PS: I did notice several seasons ago that I was facing a TON of smurfs but not really coming across them anymore or at least I haven't noticed.
This season:
Overall:
Blogs > Papabear2009 |
Papabear2009
6 Posts
PS: I did notice several seasons ago that I was facing a TON of smurfs but not really coming across them anymore or at least I haven't noticed. This season: Overall: | ||
Nakajin
Canada8764 Posts
| ||
WombaT
Northern Ireland20722 Posts
Those numbers don’t look pretty but are the games at least close or are you getting stomped around in the ones you lose? While it would still technically be balancing around a 50% matchmaking about the least satisfying possible way to do that would be half one’s games being crushing defeats and the other half being easy victories. | ||
seriosblack
Bulgaria8 Posts
Imagine if we had the technology for separate match up MMR. Just pity nobody could do it back when they introduced separate MMR per race | ||
ArtyK
France3143 Posts
| ||
Papabear2009
6 Posts
On April 15 2021 21:52 WombaT wrote: How are the actual games going for you? Those numbers don’t look pretty but are the games at least close or are you getting stomped around in the ones you lose? While it would still technically be balancing around a 50% matchmaking about the least satisfying possible way to do that would be half one’s games being crushing defeats and the other half being easy victories. Most of the Protoss I face just turtle hardcore and I'll try to get to like 80 or 90 workers when I see a low gateway count to really start pressuring them, like with a hydra bane timing then with the next remax to include some vipers then the one after whatever tech I need for the next one depending on the ground force. Usually, the hydra bane timing will do some damage, maybe kill a 3rd or deny a 4th and get some army kill. It just seems they will just produce carriers and HT after and makes feel like a brick wall. Terran is the worst, I feel like the Terrans are expanding faster than me and I have to commit so many resources to destroy these planetaries with their sim cities as well as always feel like I'm being dropped, pushed, etc. Like it just feels every single Terran is mechanically better than me. The most sense of comfort or control I do get is when I go muta bane but the timing of control I have does not last for long. | ||
Ben...
Canada3485 Posts
PvZ is so makes no sense right now. I tried to be charitable to the zergs I played against and avoided doing skytoss because it was boring for both of us, but lately I've been running into zergs who are basically forcing me to do so. I played four PvZs in a row yesterday where the zergs basically skipped most upgrades (usually they just get +1 or +2 range attack, no carapace), rushed hive and lurkers (with the range and speed upgrade), made like 10-12 lurkers and a bunch of hydras, then tried to bust one of my expansions at like 10 or 11 minutes. It's quite strong against ground compositions but is countered easily by mass voids and storm. I genuinely don't get why they're playing this way. It's not good. It's like nobody told them that void rays with attack upgrades can often beat hydras with no armour upgrades. If they went for carapace instead it'd likely be much stronger. The other Zerg style I've been seeing a lot of is playing super greedy mass expanding behind static defence with queens then trying to go for mutas, which again is countered by opening two stargate and going for air upgrades and templar archives. I've seen more mutas in the last two weeks than the last entire year before that. They even scout that I have two stargates and a fleet beacon but still the mutas show up usually die to upgraded phoenixes. I've played against some queen walks, but not that many. Those make a lot more sense than either of these other current builds I'm facing though. | ||
WombaT
Northern Ireland20722 Posts
On April 16 2021 04:36 Ben... wrote: This season has been odd for me so far. PvP is dumb as always, though the increasing number of people trying to do battery air turtle builds is getting old quite fast. I'm really bad at PvP so my numbers are never great. PvT is fine, and still where I find I have the most fun. Holding off tank pushes is interesting as is dealing with drops and stuff. There's still a lot of variety in it which keeps things fresh. PvZ is so makes no sense right now. I tried to be charitable to the zergs I played against and avoided doing skytoss because it was boring for both of us, but lately I've been running into zergs who are basically forcing me to do so. I played four PvZs in a row yesterday where the zergs basically skipped most upgrades (usually they just get +1 or +2 range attack, no carapace), rushed hive and lurkers (with the range and speed upgrade), made like 10-12 lurkers and a bunch of hydras, then tried to bust one of my expansions at like 10 or 11 minutes. It's quite strong against ground compositions but is countered easily by mass voids and storm. I genuinely don't get why they're playing this way. It's not good. It's like nobody told them that void rays with attack upgrades can often beat hydras with no armour upgrades. If they went for carapace instead it'd likely be much stronger. The other Zerg style I've been seeing a lot of is playing super greedy mass expanding behind static defence with queens then trying to go for mutas, which again is countered by opening two stargate and going for air upgrades and templar archives. I've seen more mutas in the last two weeks than the last entire year before that. They even scout that I have two stargates and a fleet beacon but still the mutas show up usually die to upgraded phoenixes. I've played against some queen walks, but not that many. Those make a lot more sense than either of these other current builds I'm facing though. That does indeed sound mystifying, especially turtling into mutas given this meta. Those sound like styles for almost meta but this meta. I guess I can’t be too critical, I imagine prior to a really, really high level Zerg is super unforgiving to play when turtle styles are popular. Players probably struggle to make good reads and changes to their style while keeping on top of things mechanically. Perhaps some of the ones you’ve faced are so sick of losing to certain styles they’re going ‘fuck it I’m doing my build’ regardless of what their opponent is doing. I’ve definitely had periods where I was so sick of PvZ I played essentially blindly like that | ||
Cyro
United Kingdom20157 Posts
On April 15 2021 21:52 WombaT wrote: How are the actual games going for you? Those numbers don’t look pretty but are the games at least close or are you getting stomped around in the ones you lose? While it would still technically be balancing around a 50% matchmaking about the least satisfying possible way to do that would be half one’s games being crushing defeats and the other half being easy victories. SC2 does have this issue because it assumes that everybody is equally proficient against all races in the matchmaking process, which isn't true. A lot of the time it's close enough, but often it's not. You can easily get yourself into a feedback loop if your effective skill in one or more of the matchups is even slightly mismatched from the others. Lets say for example that a player is a 4500 MMR overall but this is made up of a ZvP skill of 4000 MMR and a ZvZ skill of 5000 MMR. Their MMR is 500 away from accurately representing their skill in either matchup but it will get closer with games, right? No. When they match against a zerg player they will have a very high chance of winning, say 75%. They'll likely win and thus go up to 4550 MMR; only 450 below their actual skill. MMR is moving, but it's becoming more accurate for ZvZ and less accurate for ZvP at the same time - the overall accuracy is not increasing. This player was already going to be matched against Protoss players 500 MMR better than them, but now it's 550 because they won that last game. The very next game against Protoss they're highly unlikely to win (~20%) and when they lose, their MMR drops back to 4500. Now the MMR is becoming more accurate for ZvP, but less accurate for ZvZ and again the overall accuracy hasn't improved. We're still being matched based on a number which doesn't usefully represent any matchup. The change that we just made directly counters the one that happened in the game before, so we haven't gone anywhere at all even though we had two extremely lopsided games. Maybe this happens again and they drop to 4450. Now they're 550 MMR below their ZvZ skill - even further than before - so when they play a Zerg again they have even better than the 75% odds that they had last time. They win, now they're at 4500 rating. You can play over and over and over and over again, have every game be a 75% chance of ending a particular way due to matchup MMR imbalances, yet not actually go up or down in MMR at all - at the end of playing 50 games, you're still in as big of a hole as before playing them. There's no system in place to correct for this; it just indefinitely amplifies any imbalance in effective skill between the matchups. MMR is stabilized by there being three matchups, rather than two - but it's destabilized by effects like there being different amounts amounts of each race playing on the ladder. The race/s that you hit more often will drag your overall MMR around to follow your skill in that one matchup. When you hit the rarer matchups, you will very consistently find yourself over or under-matched and it will be a stomp one way or the other.. one of the largest factors in people going 60/40 or 70/30 in two matchups for an average 50% win ratio since the birth of the game. It's a statistical disaster that really ruins the matchmaking system so it's quite amazing that it didn't get fixed before sc2 went EOL. It's much more important than having a different "overall MMR" for each race that you're playing as, which did get implemented eventually. For OP, their MMR is stuck far below their ZvZ skill but far above their ZvP and ZvT skill. ZvZ is strongly pulling the MMR up while the other two pull it down and the end result is that it's not going anywhere fast even though every game is a stomp. After 54 games their MMR is still horribly mis-rated - it really is a pretty catastrophic design flaw. | ||
Charoisaur
Germany15616 Posts
On April 16 2021 05:35 Cyro wrote: Show nested quote + On April 15 2021 21:52 WombaT wrote: How are the actual games going for you? Those numbers don’t look pretty but are the games at least close or are you getting stomped around in the ones you lose? While it would still technically be balancing around a 50% matchmaking about the least satisfying possible way to do that would be half one’s games being crushing defeats and the other half being easy victories. SC2 does have this issue because it assumes that everybody is equally proficient against all races in the matchmaking process, which isn't true. A lot of the time it's close enough, but often it's not. You can get yourself into a feedback loop. Lets say for example that a player is a 4500 MMR overall but this is made up of a ZvP skill of 4000 MMR and a ZvZ skill of 5000 MMR. When they match against a zerg player they will have a very high chance of winning, say 75%. They'll likely win and thus go up to 4550 MMR; only 450 below their actual skill. MMR is moving, but it's becoming more accurate for ZvZ and less accurate for ZvP at the same time - the overall accuracy is not increasing. This player was already going to be matched against Protoss players 500 MMR better than them, but now it's 550 because they won that last game. The very next game against Protoss they're highly unlikely to win (~20%) and when they lose, their MMR drops back to 4500. Now the MMR is becoming more accurate for ZvP, but less accurate for ZvZ and again the overall accuracy hasn't improved. We're still being matched based on a number which doesn't usefully represent any matchup. The change that we just made directly counters the one that happened in the game before, so we haven't gone anywhere at all even though we had two extremely lopsided games. Maybe this happens again and they drop to 4450. Now they're 550 MMR below their ZvZ skill - even further than before - so when they play a Zerg again they have even better than the 75% odds that they had last time. They win, now they're at 4500 rating. You can play over and over and over and over again, have every game be a 75% chance of ending a particular way due to matchup MMR imbalances, yet not actually go up or down in MMR at all - at the end of playing 50 games, you're still in as big of a hole as before playing them. There's no system in place to correct for this; it just indefinitely amplifies any imbalance in effective skill between the matchups. It's especially bad because there can be pretty wide disparities in the amounts of each race playing on the ladder; the race/s that you hit more often will drag your overall MMR around to follow your skill in that one matchup. When you hit the rarer matchups, you will very consistently find yourself over or under-matched. It's a statistical disaster that really ruins the matchmaking system so it's quite amazing that it didn't get fixed before sc2 went EOL. It's much more important than having a different "overall MMR" for each race that you're playing as, which did get implemented eventually. I disagree. First of all it's unrealistic that a player is 4000 MMR in one matchup and 5000 MMR in the other. The MMR difference per matchup for the majority of players won't be higher than 200-300 MMR. Would MMR be more accurate if there would be seperate MMR per matchup? Maybe. But at the end of the day MMR won't be 100% accurate anyway because the current MMR of a player also depends a lot on daily form, having a lucky streak etc, which can also lead to a 200-300 MMR difference so I think it doesn't really matter. I think the current MMR system usually indicates the skill level quite well (not 100% accurate but generally accurate enough). | ||
Cyro
United Kingdom20157 Posts
First of all it's unrealistic that a player is 4000 MMR in one matchup and 5000 MMR in the other. The MMR difference per matchup for the majority of players won't be higher than 200-300 MMR. They are example numbers only, the actual starting point or difference between them is irrelevant. It's a hypothetical scenario to be more easily explainable. MMR will indeed never be 100% accurate but that doesn't mean we shouldn't improve it from 70% to 90% accurate if such a change were arbitrarily possible. There are definitely complexities but something like a matchup-MMR-offset which could change over time (positive or negative) in response to a scenario such as OP's would have been able to fix all three matchups in 20 or 30 games, not fix only 2 out of the 3 after 100 games. It would have improved the matchmaking accuracy for everybody in everyday play. | ||
Obamarauder
697 Posts
On April 15 2021 22:05 seriosblack wrote: My numbers aren't that extreme, but i feel your pain. I think my ZvZ is around 3,7k MMR/ ZvP- 3,2 & ZvT- 3,1. Imagine if we had the technology for separate match up MMR. Just pity nobody could do it back when they introduced separate MMR per race I can already imagine the balance whining if separate matchup MMR is introduced. "my TvZ is 5.5k but my TvP is 4.8k!! PROTOSS OP!!!" | ||
BigFan
TLADT24917 Posts
| ||
Ben...
Canada3485 Posts
On April 16 2021 10:07 Obamarauder wrote: Show nested quote + On April 15 2021 22:05 seriosblack wrote: My numbers aren't that extreme, but i feel your pain. I think my ZvZ is around 3,7k MMR/ ZvP- 3,2 & ZvT- 3,1. Imagine if we had the technology for separate match up MMR. Just pity nobody could do it back when they introduced separate MMR per race I can already imagine the balance whining if separate matchup MMR is introduced. "my TvZ is 5.5k but my TvP is 4.8k!! PROTOSS OP!!!" "My MMR in PvP is 300 MMR below my PvT and PvZ so clearly protoss is overpowered in PvP." | ||
Cyro
United Kingdom20157 Posts
On April 17 2021 03:45 Ben... wrote: Show nested quote + On April 16 2021 10:07 Obamarauder wrote: On April 15 2021 22:05 seriosblack wrote: My numbers aren't that extreme, but i feel your pain. I think my ZvZ is around 3,7k MMR/ ZvP- 3,2 & ZvT- 3,1. Imagine if we had the technology for separate match up MMR. Just pity nobody could do it back when they introduced separate MMR per race I can already imagine the balance whining if separate matchup MMR is introduced. "my TvZ is 5.5k but my TvP is 4.8k!! PROTOSS OP!!!" "My MMR in PvP is 300 MMR below my PvT and PvZ so clearly protoss is overpowered in PvP." We get far worse from people who say "My MMR is X, but i have a 70% winrate against terran and 30% against protoss! This must be because protoss is OP". But no, it's just a dumb rating system amplifying their slight strength against terran and weakness against protoss. | ||
WombaT
Northern Ireland20722 Posts
On April 17 2021 04:01 Cyro wrote: Show nested quote + On April 17 2021 03:45 Ben... wrote: On April 16 2021 10:07 Obamarauder wrote: On April 15 2021 22:05 seriosblack wrote: My numbers aren't that extreme, but i feel your pain. I think my ZvZ is around 3,7k MMR/ ZvP- 3,2 & ZvT- 3,1. Imagine if we had the technology for separate match up MMR. Just pity nobody could do it back when they introduced separate MMR per race I can already imagine the balance whining if separate matchup MMR is introduced. "my TvZ is 5.5k but my TvP is 4.8k!! PROTOSS OP!!!" "My MMR in PvP is 300 MMR below my PvT and PvZ so clearly protoss is overpowered in PvP." We get far worse from people who say "My MMR is X, but i have a 70% winrate against terran and 30% against protoss! This must be because protoss is OP". But no, it's just a dumb rating system amplifying their slight strength against terran and weakness against protoss. People seem to forget matchups are really quite different too. Not sure why. Although not just from this thread it does seem matchmaking is generally a bit wonky. I have always, always been considerably better at PvT than PvZ as I like to play safely and deflect and eke out insurmountable advantages in long games of attrition with non-committal harassment around the edges. Just how I like to play really. Up until a certain level of play that’s always going to be a style of play that is rather good against Terrans, a ‘you take the initiative and I’ll try to chip away in a long game’ is just an absolutely atrocious way to play against Zerg, outside of a few metas. Playing T where being aggressive was kind of the name of the game (and fun too) I was roughly equally as good vP as vZ, maybe slightly more the latter because being the aggressor is kind of the name of the game. I mean within reason of course, I think if a game has sufficiently varied races it’s a healthy enough sign that matchup strength variance is ok | ||
alpenrahm
Germany628 Posts
On April 16 2021 05:35 Cyro wrote: Show nested quote + On April 15 2021 21:52 WombaT wrote: How are the actual games going for you? Those numbers don’t look pretty but are the games at least close or are you getting stomped around in the ones you lose? While it would still technically be balancing around a 50% matchmaking about the least satisfying possible way to do that would be half one’s games being crushing defeats and the other half being easy victories. SC2 does have this issue because it assumes that everybody is equally proficient against all races in the matchmaking process, which isn't true. A lot of the time it's close enough, but often it's not. You can easily get yourself into a feedback loop if your effective skill in one or more of the matchups is even slightly mismatched from the others. Lets say for example that a player is a 4500 MMR overall but this is made up of a ZvP skill of 4000 MMR and a ZvZ skill of 5000 MMR. Their MMR is 500 away from accurately representing their skill in either matchup but it will get closer with games, right? No. When they match against a zerg player they will have a very high chance of winning, say 75%. They'll likely win and thus go up to 4550 MMR; only 450 below their actual skill. MMR is moving, but it's becoming more accurate for ZvZ and less accurate for ZvP at the same time - the overall accuracy is not increasing. This player was already going to be matched against Protoss players 500 MMR better than them, but now it's 550 because they won that last game. The very next game against Protoss they're highly unlikely to win (~20%) and when they lose, their MMR drops back to 4500. Now the MMR is becoming more accurate for ZvP, but less accurate for ZvZ and again the overall accuracy hasn't improved. We're still being matched based on a number which doesn't usefully represent any matchup. The change that we just made directly counters the one that happened in the game before, so we haven't gone anywhere at all even though we had two extremely lopsided games. Maybe this happens again and they drop to 4450. Now they're 550 MMR below their ZvZ skill - even further than before - so when they play a Zerg again they have even better than the 75% odds that they had last time. They win, now they're at 4500 rating. You can play over and over and over and over again, have every game be a 75% chance of ending a particular way due to matchup MMR imbalances, yet not actually go up or down in MMR at all - at the end of playing 50 games, you're still in as big of a hole as before playing them. There's no system in place to correct for this; it just indefinitely amplifies any imbalance in effective skill between the matchups. MMR is stabilized by there being three matchups, rather than two - but it's destabilized by effects like there being different amounts amounts of each race playing on the ladder. The race/s that you hit more often will drag your overall MMR around to follow your skill in that one matchup. When you hit the rarer matchups, you will very consistently find yourself over or under-matched and it will be a stomp one way or the other.. one of the largest factors in people going 60/40 or 70/30 in two matchups for an average 50% win ratio since the birth of the game. It's a statistical disaster that really ruins the matchmaking system so it's quite amazing that it didn't get fixed before sc2 went EOL. It's much more important than having a different "overall MMR" for each race that you're playing as, which did get implemented eventually. For OP, their MMR is stuck far below their ZvZ skill but far above their ZvP and ZvT skill. ZvZ is strongly pulling the MMR up while the other two pull it down and the end result is that it's not going anywhere fast even though every game is a stomp. After 54 games their MMR is still horribly mis-rated - it really is a pretty catastrophic design flaw. I'm not entirely sure on the numbers of the Sc2 MMR system but in Chess its 400 points of difference correspond to a one in a hundred chance of winning. And as the overall point distribution seems quite similar I assume this applies to sc2 as well. So your example is grossly misusing the numbers. I think for most people the difference in matchup skill is about 20 points of MMr to maybe a hundred. But certainly not 500 because that would statistically mean that they win like 1 out of a 1000 games. | ||
Cyro
United Kingdom20157 Posts
I'm not entirely sure on the numbers of the Sc2 MMR system but in Chess its 400 points of difference correspond to a one in a hundred chance of winning. It's much flatter than that in SC2 and i've definitely been at least a few hundred MMR apart in different matchups, but it doesn't ultimately matter because it's an abstract explaination. If i were to use very specific numbers from SC2 then i would have to understand the SC2 system exactly to avoid making a mistake, which i don't and i'm not even sure that i can do with public information. Copy/paste it into notepad and change the number if you want, it doesn't change the explanation. | ||
| ||
Next event in 2h 34m
[ Submit Event ] |
StarCraft 2 StarCraft: Brood War Dota 2 League of Legends Counter-Strike Super Smash Bros Other Games Organizations Other Games StarCraft 2 StarCraft: Brood War StarCraft 2 StarCraft: Brood War
StarCraft 2 • Berry_CruncH131 StarCraft: Brood War• practicex 22 • Gussbus • LaughNgamez Trovo • Poblha • aXEnki • Migwel • intothetv • Laughngamez YouTube • Kozan • IndyKCrew League of Legends Other Games |
Replay Cast
ByuN vs GuMiho
TBD vs Rogue
TY vs DongRaeGu
TBD vs Bunny
TBD vs SHIN
TBD vs Classic
ESL Pro Tour
OSC
ESL Pro Tour
PassionCraft
ESL Pro Tour
World Team League
ESL Pro Tour
Korean StarCraft League
Afreeca Starleague
hero vs Soulkey
[ Show More ] AfreecaTV Pro Series
Reynor vs Cure
ESL Pro Tour
World Team League
ESL Pro Tour
BSL
Zhanhun vs DragOn
Dewalt vs Sziky
CSO Cup
Replay Cast
Sparkling Tuna Cup
ESL Pro Tour
World Team League
ESL Pro Tour
BSL
Gypsy vs Bonyth
Mihu vs XiaoShuai
ESL Open Cup
ESL Open Cup
ESL Open Cup
|
|