| Forum Index > SC2 General |
|
DarK[A]
United States217 Posts
On March 31 2012 10:59 Apocalypse114 wrote: how do you caounter mass muta with toss? tried stalker and archon and fenix i always lose Just make more of those units. When you say mass muta what kind of numbers are you talking? When I make a muta ball I stay the fuck away from blink stalkers, stalkers with archons, and large groups of phoenixes, so yeah - I'd say just make more. | ||
|
ttfdrevil
United States16 Posts
All the other leagues are significantly underrepresented (compared to the supposed 20-20-20-20% slit for plat-bronze). So you might be able to infer a few things from this: 1) players are overall better and it is now harder to increase mmr to the artificial benchmarks blizzard set. 2) bronze is full of smurfs from higher leagues 3) overall, higher level players are dropping off faster than lower level players I prefer to think its a combination of 1 and 2, which is why I think it does tend to be harder to get out of bronze. Second, as for those of us who play regularly, but can't quite dedicate the time to play enough to cement skills, I think what I've learned is this game depends on lots of little things that add up to improved play. I have 750+ games, of which about 200 are 1v1 (since season 1). I play a lot of team games with friends, and only play 1v1 when they aren't available. What I've found is that the team games have a different dynamic that lead me to get lazy with the 1v1 skills. When I go bak to 1v1 after taking a week or more off, I generally get clobbered until I can get those skills back up to speed. So, maybe I just suck. My apm hardly ever gets over 50. But I have just learned that I don't have time to play both team and 1v1 games, and would rathe play with friends. Maybe someday I'll get out of silver 1v1, but right now I'm looking forward to season 7 silver... | ||
|
Lysenko
Iceland2128 Posts
On April 01 2012 06:20 ttfdrevil wrote: All the other leagues are significantly underrepresented (compared to the supposed 20-20-20-20% slit for plat-bronze). So you might be able to infer a few things from this: 1) players are overall better and it is now harder to increase mmr to the artificial benchmarks blizzard set. 2) bronze is full of smurfs from higher leagues 3) overall, higher level players are dropping off faster than lower level players Those statistics show none of these things. Here's what's really going on. The MMR system is zero sum, meaning that when a player drops in MMR, his opponent increases in MMR by the same amount. (Note that this is NOT the same as saying that each player gets the same number of displayed points after the match, because the system awards extra points to players whose point scores are below where they should be based on MMR, and vice versa.) This means that active players push each other's rankings up and down until their MMRs are correlated with their likelihood of winning a particular matchup. What makes bronze different is that the activity of bronze players is (on average) a lot less than the other leagues. An inactive player who places into bronze in their placements and stops playing isn't really participating in the MMR distribution process that would push more active bronze players into higher leagues by giving them wins. So, you end up with an excess in bronze league, essentially representing the players who aren't playing enough to establish upward pressure on the more active bronze players who might otherwise be silver. Note that Blizzard's MMR boundaries aren't "artificial," because as long as the shape of the MMR distribution stays the same, those boundaries should correspond to the same percentiles. The only reason to tweak the boundaries are if the shape of the MMR distribution changes, and that the higher leagues are roughly equal in size suggests that it's not changing that fast. Edit: So in response to each point, (1) MMR doesn't say anything about the quality of play, only likelihood of winning. (2) Someone who tanks their rating to wind up in bronze ends up raising other people's MMR by the same amount, yielding promotions elsewhere, so smurfs can't explain an excess of people in bronze. (3) It's precisely because bronze players place and drop out that there are an excess of them. If bronze players were more active across the board, the percentage would be very close to 20% and the higher leagues would have more people. | ||
|
minilance
Canada500 Posts
I was a mid diamond player and i downed my mmr to play in lower league, i have a 60% win ratio by walling my main with 2 nexus, building 20 photon canon and then doing a 4 gate.... | ||
|
Lysenko
Iceland2128 Posts
On April 01 2012 06:40 minilance wrote: I was a mid diamond player and i downed my mmr to play in lower league, i have a 60% win ratio by walling my main with 2 nexus, building 20 photon canon and then doing a 4 gate.... Hard to imagine you'd get very far with a 4-gate at 10 minutes or later in gold or plat, but I'll take your word for it. | ||
|
ckunkel1
United States181 Posts
| ||
|
speedything
18 Posts
On April 01 2012 06:40 minilance wrote: people in plat and under are still shit, I was a mid diamond player and i downed my mmr to play in lower league, i have a 60% win ratio by walling my main with 2 nexus, building 20 photon canon and then doing a 4 gate.... Mmmm... so 6/10 Platinum players lose to your 3950 mineral delayed 4-gate? I was going to call bullshit on this but then when I saw you had reached the lofty ranks of mid-diamond I realised there was no need to doubt such an obvious master of the game. | ||
|
ttfdrevil
United States16 Posts
I'm not sure I fully agree with you. I think my point is that the distribution has changed, but blizzard has opted not to Adjust the bars. Also, i'm not sure mmr is a zero sum game when it is manipulated by people with second accounts or by first account smurfs. At minimum it must impact mmr "confidence" for real bronze players. Also, do we know that mmr point trades are always even like for displayed points. I don't know. Need to go back and read excalibur's stuff I guess. Anyway, my point isn,t that bronze players currently make up a disproportionately large % of the player base. There are roughly 30% more bronze level players than there should be. That may be an indication that the league itself is more challenging. | ||
|
Lysenko
Iceland2128 Posts
On April 01 2012 07:09 ttfdrevil wrote: I'm not sure I fully agree with you. I think my point is that the distribution has changed, but blizzard has opted not to Adjust the bars. Unfortunately, it sounds from your comments like you didn't understand what I wrote. Player activity serves to spread players evenly through the leagues. (MMR itself may not end up spread evenly, but it's correlated to percentile among active players, so the MMR boundaries are fixed at uniform 20-percentile intervals.) A player who places and then doesn't play actively has no impact on the MMR system at all. The five placement matches yield an estimate of that player's MMR, and the player's dropped somewhere on the MMR spectrum, but in a way that doesn't move any other players around. Meanwhile, the active players' wins and losses push each other around in such a way as to distribute them consistently with the MMR boundaries. Thus, an excess of inactive players placing into bronze and ceasing to play further is all it takes for that league to be larger. Think about what happens when a player places into bronze: they lose five placement games and get a window indicating that they're in the lowest league. A player who's ambivalent about multiplayer SC2 would be far more likely to stop at that point than if they won one of their games and placed into silver. This mechanism alone would account for more people being in that position in bronze league. Also, of course, bronze players are probably less likely to take the game seriously anyway. Also, i'm not sure mmr is a zero sum game when it is manipulated by people with second accounts or by first account smurfs. At minimum it must impact mmr "confidence" for real bronze players. A player playing on a second account is indistinguishable from a second human being at the same skill level. A player who deliberately quits a game to lower their MMR is indistinguishable from a player who lost that game fairly, and for the opponent, it's indistinguishable from a win. Manipulation doesn't affect how the MMR system works, other than that smurfs have a higher chance of winning than the system estimates on those occasions when that player decides to play a game out. Remember, MMR doesn't have anything to do with actual skill, it only reflects wins and losses, so if a large number of people around you are throwing games, you'll have a higher MMR as a result. In an extreme hypothetical case, if you had 20% of the active players throwing EVERY game, any player who tried to play the game at all would wind up in silver. Also, do we know that mmr point trades are always even like for displayed points. I don't know. Need to go back and read excalibur's stuff I guess. Displayed points are not always even between the winner and loser, but MMR movement probably is. The reason is that the difference between two players' MMR numbers maps directly to their likelihood of a win. Two matched players (on average, across many matches) are going to have the same MMR. A win or loss provides additional information about the likelihood of one of those two players beating the other one, but it doesn't provide distinguishable, individualized information about one player's skill level vs. their opponent's. After a game where the two players start out at the same MMR, the players need to be moved apart by some amount to reflect that likelihood of a win or loss, but the information simply isn't there to move one player farther than the other one. Incidentally, that MMR is zero sum is strongly supported by the fact that the league boundaries' MMRs are relatively stable, and that when they make changes, people notice. If MMR were not zero sum, there would be a trend toward inflation or deflation. EDIT: Here's another way to think about the first reason that MMR is zero sum: Two players with the same MMR each have a 50% chance of a win, because that's what equal MMR means. If MMR were not zero sum, a series of losses followed by an equal number of wins would not always get those two players back to the same MMR, because whatever your scheme for assigning extra points to one player but not taking them from the other, it would cause a round trip not to work. So, two players with equal MMR, after playing a large number of games where they went exactly 50/50 vs. each other, would wind up with different MMRs despite the 50% win likelihood predicted by their starting equal MMRs being confirmed perfectly. That's a nonsensical result -- you start out predicting a 50/50 win loss split, and when that happens exactly as predicted, your MMRs move apart to favor one player or the other? That would make no sense. Anyway, my point isn,t that bronze players currently make up a disproportionately large % of the player base. There are roughly 30% more bronze level players than there should be. That may be an indication that the league itself is more challenging. All it indicates is that a large number of bronze players don't play enough after their placements to move other players' MMRs upward with their losses. | ||
|
ttfdrevil
United States16 Posts
You are right that the leagues are split based on active population. But blizzard estimates the splits and sets a fixed mmr for promotion. The last time blizzard changed those fixed boundaries they made an announcement. I believe it was season 4 and immediately after they adjusted those boundaries I was promoted to silver. I do not believe that they have adjusted them since. Also Excalibur notes that even though the mmr ranking requires to promote is fixed, the ability of a player to get to that level will depend on the skill of the player population (my original point). Also, as the leagues are only 2 months long now, I would find it surprising that the drop off rate for bronze players (from active status) is roughly 30% higher than the other leagues (hence making the league distribution look skewed due to inactivity). It could be, I don't know. So I still hold that it could also be a function of it being more challenging to reach the mmr breakpoint set by blizzard. But I don't discount that inactivity could also be the cause. | ||
|
Lysenko
Iceland2128 Posts
On April 01 2012 08:55 ttfdrevil wrote: So I still hold that it could also be a function of it being more challenging to reach the mmr breakpoint set by blizzard. On the question of inactivity: First, the distribution across the leagues isn't a sharp knee at bronze. Currently, according to sc2ranks.com, the percentage of tracked North America players per league, this season, is: Grandmaster: 0.1% Master: 4.7% Diamond: 9.3% Platinum: 16.8% Gold: 19.4% Silver: 22.9% Bronze: 26.9% I'm not even sure what effect we're trying to explain with this. I don't see some big discrepancy between adjacent leagues here until you hit Diamond. The median number of wins (this season) per player in each league (by their statistics) is: Grandmaster: 121 Master: 31 Diamond: 17 Platinum: 8 Gold: 7 Silver: 7 Bronze: 8 Now, what's interesting is looking at the data for the last patch of Season 1, 1.2.0: Master: 1.1% Diamond: 7.1% Platinum: 12.4% Gold: 15.9% Silver: 17.7% Bronze: 45.9% The Season 1 data illustrates what the numbers look like when tons of new players are placing and quitting. I'd say that today's numbers look a lot more like what you'd expect if there are no new players and the leagues are mostly sorted by how much players practice. I don't see some enormous knee around Bronze that suggests much of anything in today's data. Now, there were a couple of major ladder boundary changes in there that shifted bronze through platinum upward in league a bunch. Still, it's pretty obvious from the distributions that in Season 1, a lot of people tried multiplayer, said "wow, this is hard," and quit. Today, most people are hanging in there to some extent, but diamond and up play a lot more than the other leagues. | ||
|
ttfdrevil
United States16 Posts
Considering both games (wins) and average (not median) points per player are also about comparable, I don't see any sign that bronze players are more inactive than higher leagues (at least through plat. If anything it appears bronze may be slightly more active. So again... Implies that inactivity may not be the cause. Difficulty reaching the mmr limit appears more likely (due to increase in difficulty). | ||
|
Lysenko
Iceland2128 Posts
On April 01 2012 18:08 ttfdrevil wrote: again... Implies that inactivity may not be the cause. Difficulty reaching the mmr limit appears more likely (due to increase in difficulty). Cause of what? I posted those numbers specifically to point out that bronze is currently not that much larger than silver. I'm not sure 6% is a meaningful difference here, since those MMR boundaries are manually adjusted. It certainly does not suggest that it's any harder to get out of bronze than it was -- if anything, it's easier than ever before because fewer people are staying there. | ||
|
Monkeyballs25
531 Posts
On April 01 2012 06:40 minilance wrote: people in plat and under are still shit, I was a mid diamond player and i downed my mmr to play in lower league, i have a 60% win ratio by walling my main with 2 nexus, building 20 photon canon and then doing a 4 gate.... In which league?  I met a guy in silver who does something similar. He plays random, and anytime he spawns as Protoss or Zerg he just makes a fuckton of cannons/spines at his ramp. Not sure what happens afterwards since I'm only viewing them from the build order overview, but he still manages to win some games with it somehow... My theory is their opponents do their standard one base push, suicide it onto the static d and then panic and lose the game. | ||
|
Monkeyballs25
531 Posts
On April 01 2012 19:40 Lysenko wrote: Cause of what? I posted those numbers specifically to point out that bronze is currently not that much larger than silver. I'm not sure 6% is a meaningful difference here, since those MMR boundaries are manually adjusted. It certainly does not suggest that it's any harder to get out of bronze than it was -- if anything, it's easier than ever before because fewer people are staying there. Its probably harder for a new player to get out of bronze, since all the leagues above it seem to have gotten better. If we're debating how hard it is to get out of bronze, its really the shift in silver level play we should be looking at. | ||
|
framtidenskrig
Sweden92 Posts
| ||
|
Adamgm
Canada68 Posts
I haven't ever stopped really lol. | ||
|
Lysenko
Iceland2128 Posts
| ||
|
0neder
United States3733 Posts
| ||
|
Zariel
Australia1285 Posts
Essentially, they don't play to win. Instead they play to see their 200/200 void ray army kill the enemy or die trying. The difference in 'macro' between a bronze player than the rest of the leagues is that bronze players tend to 'zone out' to fights and like to think for like 10 seconds on 'where should i place this gateway' | ||
| ||
StarCraft 2 StarCraft: Brood War Britney Dota 2  39122Calm  7792Jaedong 3014Bisu 2922Flash  2540Shuttle 2253Horang2 1451BeSt 1156ToSsGirL 1031Zeus 797[ Show more ] Counter-Strike Super Smash Bros Other Games singsing2560 B2W.Neo1616 Beastyqt1188 hiko1114 Dendi700 Lowko686 XBOCT563 Liquid`VortiX162 ArmadaUGS125 KnowMe123 Trikslyr51 ZerO(Twitch)36 Organizations StarCraft 2 StarCraft: Brood War StarCraft 2 StarCraft: Brood War |
|
Monday Night Weeklies
RotterdaM580
Replay Cast
Afreeca Starleague
Soulkey vs Rush
Replay Cast
Kung Fu Cup
PiGosaur Monday
OSC
GSL Code S
Cure vs sOs
Reynor vs Solar
Replay Cast
GSL Code S
Maru vs TriGGeR
Rogue vs NightMare
[ Show More ] |
|
|
EPT
