k
the most difference you're going to get is 500-1k by spamming tinker and shit but if you're seriously a 2-3k player you're never gonna be 4k with just a new account
These experiments are fucking bullshit and prove absolutely nothing.
Oh hurrdurr I'm a 4-5k player give me a 2k account and I will get it up to the ranking i am at currently. See everyone, anyone can get up in rank hurrdurr.
@havoc
how so? it proves a player really at a given rating can get his rating if he's under calibrated, which basically disproves all those "hurr durr i got shit teammates for calibration im really a 5k player on a 3.8k account" complainers
it only seems retarded because its in response to the complainers, who are really that retarded
this guy documented his progress well too, using all sorts of roles and heroes, and annotated every single game he played, raising both party and solo to close to his original levels
he was never trying to say anyone can go up in rank, the whole point of these experiments is to prove those "idiots who believe they are being held back solely by their teammates" wrong.
the effort at least is commendable
havoc why such irrational rage, this is uncharacteristic of you..
That wasn't the purpose of the experiment, plus these were just a 5K player stomping trash till he hit about 4.2K then a lot of his stomps stopped. This is in no way helpful and doesn't prove anything at all except that 2-3K players are easy to beat if you are 5K rated.
If he had stated intent like this is to prove that if you are stuck in 2k you belong there that would have been fine, but this was yet another "ELO HELL" doesn't exist experiment where he was attempting to prove that it doesn't exist for players who aren't skilled enough to get out.
I too appreciate the effort but as stated here http://www.reddit.com/r/DotA2/comments/291pvl/my_mmr_experiment/ I would much prefer to see some 2K player fight and struggle his way to 5K and document how he got there and what he did. If someone did that it would greatly help out some of the players who are kind of lost about what they need to do to improve as in their minds and based on the players they see every game in their bracket, they're fine.
a 2k player struggling to 5k is something that takes year and thousands of games... and is of course possible, all of us 5k players were 2k, 1k players once, there's no need to prove something like that
its to contend all those complainers on every forum that they say stuff like they didnt take their calibration matches seriously, have shit teammates consistently, etc etc and flame the matchmaking system for underrepresenting their rating when they are really 5k players stuck in a 3k account and they can prove it, they once did better than a 5k friend in party queue and it obviously shows hes should be 5k etc etc
if you go to other forums more (gosugamers, playdota, joindota etc) you will see threads created by these people by the dozen daily, and these experiments are meant to shut them up
everybody knows its obviously possible to struggle from 2k to 5k, nobody ever started at 5k.
in statistical terms:
Null Hypothesis: A player is unable to win more than 50% of his games despite having a "true mmr" of at least 1k above the rating given by the system.
Alternative Hypothesis: A player is able to win more than 50% of his games when having a "true mmr" of at least 1k above the rating given by the system.
These experiments are sufficient to disprove the Null Hypothesis at a statistically significant level - which proves something that logical people have known all along - you will win more than you lose if your "true rating" is above your current rating.
It's just that there are some delusional idiots who don't recognize that.
oh come on havoc you know what i mean, wave was once a 1k scrub like all of us (oh the horror)
No I completely agree with you but equally a lot of these aren't as noble of intent as you claim and are just done because...
My issue with this and all the nonsense theory crafting is one of the reason I started coaching people a little because sometimes people need to hear what can actually help them either improve, or win, or both. Sometimes instead of theory training people actually need to do practical training and realize that to get good it takes work and practice.
The main flaw of this guy's experiment is the mode. CM and CD are very underpicked modes but they GENERALLY allow for proper games (trilane,offlane,mid). The hard part of climbing lower MMR is dealing with trash lineups and generally uncooperative teammates. Thats what everyone QQ's about.
Its waaaay better than juice's experiment though
that shit was fucking retarded. sure, pick mid everygame using snowball heroes. way to prove nothing. not everybody has that playstyle or wants to play that way and other roles are much more team dependent (carries,supports,offlane).
Everyone who agreed with Juice was a simplistic buffoon.
I would like to see an experiment of a 4.5k player working up from 3k playing all roles in all pick. That would be interesting.
jing stopped at 4500 because once u get within -500 from ur "real rating" the games get pretty hard (i'm 4600-4800 solo rating usually and this account's at 4300 after loss streaks experienced months ago)
http://dotabuff.com/matches/740652292
my most recent game, i don't think a 4300 player would've won this one. (i learned the refresher + sheepstick build from wagamama and that's what won me the game, if i went for full dps items which is what most 4k players would do it would've been a definite loss)
"The main flaw of this guy's experiment is the mode. CM and CD are very underpicked modes but they GENERALLY allow for proper games (trilane,offlane,mid). The hard part of climbing lower MMR is dealing with trash lineups and generally uncooperative teammates. Thats what everyone QQ's about."
lies, when i played cm and cd i got shit like support faceless, carry lina all the time. it's as much of a gamble as all pick, he simply picked that mode so that he wouldn't get to choose his own hero
as for the hard part of climbing "lower MMR", that is applicable to all echelons of play, even page 1 games you have retards who last pick carries because they don't want to support. its a point irrelevant to MMR.
"Everyone who agreed with Juice was a simplistic buffoon."
yea only idiots who misunderstood the experiment and extrapolated it unnecessarily, it basically showed that if you're good at a certain set of heroes and you keep playing it you'll definitely get to your real rating
people who disagree with that are simplistic buffoons.
What if a player smurfs and gets his smurf rating higher than his main rating(ex main rating 3500, smurf 4100). What does this prove?
learn from the master, not the student
http://dotabuff.com/matches/601882387
my hero damage is low cuz i wasn't very efficient at moving around the map at times, didn't pop bkb at the right moments, + they had tinker eb, malefice, 3 sheepsticks, etc. i didn't say i played well that game - i just said "i don't think a 4300 player would have won this"
@ rhapsarus
it doesn't mean anything, the difference between 3500 and 4000 isn't very significant. furthermore smurf detection isn't 100% accurate, you need to play enough games for ur rating to settle within a range.
that's quite easy
it proves that he gamed the system for anti-smurf calibration, but if he plays enough games he'll get back to his "true rating"
don't forget that people still improve while playing the 100s of games to calibrate smurf accounts too, so in that time his "true rating" might have been going up
I get it tho why some people can't climbe. Let's make this analogy. If you have say 3000 games you are at some level. If you make a new account the first matches are piss easy(you can carry them no problem). But when you can't carry those noobs anymore, when your matches arent easy stomps anymore, it means that the mmr finally found your rating.
My analogy goes with a perfect world where anyone wants to carry but you get the ideea.
the point of the matchmaking system is to provide "equitable and balanced" games. whether or not it succeeds in doing so is debatable - the goal isn't so every 2k/3k player "learns quickly" and advances to 5k. i've been playing this game for 6 years on and off and my rating is ~4700 solo/5000 party (taking the higher of two numeric values from both accounts). i'd say i'm a bit above the curve but nothing special. obviously there are naturally talented players who are 5500+ even though they've never played dota 1, HoN, or league, but they are few and far in between.
it doesn't matter if you're in the top 1% (me), top 0.1% (zenoth), or whatever, you're still going to feel inadequate as long as u find the willingness to improve.
why are u in such a rush to play at a higher rating? ur not planning on going pro so treat it as a hobby, not a chore. i'm actually going to calibrate an account sometime within a week or so (i'm hoping i can get 5k solo, but more likely it'll be around 4700) and if ur that desperate to i can lend it to you for 10-20 games, but i'm telling u you're just going to feel frustrated by the skill gap (unless u want to play a passive support and pray ur teammates carry u every game, which is hardly my idea of "having fun")
I think the problem with climbing for people who are not 5/6k players is that you simply have no fucking clue what to do better - which is why people like Havoc are awesome for people c. 1000 mmr below.
If you're 3.5k you face other 3.5k people. You do well you do bad but at the end of the day you're all 3.5k players of roughly the same level. That means that you don't learn any of the skills a 4.5k person because the people you are learning from/with are fucking 3.5k players - you don't see any of the 4.5k moves/reflexes that makes a 4.5k player a 4.5k player.
Here's an example: I was a high diamond sc 2 player during seasons 1 and 2 (equivalent of 5k). One of my mates from my old brood war team became GM when they introduced it at season 2 (equivalent of 6k). When I was playing against other high diamond players, I found it very hard and balanced. But if ever I played against a friend who was gold or anything, I would absolutely shit on them - I could do absolut troll build and still win on the basis of strong mechanisms and game awareness.
When my friend who was GM was watching me play vs other diamonds players however, he kept pointing out mistakes/opportunities that I wasn't even seeing. The reason he was seeing them and I wasn't is that because in his games at his level these sort of things didn't even exist. So when someone did what at my level was a slight mistake, for him it immediately stuck out like a big red flag and he could instantly capitalise on it.
I had numerous games with him coaching me where I ended up winning because he was poiting out shit that I wasn't even seeing. The problem is, because for me these were "normal" movements/actions in my bracket, I did not even see the mistake being made. Thus I never/very slowly learned.
I think that analogy holds true in dota: it's really hard to understand what you're doing bad if you don't know that what you're doing is bad to start with. And how do you learn that? By watching what good players you play with do. If you all play at the same level, then more often than not you all do the same shit, and no one learns anything.
Sorry for the long post.
Juice's experiment wasn't to show if you were good at a certain set of heroes, you would rise in MMR. That is your interpretation of his experiment.
His experiment was PURELY to 'show' how ELO hell doesn't exist by climbing through the ranks at a fast rate.
Thats why his experiment was retarded. It was narrow minded and simplistic beyond belief (pick tiny pool of heroes in mid, repeat for 100s of games over three months). And people were taking it like it was some well though out shit and using it against every1 who moaned elo hell which I do NOT believe exists but it doesn't change the retardacy of using juice's experiment as some holy grail.
Do not be a simplistic buffoon.
#shotsfired
and i have no idea what sorta CM matches you played at high 4k or 5k (you're indirectly suggesting higher mmr does not mean smarter people). Even at 4k bare my captain's mode had a far more higher constancy of proper lineups and plays. I played well over a 100 games so its not baseless assumption.
"I think that analogy holds true in dota: it's really hard to understand what you're doing bad if you don't know that what you're doing is bad to start with. And how do you learn that? By watching what good players you play with do. If you all play at the same level, then more often than not you all do the same shit, and no one learns anything."
this is true, but the 4.5k players in my game will pick 3 mid heroes, flame the 5.6k player for not being able to support 3 of them successfully, then QQ on forums about shit teammates
>true story
>people who should be learning from the better players flame them all the time instead
And no, you're wrong regarding consistent lineups not being related to MMR.
Are you seriously saying 2k games have better lineups, coordination and warding whatever shit than 5k games... if you say yes then I cannot argue further since I don't play at 5k.
It happens, so? It happens MORE at lower MMRs.
oh please you are hopeless
it doesnt matter what he says about it, the crux of the issue is about how you interpret it in a meaningful way
yes, he himself is guilty of over-extrapolating the data from his experiment, no one is denying that. but my point still stands. that it proves playing the heroes that you are good at will bring you to your actual rating.
at lower rating the skill level is so horrendous that the lineup doesnt matter. the level of play affects the game much more than the lineup could possibly hope to. meanwhile you come up with the shitty argument that "Are you seriously saying 2k games have better lineups, coordination and warding whatever shit than 5k games... if you say yes then I cannot argue further since I don't play at 5k."...
right i never said that, i said it doesnt matter because there are always assholes who pick what they want, and that at lower ratings where people are too ignorant to pick better the lineup is hardly one of the problems in the game.
meanwhile havoc makes a post on reddit complaining about these experiments, still completely missing the point of them.
havoc you want to see a documented series of games from 3k to 5.6k? I have a sample size of 2000 games right here, it's called my game history.... every single 5-6k player is a living "experiment".
these "experiments" have never been about telling weaker players how to improve. it is about telling them to wake up their bloody idea, accept that they are bad, and from there learn to improve, rather than forever blaming everything but themselves and getting stuck in a plateau due to their mentality.
if you misinterpret raw data and try to form a wrongful conclusion from it obviously the data will seem invalid to you.
havoc i get that you're a nice guy and out to help people but all the replies on reddit have basically said the same thing i did. that you are misinterpreting the data. just lighten up man... you've been seeing the nicer side of things, the people who want to improve but don't know how. these experiments are not targeted at them nor meant to help them, but to disprove the delusional idiots who keep complaining that they are underrepresented by the system.
"If you're 3.5k you face other 3.5k people. You do well you do bad but at the end of the day you're all 3.5k players of roughly the same level. That means that you don't learn any of the skills a 4.5k person because the people you are learning from/with are fucking 3.5k players - you don't see any of the 4.5k moves/reflexes that makes a 4.5k player a 4.5k player."
You win 3.5k games by killing the retarded 3.5k players on the other team. You find out their weakness and exploit it.
And yes, Dota 2 does give you the option to learn from better players, via streams/replays/live games/partying up with higher rated players. It's your choice if you want to learn or just grind games without gaining any new knowledge.
I once played on my friend's 2k mmr account, i gotta tell ya, i played 4 games and in all of them i had 20+ kills and a easy win. One of those games i was the Crystal Maiden.
No you're hopeless...
Don't put words into my mouth nor change your own words.
Juice's experiment was, I repeat, was to show how ELO hell didn't exist. It was not to show how to climb MMR or anything like that. His method of conducting the experiment I've explained earlier, was simple and somewhat stupid. Therefore, anyone who agreed with him from this experiment was a RETARD. The funny thing is, from what you've said on the way you perceive Juice's experiment, my simplistic buffoon comment wasn't even directed at you.
So you came in out of nowhere and shot bullets at me.
Also, don't change the god damn subject about how lineups don't matter. Maybe it doesn't when you're a leaderboard player playing among noobs, but it certainly does matter to players who play similar to the rest among them or slightly better (hence being held back). Thats why the bloody term 'outdrafted' exists.
Once again, regarding this, you came out of nowhere and shot me. You don't even have freaking CM mode on your profile! And you shoot down someone who has played 200 freaking games in that mode.
So honestly, it was fucking dickish of you to act like a know-it-all here. Before your ego blows up, just reflect on what I've just said, okay?
I didn't say this experiment was crap. I only pointed out a variable it didn't account for, hence it being the main flaw. Juice's experiment on the otherhand, was stupid. Just because you got something out of his experiment doesn't mean it was good! For its intended goal, it SUCKED. But plenty of people didn't realize that and took it as if Juice's experiment was a 100 page documentary explaining why ELO hell didn't exist, hence these people are "SIMPLISTIC BUFFOOOOOONS"
Sheeesh, starting this pointless crap out of nowhere.
i never said his experiment showed his intentions, dumbass
i simply said his experiment proved one thing, which was that picking heroes you are good at will bring you to your actual rating
let me quote you here
"that shit was fucking retarded. sure, pick mid everygame using snowball heroes. way to prove nothing. not everybody has that playstyle or wants to play that way and other roles are much more team dependent (carries,supports,offlane).
Everyone who agreed with Juice was a simplistic buffoon."
i disagreed with proving nothing, i disagreed with "that playstyle" - because it simply showed that playing what you are good at will bring you up to your rating. Which renders your following point redundant. Several other players followed up with Supports only experiments, Chen only experiments, and they stomped like 20 games consecutively and got sick of it, because it proved the same thing - playing what you are good at will bring you to your rating, be it mid, offlane or support.
Therefore your blanket statement of "everyone who agreed with juice being a simplistic buffoon" was obviously something I disagreed with. Yes, you can say that he over-extrapolated the results of his experiment, but denying that there was some truth in it? That the data was at least applicable to same proportion of the population? Now that is just stupid.
As for your other points:
Let me assure you, the lineup at most fucking ratings don't matter. Even at 5k+ the lineup seldom matters. The team that seems stronger on paper never has the advantage in the game itself. The only thing which even has any effect on the team's chances is the way farm is distributed. Five fucking carries will obviously disadvantage the team. But the way you even lane your heroes itself has a far bigger impact in pubs than the fucking lineup. Which is why I contested your point about CM and CD. I do actually have 100 odd games of Captain's Mode, http://dotabuff.com/players/68379658/matches?date=&duration=&faction=&game_mode=captains_mode&hero=&lobby_type=®ion= and when solo queueing you are just as likely to get some funky lineup with five carries as you are in ALL PICK. Unless you take captain EVERY SINGLE TIME and make sure you draft a competent lineup. Which this guy didn't. He let his 2k/3k/4k teammates do the drafting. He deliberately put himself at the mercy of his teammates for the purpose of this experiment, which you seemed to have conveniently overlooked.
I never acted like a fucking know-it-all. I pointed out the flaw in your argument and your loaded statements, and you couldn't handle it and blew up at me without bothering to argue in a rational matter. That's your own loss.
Why would I call you hopeless? Instead of trying to come up with a valid point, you extrapolate my argument to outlandish proportions and proceed to beat it down, which is a pitiful way of arguing. If you cannot even recognize that then I'm afraid its pointless trying to hold a logical debate with you.
Your engrish is suffering. "Everyone who agreed with juice being a simplistic buffoon" refers to people who agree with juice on his experiment for its INTENDED goal. Holy fucking hell, you do realize you don't agree with Juice, right? You just got something out of his stupid experiment. HENCE I DON'T SEE WHY YOU FELT THE NEED TO RESPOND TO ME WITH AN INSULT.
I said "way to prove nothing" because once again, holy fucking hell, Juice was abusing snowball heroes on people who was far below him in mechanical skill and game sense. Do you see what I'm getting at. Say he used a non-snowballer, it would have been harder for him to win. A Venomancer is not going to stomp that Faceless Void even with double the farm. His experiment proved NOTHING. I repeat, not everyone has a playstyle of solo mid snowballer. It would have undoubtedly been MUCH MUCH more difficult if he didn't get heroes that could 1v3 and more.
I swear to god, I usually agree with most of the shit you say but LISTEN TO YOURSELF
"LINEUPS DON'T MATTER"
WAY TO DENY THE VERY CONCEPT OF DOTA. ARGHHHHHH. Shit, screw this, counterpicks and lineup strategies don't matter at all buds, welcome to league of legends.
Just because a lousier lineup wins a better one, doesn't fucking mean the lineups don't matter! If I win you in a game of poker with pair of 2s and you lose with a pair of AAs, are you gonna fucking say the cards don't matter??
If I somehow win the game with 1/3 of your CS, are you gonna say last hitting doesn't matter??
If I won a damn game by building a carry support, are you gonna say freaking roles don't matter?!?!?>!?!!?
You can agree with aspects of a person's point, even if you don't agree with the whole thing.
"Way to prove nothing" - he was simply playing the heroes he normally played on his main. The fact that you are still arguing his experiment proved NOTHING is quite pitiful.
Now you're quoting me out of point again. I am not saying that lineup has no effect on the damned game. My point is that in pubs in general, the impact of the lineup is far eclipsed by all the various other aspects of play. This point is only relevant to your suggestion that "Are you seriously saying 2k games have better lineups, coordination and warding whatever shit than 5k games... if you say yes then I cannot argue further since I don't play at 5k."
The thing is that you blew my point out of proportion, which is why I replied as such. That was never my point. My original post in the first place was about how the person who did this experiment picked CM, CD so that he wouldn't be able to pick his hero, and you argued that by picking those modes he skewed the experiments because there would be "better lineups". That was the point I was contesting.
I do apologize for over-arguing about lineups, yes they do have a significant effect on the game but I do believe that it is still far eclipsed by other aspects of play (given the same number of cores on both teams). I was detracting from my point.
Another fucking experiment ? holy fucking shit this is getting tiresome
I do think that sometimes people doing this just wants to attention whore for a bit
Yes if you tryhard and you are a fucking 5k player you will go up in rank and win games, i dont see what you need to prove here.
btw you guys realize that he is literally ruining games right ? If the 3k player is playing like a 5k player the balance of that match is broken.
That its ruining games
I only got angry because out of nowhere and basically nullified my entire opinion while adding a bloody insult to it.
His very experiment was pitiful. He failed to take into account his playstyle was the ONE playstyle that would feel extremely overpowering against people of lesser skill. His experiment was simply like Havoc said or implied, "to show off". The freaking experiment only proved 1 thing that everyone already knew, snowballers can carry games.
Regarding the second point, I don't think it matters why he chose CM/CD because the crux of the situation remains the same. The dude wouldn't be hampered by stuff like no courier or farm being contested because CM/CD games are far less likely to have shit like unwilling supports or 5 carries. You just had one game where you went 0-9 as TB because someone lastpicked Furion and proceeded to be a dickhead right? I imagine these sort of scenarios would occur at much higher rate in AP lower mmr.
However, I did agree his experiment was far better than Juice's. Its just that it isn't really that useful because a majority of people complaining about ELO hell play ap and whine about horrible lineups, no wards and blablabla.
Thats why at the end of my comment, I said it would be more interesting to see a 4.5k player climb up playing all roles in AP from 3k because I'm sincerely curious what his win rate progression would be.
I understand, zenoth, that it must be very frustrating to still see people bitching about being better than everyone, but i dont think people need to still do this kind of experiment, its pointless already, people is NOT going to change their minds.
@ Sampson
So I haven't done a smurf experiment before, but I have a better win rate with supports and offlaners by far than carries. You're telling me if I had a new account, played nothing BUT the characters I have over 55% to 75% win rates on them I would dwindle down back to my original 2k trash mmr?
I think that your logic is flawed sorry bro.
EDIT:
Just to be clear, I'd rather not do a "smurf" account and see my skill rise or just stay the same I'm not worried about it so all you 6k~4k players out there don't worry I won't be a part of the "problem" since everyone screams bitches moans about "smurfs ruin my game blah blah blah" I'll stay in my waste basket of 2k crap thank you.
Reading the notes in the guy's excel file he provided from the experiment is quite useful. If you look at the way he played support in the lower mmr bracket, he basically never played the kind of hard 5 position he might usually play, if he'd picked the same hero in the same line-up in a higher mmr game. He played very greedy and took the carry's farm when the carry wasn't farming efficiently, essentially ensuring every hero he played had the potential to carry the game. This might be somewhere that a lot of players complaining about their mmr and their team mates are going wrong.
^
What Karma said.................. I'd like to see someone do a hard 5 slot support smurf instead of the usual slark bullshit buff smurf.
The support or not support thing is bullshit. He just played carry because he is BETTER than the enemys and if he plays support he puts his fate in his teams hand after the mid game.
For the people who are not smurfing it doesnt fucking matter if you support or carry. If you suck then you suck and it doesnt matter if you miss 30cs on a carry and get poor gpm or if you roam around useless as a support and archieve nothing.
I used to support on 5k mmr pubs and it was no fucking issue. I still won about 55% and if I carried in solo que it was similar.
"if he plays support he puts his fate in his teams hand after the mid game."
Yes, but some people doesnt want to accept this.
Pro psaní komentáře se přihlaste.
http://www.reddit.com/r/DotA2/comments/28zclp/my_mmr_experiment_solo_mmr_2278_party_mmr_2196_to/