Toss hits 43% in GM

Toss just hit 43% in GM. They set a record 45% in 2020. Will Protoss beat the previous record? I guess we will find out. Even during the brood infestor era of Wings of Liberty, Zerg’s representation peaked at 40%. Protoss apologists will continue downplaying the issue, even though we all know current toss is more busted than brood infestor ever was. It’s just the facts. Toss turtles behind cannons while rushing straight to carriers, and they do this in both PvT and PvZ, and that’s exactly how brood infestor worked in Wings. You skip the early and mid game, slowing down only to get enough D to survive, and head straight to the free-win condition.

The question is if the balance counsel can man up, admit the pro scene has a talent issue on the protoss side, and start to prioritize balance instead of using affirmative action to give a protoss premier win. If a protoss won a premier under the current balance conditions, people will always be wondering if he could’ve done it if the game had been balanced. In other words, did he win it, or did protoss win it for him? The tragic thing is that if a toss did win it, and he deserved it, balance would cast doubt on his victory.

Buffing protoss to the moon is not going to make protoss win a premier, it will rob a protoss of the prestige of winning a premier. It’s going to backfire. I am calling it now. Protoss wins and 3 2 1 0 we will have people saying he didn’t deserve it.


The difference to Bl/ infestor is that Zerg has to actively play for defense.
While toss is more passive. only one worker for cannon/ sb / nexsus (all minerals only) spam and gas in tec/ units.

1 Like

Good points. Also, the brood infestor army was more positionally vulnerable because it moved slower & didn’t have recall. Toss now have 2 types of recall. Brood/infestor was basically limited to turtling on creep. Broods also were heavily specialized anti ground, all skytoss units attack air & ground. The range advantage of tempest/carrier also exceeds what broods were capable of achieving (9.5 vs 14). Toss also has cloaking, which is something that the brood army never had. The two just don’t compare. It’s obvious why toss are hitting 45% in GM meanwhile brood infestor only hit 40% historically.

People called broods imba because they were. Skytoss is much more imbalanced than brood infestor but nobody calls it out. The most I’ve seen is Harstem say PvZ favors the toss 55% over the zerg’s 45%. If you did a proper statistical analysis, the actual win-rate is closer to 70% for toss and 30% for zerg. That can be calculated with 3 different methods and they all give the same result. Most people don’t realize that there are second and third order effects that have to be accounted for before you can measure balance. You can’t get an accurate result by looking at a win-rate because the ladder demotes people to equalize their win-rates to 50/50. It’s complex math, and it’s hard to do the calculation right, which is why there is so much confusion out there about how busted protoss is right now.

Nobody seems to realize that this level of dominance has literally never happened before. Nothing has been more imbalanced, in the history of the game, than current protoss. David Kim used to dish out nerfs if win-rates went above 55%. Lmao. It’s wild how bad the state of the game is at the moment.

1 Like

Hero needs to stop giving Patience the controller for the first 5 minutes of the game like in the GSL finals.

1 Like

Stats was more interesting than hero. A 6200 mmr toss a-moves carriers through a 6900 mmr zerg, but then is bopped in a PvP by a 6600 protoss. Stats lost to the same toss who proceeded to misrally a free colossus into marus army, blind blinked up a ramp into a meat grinder, and other kinds of terrible mistakes.

Yes, skytoss is so insanely busted that it boosts a 6200 toss +700 mmr to beat 6900 zergs. Balance in sc2 is a meme. The current balance model is more or less an oligarchy of sorts, and what a perfect example it has become for why oligarchies are terrible forms of governance. David Kim would nerf a race if they got to 55% winrate, but current toss is so far beyond that and, despite this, I guarantee the balance counsel is thinking of ways to buff toss even more.

It’s the worst balance the game has ever seen, and they are concerned about if the outlier of outliers wins a best of 7. The amount of bias is hysterical. I might even use the word delusional to describe it. It’s the funniest thing. They are professional players and yet they clearly do not understand the game. That’s the problem with designing sc2 around apm spam. Top leagues are filled with apm spammers who know nothing about the game. They are only there because they click fast, and that’s it. That’s who we have balancing the game. Face palm.

I mean, while that might have been the end of that particular game/series (I assume you’re referring to the GSL), Stats vastly outplayed Dark with constant multi-pronged pokes and harass the entire series, sniping key units, bases drones and generally being a nuisance. He played extremely well, regardless of his current ladder MMR.

1 Like

Curious that stats was able to do that in a pvz, but not a pvp. If skill were the cause, a 6600 toss should be easy after beating a 6900 zerg.

To be fair, they’re both completely different matchups, and HerO’s PvP is generally considered to be pretty good while Stat’s PvP has historically been his weakest matchup because of the general volatility of the mirror.


Hmm. So there’s a difference between the matchups that causes a drastic performance change. Interesting. I wonder what could be causing that. Oh wait, I got it, it’s the Synergy between carriers and storms.

That should be obvious. The differences in economy and the units available in each match-up change the way that the match-up should be played. Since different players tend to have different strengths, this usually results in a different win-rate in each match-up.

There are a lot more differences than just the performance of Carriers and Storm against Protoss/Zerg units respectively.

If I understand the match-up correctly, Gateway units can also be relatively more dangerous in PvP, and with Warpgates that can result in a very volatile match-up.


Let’s put your reasoning into another context to see if it makes sense. Sally takes a Tylenol to treat a headache, but it doesn’t work. Sally tries taking an Ibuprofen, and it works. Your conclusion: Wow, sally is more skilled at taking Ibuprofen!

When the matchup changes and there is a corresponding performance change, that means the matchup is the cause. To correlate performance with player, you have to change the player. We’re not doing that. We have a player punching 700 mmr higher than his typical performance, but only in one matchup. The performance is matchup dependent, not player dependent.

The correlation between symmetrical matchup performance and asymmetrical matchup performance is very high. It’s like 0.97 or 0.98. It’s been awhile since I calculated it. The point is, matchup performance is just player skill. Individual matchups don’t vary enough to be distinguishable from a player’s skill in general. If a player can beat a 6900 zerg, he should be able to demolish a 6600 protoss because 6600 is way weaker than 6900. That’s not what happened. Protoss won the PvZ because the asymmetry of PvZ gives protoss an advantage, and that advantage doesn’t exist in a PvP, because it’s symmetrical, so he lost the PvP.

The only variables that affect a mirror match are skill and luck, because the players have access to the same tools. Asymmetrical matchups can be affected by the differences between the races. Protoss in general are very strong against zerg. It’s not like Stats is the only one experiencing insane performance boosts with protoss.

What Stats proves is that one of the excuses, used by the “balance counsel,” is totally wrong. They said carriers only affect GM and that only “bad” players are struggling with carriers. They are so flipping wrong. That’s why Stat’s performance is so interesting. It shows, once again, the balance counsel’s intuitions are totally wrong. They just don’t know what they are doing. They think toss is underpowered and that’s a literal inversion of the state of balance.

Frankly it’s amazing to me, as a scientist who is fascinated by human psychology and sociology, that “experts” can have beliefs that are so radically detached from reality. It’s amazing. We’re talking flat-earth levels of delusion from people who are supposed to be the world’s smartest people when it comes to SC2. How do you explain that, scientifically? That’s why it’s fascinating. It’s a case study in human bias specifically how it applies to experts.

The best explanation is that game knowledge isn’t predictive of SC2 performance. That’s the only way to explain this. They are good at SC2, but they don’t understand SC2, so game knowledge isn’t what causes their talent at the game. So being talented at the game is a distinct concept from understanding the game, and that’s interesting. So, what is the skill that SC2 requires? It’s multitasking. You don’t need good game understanding, you just have to follow a build order & have good multitasking, and that’s it. That’s SC2 in a nutshell.

PiG was doing a cast wherein he talked about a conversation he had with a professional protoss. This protoss refused to play on the new maps. He didn’t like them. PiG convinces him to try a map that is insanely busted in favor of toss, and this protoss demolishes his zerg opponent despite not having played on the map. He was able to do this because PiG pointed out that there was a particular attack that can be done at a certain base which was virtually unstoppable. So this protoss, a PROFESSIONAL, clearly can’t look at maps and theorize strategies to win. Professional SC2 gamers are literally APM spammers who have near zero game understanding. Literally. :exploding_head: :exploding_head: :exploding_head:

A lot of streamers outside of SC2 understand this. Asmongold is a good example. He said he isn’t going to stream SC2 because, more or less, it’s too emphasized on mechanical skills like multitasking. The guys who designed League understand this. That’s why they took the RTS genre, dialed the multitasking back, and, BAM, they have a super successful game. Players like games that emphasize unit interactions, micro, aka tactics & strategy, and that’s not what SC2 emphasizes: it emphasizes multitasking. That’s why league is literally over a thousand times more popular.

Because they buffed the economy in LotV, unit interactions don’t matter. Unit retainment is what drives the need to micro. If you have 10 marines, then every marine matters. If you have 100 marines, then individual marines don’t matter. This is why the micro in modern pro play is so incredibly bad. The strength of the economy makes unit retainment much less important, and that means it’s not necessary to micro your heart out to get maximum value out of each individual unit. In HotS, there was a terran who killed 8 zealots with 3 marines. That’s amazing. In this GSL, there was a terran who threw away 5 medivacs, full of marines, to a single parasitic bomb, and he won the game none the less. Unit retainment is irrelevant, that means players don’t have to micro, and the lack of micro makes unit interactions in current SC2 extremely boring. Do you want to know why SC2 is dying? This is why.

If the game is to survive, they need to be realistic about balance, aka nerf protoss, and they need to nerf the strength of the economy so that unit retainment is important again. If this doesn’t happen, SC2 is toast.

I am a game developer and I am putting these ideas out there because I like SC2 and I don’t want to see it turned into a meme that the gaming community laughs at. I don’t have to do this. It takes time, and it’s against my financial interests. It would be smarter for me to make my own RTS game than to give out free game design consultations. I guess at the end of the day I don’t feel like I’ve lost anything because the pro players will never take this advice anyway. They are too biased. SC2 is toast. That’s just reality. Game developers know this and that’s why there are new RTS games entering the market. Other companies smell weakness. They know SC2 is in a bad spot and can’t retain its player base. Players will switch games if the current memery is allowed to continue, and it will continue because APM spammers don’t want to nerf themselves. That’s really what it boils down to. Pro players won’t vote against their own financial interests.


That is an absurd comparison, since it is purely biological and has nothing to do with a person’s skill or input.

A more accurate comparison might be to compare two people playing different instruments, such as the Guitar and the Violin. Both require some skill, and they are different enough to each other that a person can become better at one than the other and vice-versa.

A person can be more skilled at using and controlling one playstyle than another, and the differences between races can and often do favor one set of compositions or playstyles in one match-up and a different set in another.


Biological or not has nothing to do with it. It’s a process which has an element of luck and we want to know what variables control the outcome.

Experiment 1:
Variable 1: True
Variable 2: True
Outcome: True

Experiment 2:
Variable 1: False
Variable 2: True
Outcome: False.

Your conclusion: variable 2 controls the outcome even though the outcome changed when variable 2 stayed constant. Aka, your theory contradicts the data. It’s that simple.

In case you are wondering, variable 1 is “true or false, is this matchup asymmetrical”, and variable 2 is “true or false, is the protoss player stats”. The outcome is, “True or false, did Stats win?”. The player stays constant, but the outcome changes. The outcome changes at the same time as the symmetry of the matchup changes. The symmetry has a correspondence with the outcome, the player does not.

Definitely false. The variability at the pro level can’t explain an effect this large. At the pro level, matchup performance correlates with a player’s skill in the ballpark of 0.97. That means any variability in performance, between matchups, is within the standard mean error. The standard mean error represents the inherent volatility of an experiment. A player’s performance per matchup varies so little it’s indistinguishable from background noise.

If a player is winning at a 6900 mmr level in PvZ and losing at a 6600 mmr level in PvP, it means their true skill level is <= 6600 and the asymmetries of PvZ gives a >=300 mmr advantage. Amazingly, that perfectly corresponds with numbers generated using 3 separate calculation methods on the GM population (here, here).

If it walks like a duck, quacks like a duck, looks like a duck, it’s a duck. This isn’t variability in matchup skill. It’s Protoss being advantaged.

A player can be good at PvZ and bad at PvP. You’re completely discounting this as a possibility, with no reason or evidence.

Like most of your posts, you’re saying something that’s complete nonsense and pointless.


I was wondering when you’d chip in. You’ve been keeping your distance. I’m telling you now this post will age like milk. You could just ask for the charts like a normal human being, but you gotta do you, lmao. Watching internet randos attempt debate is like watching a skateboarder lose his teeth to the concrete. You’re gonna pick a fight with one of the worlds leading experts on rts game theory. Bold move, cotton.

Alright, time for your dose of reality. SC2pulse reports Stats’ mmr is 6254. Dark’s is 6894. The mmr algorithm is detailed in the AlphaStar paper (page 9):

P[0i = 1] = cdf((r1 - r2) / 568))

We are going to rearrange this to make it easier to work with:

P(Win)=1/(1 + exp(X / 568)) 
where X is the mmr difference.

We can use this to calculate the odds of a 6254 mmr player beating a 6894 mmr player, like so:

P(Win)=1/(1 + exp(640 / 568))`
P(Win) = 0.24

So Stats has a 24% chance to beat Dark in a best of 1. But, they played a best of three. What are the odds of winning that?

P(WinningSeries) = (P(Win) & P(Win)) | (P(Win)  & P(Win) & P(Loss))

Solve for the first term:

P(Win) & P(Win)) = 0.24 * 0.24 = 0.0576

And the second term:

P(Win) & P(Win)) = 0.24 * 0.24 * 0.76 = 0.043776

Combine them together:

P(WinningSeries) = 1.0 - ((1.0 - 0.0576) * (1.0 - 0.043776))
P(WinningSeries) = 0.0988545024.

So, Stats has a 1 in 10 odds of beating Dark, which meets statistical significance. What if Stats’ performance is biased by Protoss by the same amounts as measured on GM trends? Well that just bumps his mmr up by ~300. That makes the previous calculations:

P(Win)=1/(1 + exp(340 / 568))`
P(Win) = 0.35

His win-rates in a best of 1 goes up by ~10%. Let’s do the calculation for a best of three:

P(WinningSeries)  = 1.0 - ((1.0 - 0.126) * (1.0 - 0.082))
P(WinningSeries) = 0.2

His win-rate doubles, but is still a 1 in 5 odds of occurring.

The odds of Stats beating Dark is 1 in 10 if you assume Protoss has no advantage, but it’s 1 in 5 if you assume Stats receives a +300 mmr boost in PvZ like what happens in Grandmaster. Even with a balance advantage, it’s still an unlikely event. Thus my point is proven: the balance issues of carriers affects the highest levels of SC2 pro play. The “balance counsel” is wrong in their assertion that only low level zergs struggle to beat carriers, and we can say that with a 90% confidence. Carriers need nerfs. It’s just a fact of this thing called R E A L I T Y.

Not only am I right, I proved it. You were provided with the mmr numbers, which were sufficient to prove my point, you just didn’t know what they meant. Please don’t think that an argument is wrong merely because you can’t understand it. If you wanted me to explain it to you, I’d have been glad to do so, but, instead you are all “your argument is nonsense and pointless” and “you’ve provided no proof” aka literal untruths so detached from reality they are bordering on a lie. How am I supposed to respond to that.

This is why I don’t engage with the SC2 community very much, and why I avoided streaming for so long. The “neckbeard syndrome” is intense inside the sc2 community. All you had to do was ask, but, no, we couldn’t be friends because I had opinion about sc2 that you didn’t like. Lmao. Unreal. Oh well. It is what it is. Not everyone thinks scientifically and I am just going to have to accept that.

1 Like

This isn’t me arguing, i’m just discussing Protoss. Cannons, probes and carriers are so simple that GM is full of them. To beat it as Terran or Zerg you have to be genuinely gifted. That is why, I believe at the top level Protoss is underrepresented, but it is such a small percentage of players that it doesn’t feel fair.

Nerfing these tools and giving Protoss creative ways to win might be nice, say… I would nerf EMP. Maybe Toss is lazy, but even impressive players like Creator and Hero have problems winning. It shows to me that braindead toss is too strong (no offense) but skilled toss (blink stalkers) are too weak. Probably Nerf T and Z answers to that style, but nerf noob toss.


You are headed in the right direction. It’s the scale parameter that is messed up in addition to the mean. Average toss performance is higher, but it has a lower variance, which clips it at the highest levels. The performance is tightly clustered in other words. It’s not spread out enough to allow for super high performance players.

The problem is that the scale parameter is probably caused by the same issue as the mean. Toss is easier to play, this gives reliably higher performance, but it lacks the refinement capabilities needed for super high level play. By making it easier, they removed the refinement capabilities. A carrier represens 6 supply. A Zerg could choose to make 3 hydras, 2 hydras and 1 roach, 2 roaches and 1 hydra, 3 roaches; 12 lings, 8 lings and 1 roach, 4 lings and 2 roaches, etc. Protoss units are very easy, but they lack refinement, and the reason they lack refinement is because they are easy. The ease of use causes higher performance, but it reduces the peak performance by causing performance to be more clustered.

That’s why I’ve been saying, for years, they need to nerf a-move mechanics and buff the high skill mechanics. They’ve been doing the opposite. They went on a mission to nerf the blink stalker over and over again, while buffing skytoss. Huge mistakes were made by the balance designers.


By the way, since you ignored my previous posts talking about how individual matchups don’t deviate much for a single person, and then lied that I didn’t provide “reason or evidence”, we can take a look at Stats’ aligulac page. It provides per-matchup ratings. Let’s see if Stats is better in PvZ:

  • PvP: 3019 +/- 129.
  • PvT: 3033 +/- 131.
  • PvZ: 3071 +/- 131.

As I was saying, it is unusual to have large differences in performance, between matchups, without a balance issue. It can vary a bit due to random fluctuations, luck, etc, but generally it’s going to correlate very strongly with the player’s skill level. That’s because a player’s performance is decided by their skill, and not the matchup. Clearly that is the case. All of Stat’s ratings are within the margin for his other ratings. In other words, they are identical.

So now we can put to bed the theory that Stats is hiding a super secret magical PvZ talent that drastically outshines his other matchups. Carriers are just really flipping strong. The balance counsel’s theory, that only low GM’s zergs struggle with carriers, is obviously false.

Ironic because to make it deep into tournaments you have to have exceptional vP performance.

1 Like