you realize pros spam apm all the time, you can infact see it, you arent proving anything. if 200 apm player beats a 500 apm player, whos better? the 200 apm player played better. shhh
2 Likes
you all talk, oh plz you post nonsense and prove how stupid you are. you can get high apm spamming and apm only matters if you micro well, you can litterally spam group hotkeys to get high apm and location camera view keys. like i said if a 100 apm beats a 500 apm player who better? obv the 100 apm player
1 Like
A player’s average APM across all their games is a very different concept from their average APM after one game. This isn’t complex, guys. Reading comprehension is your friend.
What, do you think pros will only spam APM once in a blue moon or something?
Yeah I do have a permanent injury to my left middle finger which makes it painful to play long games. Even so, I get into Grandmaster ez pz with drone pull allins and other comparable nonsense. This game is too easy. They need to nerf Zerg. The apes have 2 races to play so if they nerf Zerg then the geniuses can have something to play too. While they are at it, they should add official maphacks to the game. If you need maphacks to win, no judgement. We need hackers to make the game harder for people who think the game is just too easy. Most streamers cry about hackers and streamcheaters. I welcome them. I want all the streamsnipers to hit my stream. Maybe then the game won’t be so easy.
I mean, it’s a video game for crying out loud. If they actually made it hard, people wouldn’t play it. It’s a video game so, by definition, it is easy as pie. One, two, three, count with me, and we will make it to Grandmaster league!
They did. Its called zerg.
Expecting a reddit thread saying didn’t think I’d cry. Congrats on GM, champ.
What are you talking about? Oh. I see. It’s the start of the season GM promotion. Those don’t count, man. End of the season promotions don’t count either because players stop playing once their MMR is high enough for GM then coast to a finish. Only when lots of people are actively playing, and specifically the account that has the GM badge, does a GM badge count.
That’s the thing you have to realize about GM promotions. People cheat the system in a variety of ways and so this skews the statistics and that changes how you have to interpret the data.
By your own logic the hardest time to achieve GM will be at the end of the season. May as well have set the new requirements for a GM player. Must play the race with the least GM representation, must be on the hardest server (EU) and you can’t play the same builds more than 5 times.
It depends on how often you play. If you play often, you need your average performance to be well above the threshold. If you play infrequently, you need your peak to be above the threshold: as soon as your mmr peaks in Grandmaster, you switch accounts or play unranked or you off-race. The vast majority of people are in this category.
You have to visualize Grandmaster as a distribution of peak MMRs, not average MMRs.
I don’t think Blizzard’s algorithm is very robust. It’s very easy to exploit, doesn’t have a way to cope with smurfs, etc. There are people who queue games against their own accounts to give themselves 8000 mmr. The MMR algorithm is inherently volatile, meaning a player’s rank will fluctuate by +/- 500 mmr.
A simple solution to the volatility would be to smooth the MMR rankings. Your MMR ranking becomes the average of your past 100 games, for example. This would take care of quite a few ladder exploits. All the people who peak their MMR into GM and then camp there without playing any games would never get into GM to begin with. The odds of going on a 10 game winning streak and getting into GM is pretty high and with enough people it happens basically all the time. The odds of going on a 100 game winning streak is basically zero unless you really are GM level.
Another thing I wish they did is separate MMR rankings per matchup, then combining the separate rankings into your final MMR ranking. This is useful because it puts equal weight on each matchup. Right now vP is the most important matchup to get into GM because half of GM is Protoss so half your games will be against Protoss and that causes vs P to dominate your MMR ranking.
I call bs, unless someone plays when they’re blind drunk or something. I’ve played the same people that have stayed at the same mmr for literal years. Upatree for example hasn’t fluctuated by more than 300 in the entire time he’s played.
How to kill the ladder featuring batz. You must understand simple minds get dopamine when number goes up. The faster the rise, the more dopamine.
Agreed, all we have is win rate so I can only boast about beating terran 70% of the time but those terrans are so low.
PS have you noticed Atlanta had APM displayed on the main stage throughout the game? The highest I saw was Maru with 805 against Byun.
https://sc2pulse.nephest.com/sc2/?type=character&id=1606235&m=1#player-stats-mmr
this is a perfect example. His MMR bounces around until he hits 5.6 and then he quits playing.
There are three random variables at play. A player’s performance will vary. A player’s opponents will vary. External factors will vary like lag. The variability of a player’s ranking will be equal to the sum of these three multiplied by the MMR’s “k factor” which literally scales the variance. Usually a k-factor of 32 is used. That means the natural variance is multiplied by 32. They do this so that the algorithm can quickly adapt a player’s ranking to new performance changes. Regardless, a player can have any combination of “good day” and “bad day” and “easy opponents” and “hard opponents” etc and statistically some of these will land on “easy opponents, good day, no lag” and they will pop up into GM league just long enough to secure a spot, then they stop playing.
You don’t have to show them their real mmr number for crying out loud. Just use the real one for matchmaking and league ranking.
I did not notice, but it doesn’t surprise me. Maru is the best SC2 player after Serral. Innovation had him beat until he retired, but Maru is definitely #2 now. Since SC2 is an APM spam game, it basically predicts all the top players will be APM spammers. So, finding out that Maru is spamming 800 apm really is not surprising at all. 
I actually forgot that a tournament was going on. How many people are tuning in to see APM spammers spam clicks at other APM spammers? Golly, I just can’t wait to tune into that. When was the last time somebody pulled out interesting builds? I can’t remember the last time I saw a thor drop. Strategically, this game is dead. It’s all about spamming APM faster than the other guy.
LOL you dumb? Maru… and pros spam apm always not once a while ya dumbass, reading comprehension, you need a proper education to understand that Apm isnt important, everyone spams apm all their games not just once, go actually watch pros stream and even maru who streams, he spams apm alot so he can keep that 800 or whatever apm as his averge apm. like i said if a 500 apm player loses to a 150- 200 apm player whos better? answer the question dumbass
The fact that all those factors like rapid-fire, outliers, and racial differences impact apm yet it is still so tightly correlated to rank is a testament to its strength as a skill metric.
1 Like
Well, MMR will fluctuate on its own just due to luck. Some days you will play well, other days you won’t. Some days you will have easy opponents, some days you won’t. The fact that MMR correlates with APM at p=0.65 is absolutely insane considering how much MMR varies on its own. APM is quite likely very near the limit to how tightly a variable could correlate considering the natural variability of MMR.
I’ve been meaning to do a mini study that correlates average MMR against average APM but haven’t found the time. It’s a bit complex because you have to estimate certain factors that we don’t have access to. The matchmaker, which matches players together, will be a huge factor in deciding the variability of MMR but we don’t have access to how it works. It would have to be estimated from match histories, and the player match history doesn’t store MMR values. You could estimate against replays, except the replays are white-washed of any player information. So, no matter where you turn it’s lacking crucial information.
So, what you have to do is estimate what the most likely values are for these variables and then give a confidence interval, e.g. MMR fluctuation is likely 236 +/- 54 in 95% of cases. You have to build a probability model making assumptions about what variables there are and how they behave, simulate it with a monte carlo simulation, and compare its output to ladder data. If the data doesn’t match the model, then you have to redo the model. All that is quite the project. Once you have a working MMR model, you can then calculate what the most likely average MMR fluctuation is, and that tells you what percentage of the correlation between apm & mmr is due to mmr’s volatility.
APM means a lot. Someone with 200 apm is going to annihilate someone with 100. That still doesn’t mean that someone with 400apm will beat someone with 200apm though. Definitely a steep drop off. Basically, if you aren’t Diamond 1 or above, APM means a lot.