so after 4 wins my MMR went from 3968 to 4042, pretty solid i think, guys expect they will get +100 MMR after one win, thats just plain stupid :D
what I like most is the quality of games, rly good games, a lot better then when there was no visible MMR, no shit talking, stupid plays etc, now I feel like some players I play with are even better then me, that in most cases never happened before, maybe I just got lucky during calibration stage...
@679
When game was still balanced, I was like 1-5-3 something. Starting from that point, game was pretty much already won. If system would count only the end game stats, then fountain campers would have the highest MMR.
also I think that everyones MMR will go down alot as time passes and other players join ranked matches. I dont think many people started playing ranked matches yet (noone in my friend list at least). so I think dendi pwns some nabs and then MMR will settle :D
so it seems pretty clear that ones normal match making rating is where you will start out when going into ranked..
Doesn't it then make sense to try and do as well as possible in normal MM for awhile before going into ranked? maybe try and play with some good players in normal for a few games and win, thus pushing your rating up?
^ It should be noted that it's mostly very good players who post here, or at least ones who strive to be good, and not average Dota players who might play it casually.
There is a difference between a general dota population average and an average of dotabuff posters who OPTED to post their scores WITHOUT any guarantee of honesty.
+Dotabuff is a smaller sample of players than the general player base, and it could be argued they tend to be better.
+People will be compelled to post their scores if they are good and not post them if they are bad - another skew in the average.
+People will be compelled to lie, because they value their ego.
All 3 of these aspects (and probably others) make the average of the group of dotabuff members willing to share (without checks for honesty) way higher than that of the general population.
Also, as far as I can see, no one has a score of 6000/6500 (which just puts a nail in the coffin of your analysis). The professionals are reporting ratings of about 5000-5500.
heard somewhere that top 1% is around 70k players (mby 7k anyways....)
at ti3 there were 16 teams a 5 players -> 90 pro players+ lets say a couple more that are good, roughly 150 pro players
150/7k and u have a new distribution ranging from 4100 to infinity xd
thats why u find all the pros and stacking bitches at 5-5,5k mmr
well now i can play seriously (since mmr is visible and i know what type of players i meet) so ill steadily reach 5k :)
You might be right. I was just surprised to see so many getting high ranks only in this topic. Nothing posted about the "average" . Maybe people don't want to, but rating don't matter. = I mean, now you got your rating, then you can try to improve from there.
What rating is very high? Before ratings when I checked replays I almost always in very high, and my MMR is not good (according to this topic anyway) 3900.
What I am interested in knowing are the MMR ratings of supports that didn't go in 4-5 stacks, I went 7-3 in my 10 rank games, played all 10 as a party of two. My MMR is only around 3400. Yet I did quite well (from a support point of view) in all 10.
I got my wards, courier, Mek the lot, yet given the nature of being a Hard support (and often sole support) my gpm and xpm is quite low compared to the Hard carries.
i started with 4513 MMr, lost a game STOMPED as oflane timbersaw, when i saw my rating after that game it was WOOPING 4670 ( got more than 150 elos for LOOSING), i played well that game, and the system somehow managed to see that. my second game i won, didnt play so well, went mid as invoker but my teammates were good and we won. my ELO went up to 4711, 41 points for winning. played only 2 games, so far i liked the new sistem
my invoker game i was the person on my team with the most farm, but i didnt play very well and somehow the sistem saw that to and gave me only 40 elos . cant wait to see how many points will i have after the 10 games
I have a rating of Party MMR of 3900 would that be considered as Decent ? Or is my MMR "on Par" with most of the community, if i could put it that way.
Many on dotabuff have extremely high scores... or are at least saying they do. But that is not surprising if you actually watch ordinary pub games. People who you might think are "bad" are actually far far better than your average pub dota player. If you want to compare to more serious players consider the distribution for tMMR for teams active in the last few months.
tMMR (would correspond closely to ranked party MMR)
Its a matter of perspective. If you saw what the true "average" pub player (2250 MMR) did you would be shocked. Millions of people never get past a level you might consider terrible... but they really can't. Its because they are physically unable to perceive and process the game and react at the speed required to play at a high level. Millions of pub dota players are not stupid, they are merely slower.
The fastest movements, once practiced into muscle memory are not even controlled by your brain. Just like with the reaction to touching a hot surface, your hand will pull back instantly and inform your brain afterwards of what happened. People tend to forget that all decision making is not in the brain. If you have faster nerves you are just at a permanent advantage over ordinary people in dota 2. Consider yourself superhuman. It will feel good.
2 losses team rank - Good ES game + horribly bad alch game - 4.1k
1 solo loss - good ES game - 4k even
Dunno how my solo isn't higher than team considering how bad my 2nd team rank game was. I played pretty poorly after playing so much wraith night, see where it moves over the weekend.
Only played 2 games so far, 1w 1l, alot of bad players in both.
3045/770
3067/728
There is no way I playing with the top 10~12%, that or holy shit people are bad at Dota. This new ranked system should have had a revamped MMR instead of the same garbage based on the same garbage for years.
I just lost 20 for winning...
1st game - win- 3350-uncertainty 682
2nd game - win - 3330 - uncertainty 652
i was roaming 2nd game so i guess my individual metrics were bad, did land 52secs of stun not sure that is good or bad but it seems you have to play selfish if you want rating points?
On the dota 2 blog about how they do matching it is explained what really matters for MMR points besides actually winning or losing the game.
They say "Measuring success in matchmaking is difficult"... "We used a statistical tool known as logistic regression, which essentially works by trying to create a function that predicts the odds of victory. This function contains several coefficients which determine the MMR bonus given to players in a party. Then we use numerical techniques to solve for the coefficients that produce the function which is most accurately able to predict the match outcome."
Balance of the match depends on the teams net gold as shown on the graph.
"metric measures balance, based on the difference in gold farmed. To be more precise, it’s the time integral of the gold difference, measured since the last point in the game where the difference was zero"
This (and other mysterious coefficients) is how they decide if you lived up to your MMR and deserve more points. If your team MMR avg was higher, you are expected to win by a larger amount. Armed with this information some people may decide they want to win by the largest amounts possible to get higher MMRs... others may actually want to win close games (ie almost throw) so their MMR goes down after winning making future games easier.
Or you could just play and enjoy the game. Most people won't bother trying to manipulate MMR, but don't be surprised if you encounter some who do.
It depend how long it takes. Because it is an integral a spike in gold at the very end won't do much. Being way ahead the entire game is worth far more. A graph that looks like a diagonal line up across the whole game... that will be worth the max MMR.
I have 3279/567. It seems that a wide range of players are shuffled together. I played mid vs Pudge who got bottle first and spent like all time running for runes across the map. A flashback of my first matches...
i just played 2 terrible games after 7 solo rmm win streak. in the first one a fucking peruvian trashcan stole my midas and didnt gave it back to me while yelling at me through the mic (only because i said well played after he got the courier killed with his bottle inside it). the other one http://dotabuff.com/matches/423663496 my team fed soooo hard that i actually won 9 pts after it.
#forced50%
Valve stated that average MMR for regular pub games is 2250. The average for people who play team games according to dotabuff is 3020. It has been slightly over 3k for the entire year. Probably the average for ranked MMR is closer to the 3k number.
your MMR definitely seems to be tied more to individual performance.
my 2nd game i got -20 for roaming and winning the game with bad stats
3rd game got +40 for playing mid and losing with good stats.
not entirely sure though, 3rd game had way better opponents but even than losing points for winning seems to encourage playing selfishly rather than for the team
Has anyone noticed that it takes a really long time to find a match? It's taking me 3-5 mins where before in non ranked it took 1-2. And that search range bar is almost fully expanded.
I think this may be a big reason people are skeptical of rankings because they're being matched with players worse than them or against better players than them or maybe both in a match.
so after 4 wins my MMR went from 3968 to 4042, pretty solid i think, guys expect they will get +100 MMR after one win, thats just plain stupid :D
what I like most is the quality of games, rly good games, a lot better then when there was no visible MMR, no shit talking, stupid plays etc, now I feel like some players I play with are even better then me, that in most cases never happened before, maybe I just got lucky during calibration stage...
u didint went 100 mmr in 1 game there is 10 calibration games
only after u play ur 10th game u see ur true mmr number and then it decreases and increases as u play on
@679
When game was still balanced, I was like 1-5-3 something. Starting from that point, game was pretty much already won. If system would count only the end game stats, then fountain campers would have the highest MMR.
also I think that everyones MMR will go down alot as time passes and other players join ranked matches. I dont think many people started playing ranked matches yet (noone in my friend list at least). so I think dendi pwns some nabs and then MMR will settle :D
omg dendi omg
It seems you get ~20 MMR for a win. At least I got +20 MMR for my last 2 wins.
mmr can raise a lot in one game too, my mmr increased 150 in one game :P
so it seems pretty clear that ones normal match making rating is where you will start out when going into ranked..
Doesn't it then make sense to try and do as well as possible in normal MM for awhile before going into ranked? maybe try and play with some good players in normal for a few games and win, thus pushing your rating up?
stronK charizard stron[K]
i got 3512. where do i stand ?
^in heaven
5% 1100
10% 1500
25% 2000
50% 2250
75% 2731
90% 3200
95% 3900
99% 4100
5337 points with 30% on solo mm
5% 1100
10% 1500
25% 2000
50% 2250
75% 2731
90% 3200
95% 3900
99% 4100
I really really doubht this. :) According to this topic average is 4.000(4500)
I think they re-calculate those ratings and we will see what average is. I am pretty sure "99%" will be like +6000/6500. :)
Don't tell me that so many in this topic alone is "1%" that is just nahh . :)
^ It should be noted that it's mostly very good players who post here, or at least ones who strive to be good, and not average Dota players who might play it casually.
casuals r fags
@vince masuka
There is a difference between a general dota population average and an average of dotabuff posters who OPTED to post their scores WITHOUT any guarantee of honesty.
+Dotabuff is a smaller sample of players than the general player base, and it could be argued they tend to be better.
+People will be compelled to post their scores if they are good and not post them if they are bad - another skew in the average.
+People will be compelled to lie, because they value their ego.
All 3 of these aspects (and probably others) make the average of the group of dotabuff members willing to share (without checks for honesty) way higher than that of the general population.
Also, as far as I can see, no one has a score of 6000/6500 (which just puts a nail in the coffin of your analysis). The professionals are reporting ratings of about 5000-5500.
heard somewhere that top 1% is around 70k players (mby 7k anyways....)
at ti3 there were 16 teams a 5 players -> 90 pro players+ lets say a couple more that are good, roughly 150 pro players
150/7k and u have a new distribution ranging from 4100 to infinity xd
thats why u find all the pros and stacking bitches at 5-5,5k mmr
well now i can play seriously (since mmr is visible and i know what type of players i meet) so ill steadily reach 5k :)
Vandal:
You might be right. I was just surprised to see so many getting high ranks only in this topic. Nothing posted about the "average" . Maybe people don't want to, but rating don't matter. = I mean, now you got your rating, then you can try to improve from there.
What rating is very high? Before ratings when I checked replays I almost always in very high, and my MMR is not good (according to this topic anyway) 3900.
What I am interested in knowing are the MMR ratings of supports that didn't go in 4-5 stacks, I went 7-3 in my 10 rank games, played all 10 as a party of two. My MMR is only around 3400. Yet I did quite well (from a support point of view) in all 10.
I got my wards, courier, Mek the lot, yet given the nature of being a Hard support (and often sole support) my gpm and xpm is quite low compared to the Hard carries.
lel. 5200 now. slowly climbing.
"(from a support point of view)"
STOP BEING SO FUCKING STUPID!
i started with 4513 MMr, lost a game STOMPED as oflane timbersaw, when i saw my rating after that game it was WOOPING 4670 ( got more than 150 elos for LOOSING), i played well that game, and the system somehow managed to see that. my second game i won, didnt play so well, went mid as invoker but my teammates were good and we won. my ELO went up to 4711, 41 points for winning. played only 2 games, so far i liked the new sistem
my invoker game i was the person on my team with the most farm, but i didnt play very well and somehow the sistem saw that to and gave me only 40 elos . cant wait to see how many points will i have after the 10 games
I have a rating of Party MMR of 3900 would that be considered as Decent ? Or is my MMR "on Par" with most of the community, if i could put it that way.
Valve already announced the distribution of pub MMRs and it appears you start at that level for the ranked games.
5% 1100
10% 1500
25% 2000
50% 2250
75% 2731
90% 3200
95% 3900
99% 4100
Many on dotabuff have extremely high scores... or are at least saying they do. But that is not surprising if you actually watch ordinary pub games. People who you might think are "bad" are actually far far better than your average pub dota player. If you want to compare to more serious players consider the distribution for tMMR for teams active in the last few months.
tMMR (would correspond closely to ranked party MMR)
5% 1980
10% 2200
25% 2560
50% 3020
75% 3495
90% 3990
95% 4270
99% 4485
99.5% 4865
99.9% 5100
^^^^^^^^^^^^^^^^^
That is most likely closer to the group of players actually playing ranked matching games.
im on 5200 and i dont feel like playing a "good games" .... not even first page. I think 5500-5600 is decent rating.
Its a matter of perspective. If you saw what the true "average" pub player (2250 MMR) did you would be shocked. Millions of people never get past a level you might consider terrible... but they really can't. Its because they are physically unable to perceive and process the game and react at the speed required to play at a high level. Millions of pub dota players are not stupid, they are merely slower.
The fastest movements, once practiced into muscle memory are not even controlled by your brain. Just like with the reaction to touching a hot surface, your hand will pull back instantly and inform your brain afterwards of what happened. People tend to forget that all decision making is not in the brain. If you have faster nerves you are just at a permanent advantage over ordinary people in dota 2. Consider yourself superhuman. It will feel good.
Also here is something to watch for fun.
https://www.youtube.com/watch?v=dGhbqAEJVKs
https://www.youtube.com/results?search_query=superhuman&sm=3
2 losses team rank - Good ES game + horribly bad alch game - 4.1k
1 solo loss - good ES game - 4k even
Dunno how my solo isn't higher than team considering how bad my 2nd team rank game was. I played pretty poorly after playing so much wraith night, see where it moves over the weekend.
Oh god i never played this bad in my life, i literally lost the game for my team.
http://dotabuff.com/matches/423575816
Worst OD game by far.
my mmr droped a bit, but it should have droped at least 100 after this game.
I am fucking done with dota, gonna try lol its probably easier.
Only played 2 games so far, 1w 1l, alot of bad players in both.
3045/770
3067/728
There is no way I playing with the top 10~12%, that or holy shit people are bad at Dota. This new ranked system should have had a revamped MMR instead of the same garbage based on the same garbage for years.
and yet... really you are top 10% of pubs. Or you are top 50% of serious players whichever what you want to look at it.
https://www.youtube.com/watch?v=j-DHZG5u0ak
^^^^^^^^^
actual average pub players.
I just lost 20 for winning...
1st game - win- 3350-uncertainty 682
2nd game - win - 3330 - uncertainty 652
i was roaming 2nd game so i guess my individual metrics were bad, did land 52secs of stun not sure that is good or bad but it seems you have to play selfish if you want rating points?
On the dota 2 blog about how they do matching it is explained what really matters for MMR points besides actually winning or losing the game.
They say "Measuring success in matchmaking is difficult"... "We used a statistical tool known as logistic regression, which essentially works by trying to create a function that predicts the odds of victory. This function contains several coefficients which determine the MMR bonus given to players in a party. Then we use numerical techniques to solve for the coefficients that produce the function which is most accurately able to predict the match outcome."
Balance of the match depends on the teams net gold as shown on the graph.
"metric measures balance, based on the difference in gold farmed. To be more precise, it’s the time integral of the gold difference, measured since the last point in the game where the difference was zero"
This (and other mysterious coefficients) is how they decide if you lived up to your MMR and deserve more points. If your team MMR avg was higher, you are expected to win by a larger amount. Armed with this information some people may decide they want to win by the largest amounts possible to get higher MMRs... others may actually want to win close games (ie almost throw) so their MMR goes down after winning making future games easier.
Or you could just play and enjoy the game. Most people won't bother trying to manipulate MMR, but don't be surprised if you encounter some who do.
so ur saying if we fountain camp oppos we win more mmr ?
It depend how long it takes. Because it is an integral a spike in gold at the very end won't do much. Being way ahead the entire game is worth far more. A graph that looks like a diagonal line up across the whole game... that will be worth the max MMR.
okei
I have 3279/567. It seems that a wide range of players are shuffled together. I played mid vs Pudge who got bottle first and spent like all time running for runes across the map. A flashback of my first matches...
I have played 5 party games, won all 5. 1st game I had something like 4520, now I have 4620, so since the first game almost no change
Well, I guess there are throwers in ranked too!
http://dotabuff.com/matches/423382168
"Jungle" pudge, lvl 2 at min 10.
i just played 2 terrible games after 7 solo rmm win streak. in the first one a fucking peruvian trashcan stole my midas and didnt gave it back to me while yelling at me through the mic (only because i said well played after he got the courier killed with his bottle inside it). the other one http://dotabuff.com/matches/423663496 my team fed soooo hard that i actually won 9 pts after it.
#forced50%
thx for sharing ur cool story with us bro
well this is kinda the point of this thread
so in conclusion what's the average MMR? i started at ~3800 and slowly rised to 4011 currently
^2500-3000 i think
I mean there are still people who get stomped by bara/drow picking
I'm stuck at 5k. Can't go up, can't go down.
Valve stated that average MMR for regular pub games is 2250. The average for people who play team games according to dotabuff is 3020. It has been slightly over 3k for the entire year. Probably the average for ranked MMR is closer to the 3k number.
your MMR definitely seems to be tied more to individual performance.
my 2nd game i got -20 for roaming and winning the game with bad stats
3rd game got +40 for playing mid and losing with good stats.
not entirely sure though, 3rd game had way better opponents but even than losing points for winning seems to encourage playing selfishly rather than for the team
Has anyone noticed that it takes a really long time to find a match? It's taking me 3-5 mins where before in non ranked it took 1-2. And that search range bar is almost fully expanded.
I think this may be a big reason people are skeptical of rankings because they're being matched with players worse than them or against better players than them or maybe both in a match.