DOTA 2
- Thread starter Bels_sl
- Start date
Dandain
Trakanon Raider
- 2,092
- 917
The International - Dota 2 Championships
TI invites - honestly I can't complain about the guaranteed list I think its accurate.
Europe:
Na'Vi
Alliance
Fnatic
Empire
China:
DK
IG
VG
Newbee
SEA:
Titan
NA:
EG
Cloud9
I think its pretty sweet they immortalize the 5 man roster of the closeout game at TI in the next years posters and such. Wisp, Prophet, CK, CM, Puck.
TI invites - honestly I can't complain about the guaranteed list I think its accurate.
Europe:
Na'Vi
Alliance
Fnatic
Empire
China:
DK
IG
VG
Newbee
SEA:
Titan
NA:
EG
Cloud9
I think its pretty sweet they immortalize the 5 man roster of the closeout game at TI in the next years posters and such. Wisp, Prophet, CK, CM, Puck.
- 11,058
- 2,262
All the good NA players are playing lol, where they perpetuate the starcraft tradition of losing to mid tier koreans. But yeah the NA qualifier looks pretty awful, even the SEA qualifier looks more appealing even though it's probably not going to be very high level dota, there's some big names, some big teams. NA has like Team Liquid and then a bunch of pubs and a few pros mixed in all without sponsors or a proper team. And even TL is in a pretty fucking shitty state.
Can't wait for some 6.81 matches though.
Can't wait for some 6.81 matches though.
What region is Orange?
Also I just reinstalled dota2 (9GB) and I still can't move items in my invetory/stash. It's brutal. played lone druid and I just wanted to alt+f4.
Also I just reinstalled dota2 (9GB) and I still can't move items in my invetory/stash. It's brutal. played lone druid and I just wanted to alt+f4.
Dandain
Trakanon Raider
- 2,092
- 917
This is unique to your experience, I have had zero bugs like this. Have you tried uninstalling steam or made sure all key bindings are reset (since they are stored in cloud).
Here's how you know dota2 matchmaking is pretty borked.... durnig a recent SingSing stream while he was partied with one of his random friends the algorithm decided to add two other C9 players (EternalEnvy, Bone7) and Grizine (pro on a t2/t3 team). Game didn't last long didn't last long despite pretty bad randoms.
Dandain
Trakanon Raider
- 2,092
- 917
Unless you're around 4500 solo ranking, the matchmaking you see on pro streams is absolutely nothing like it is for the rest of the dota population. And if you're not 5k+ you won't see disparate MMR's near as often or as wide. Its pretty simple to understand what the system tries to do and I am not certain there is a better solution than what they have. The big problem at the top end is that there just isn't that many players above 5k. We're talking maybe a few thousand per region? Maybe a little more, But hardly any when you compare it to the rest of the MMR universe. And keep in mind all the combinations of group size. Solo queue can end up in games with duo or triples or all other solos, Duo/Triple End up with solo players to fill teams. In unranked 4 Man stacks and all their combinations, and lastly 5 man ranked with disparate MMR are hard to matchmake for.
If I queue with a 3k rated player, I'm basically 4500. So 1500 difference + the group queue MMR boost. The system assumes we will perform as if we were two 3800 players. So the matchmaker will create a game where the average MMR is ~3800. The breakdown can happen in multiple ways. At 3800 MMR there is enough player saturation that the game will actually launch with a very low MMR spread for the solos.
Here is an example where the duo queue in the game is 5300-4600. Notice how the 3800 average game is filled with 3800 rated solo players, while this game is not. That's just because of the limited player pool that qualifies to be slotted into the game and there isn't enough players at 5k average to fill the game with a small spread set of solos. This problem only gets worse at higher ratings, but there isn't too much valve can do about it because its not an algorithm problem its a wait time + lack of player density at the extreme of skill.
If I queue with a 3k rated player, I'm basically 4500. So 1500 difference + the group queue MMR boost. The system assumes we will perform as if we were two 3800 players. So the matchmaker will create a game where the average MMR is ~3800. The breakdown can happen in multiple ways. At 3800 MMR there is enough player saturation that the game will actually launch with a very low MMR spread for the solos.
Here is an example where the duo queue in the game is 5300-4600. Notice how the 3800 average game is filled with 3800 rated solo players, while this game is not. That's just because of the limited player pool that qualifies to be slotted into the game and there isn't enough players at 5k average to fill the game with a small spread set of solos. This problem only gets worse at higher ratings, but there isn't too much valve can do about it because its not an algorithm problem its a wait time + lack of player density at the extreme of skill.
So you're saying you see no problems with a system that puts 4 pro players on one team and 5-random players on the other?
Yes, there is no problem. I'm sure the 5-random people had similar MMR as sing and EE. Is valve supposed to flag people who are teammates? Are you retarded?
If theres pros in a game you have to go through 20 dodges before you get a game also so the range gets wider and wider
... the point is MMR is broken. The fact that they're on the same team is immaterial (though dota2 does in fact have a teams....), if you grab a handful of pro players from T1 teams they'll dominate barring a try-hard stack with a hero advantage. This should tell you that the method used to determine MMR is not particularly effective.
Watch any of the streams, they'll nearly always win their lane despite hero mismatches and frequent gank attempts. (Everyone wants to claim they shutdown soandso)
Dandain
Trakanon Raider
- 2,092
- 917
You are arguing an extreme edge case of the matchmaking system. Any MMR or ELO or whatever system depends on having a shitload of players to make it work correctly. As people approach the top end of skill the amount of potential matches shrinks. At 6am EST occasionally the matchmaker will put me in a game with 6000 rated solos. But each team gets one, and people would rather play than wait all night for a game to start. This only affects the top .5% of players much, and they've been dealing with it from 4500+ - its just how the MMR system works when there becomes an insufficient number of players. You're also talking about the players that have substantially better than 50% winrates because the system cannot put them into enough games that they lose. Part of Dota is that some games will be hardcore stomps, that happens at all skill brackets, including pro tournament games.
The system works, people can improve and climb the ladder in all brackets, no one person gets more shit games than any other. The matchmaker is equal opportunity. And most of us win 50% of our games give or take a couple %. That's the goal of the system. It can't account for those few players who have outlier winrates.
http://dotabuff.com/players/verified
Look through the list, notice that every Dota personality has roughly 50% winrate (bruno/2gd/cyborgmatt). Draskyl is an outlier for a caster - he's a 5k rated player - his winrate is 54%. Godz (53%)
Then look at the pros - anywhere from high 57%'s to 73% winrate. They bust the system, its not the system that's broken.