ok then let me ask you - what would be balance to you? ive never said, that winrates were perfect. i said multiple times, that we had some 45% periods. and i said, that i want the winrates to be as close as possible to 50%. wheres the problem with that?
i believe, that youre salty, because terran never got that 55%. if you would be truely interested in a balanced mu, you would want to reach 50% winrate, but instead, you want your own 55% for terran. that would make you happy, right? because that would be âfairâ to you, because zerg had 55% multiple times - right?
it does.
i said 47-53% which is⌠actually the situation were living in. maybe you should look more closely, there were times, where terran was favoured.
obviously. because to have that 50% overall, it would mean, that terran has to have periods of 55% or even more their own. or you would need a really long period close to 50%, which isnt possible.
how often do you wanna hear it? i would be pretty ok with a 53% winrate for terran.
btw bourne, whats your opinion about the balance in tvp?
I think for win rates to matter you have to control for the strength of the player.
Win rates only matter if the players are similarly ranked.
A good way perhaps for balance testing would be to use something like alphazero and if the win rate of alphazerg is way too high vs alphaterran or something then you make specific changes until itâs normalizes.
Using Alphastar for balance is a terrible idea. The game is for Humans and Alphastar not being a human plays the game so weirdly at times that it becomes a completely different game. Balancing around that is akin (probably much worse) to balancing the game around the Meta of 5 years ago. If you want the game to be balanced around human players playing in todayâs meta you should use statistical tools + careful analysis of the game design to understand what needs to be changed if at all. This is difficult enough as is. You will not magically achieve this by going after a completely different goal.
Unfortunately, there is a loop within that method. Imbalances in the race that a player plays translate into wins, league placement, performance in tournaments, etc; so the skill that you are trying to control for is affected by the state of balance you are trying to measure.
Well, 55% should be rare. Youâll notice that it happened in TVZ⌠almost 10 years ago. In ZVT itâs happened 3 times in the past 15 months⌠Thatâs an awful strain of bad luck, if thatâs what it isâŚ
Again, strawman, typical worthless Zerg nonsense. We are currently in a sub 45% period and youâre acting like I want my own 55% period.
If the match up were incredibly volatile (rather than incredibly one sided), you might call that balance, but itâs crappy design.
Very, VERY rarely. In fact, Zerg was favored 4 times as often (and much more heavily, on average) as Terran was.
Furthermore, you find that exact same trend in PVZ.
Then you look at Premier Tournament wins. If they were central to 1 or two player, you could dismiss them as anecdotal, but instead you have 5 different players winning a total of 65% of tournaments. To put that in perspective, Terran has had 5 different winners total since 2016. Youâre talking about 4 years of Terran success crammed into one single year of ZergâŚ
Well, thatâs definitely not what we see. And I would not (one or two periods would be fine, but not for an extended period), because that would mean an extented period of Terran dominance when it came to the margins. I want whatâs fair, not beneficial.
North of 55% for Protoss across over 1000 games. Looks pretty bad. Again, likely maps, not units.
That being said the balance team has absolutely no valid excuses for the completely atrocious state of the game at blizzcon (it remains to be determined if the last patch helped - my guess it solved only some mild problems and the serious problems with Zerg are yet to be addressed).
People were âwhiningâ about Nydus being broken since early 2019. Everyone told these people to shut up and that Nydus is not really op blabla. Everyone knows the rest of the story. On the other hand it took a single GSL ST to give Zergs a completely unnecessary scouting buff reverted only after blizzcon.
The moral of the story is, game design is important, winrates are important, balance team doesnât seem to pay attention to either. It must be occupied with listening to âconsultantsâ like Catz not shutting up about their stupid biased opinions like observers being imba or the Adept nerf being a buff (check his twitter if you donât believe me).
This is from November, in the current period PvT is at 44.72% and we still have 9 days to go.
Anyway as we all saw even with the changes zergs found a way to be competitive because their race as a whole has the tools. It is a complete race that if you play well you can win when it is not necessarily the case for protoss or terran.
Zerg has always been broken, I play this since the very begin and it has always been like this. People did not want to admit it, but Zerg is overtuned, always been.
At some point zerg received only buffs, even though they were already over the top OP and Protoss received hardcore nerfs, like the removal of the HT amulet, Flux Vanes (now back), among other heavy nerfs, without any compensation.
Zerg can´t have strong units with the actual mechanics they have, tech switches are crazy, Zerg economy is crazy, Zerg has the faster and stronger units that can be created much faster, this is total BS.
Ask yourself why all Zerg Pro 3Base gas less play. How does zerg want to play early aggression without falling behind ecco? What does zerg have on hatch? ling rough
lair ? muta (100/100) hydra (need 1-2 upgrades)
all units need upgrades or are behind hive or through need min 6-10 of them. so you need larvae too.
you can play 2base or just 1base but then you have to surprise your opponent. and then you have more of an all in.
thats right. but zerg got nerfed most of the time that happened and then, winrates went back to ~50%
yep, thats the feeling i get. and like allrdy mentioned, the latest data is from november. zerg got nerfed in the beginning of december, so lets wait and see, how the % develop.
lets bring up another point - what does the history even matter? even if zerg was 20 times more favoured - wouldnt that even matter, if the recent win% would be around 50%? i would say yes. so if you say something âbut zerg was 4 times as often favouredâ you basicly imply, that you want terran to be favoured as often as zerg.
tbh, thats the case. and those guys are serral and dark. you can easily argue about the other three zergs, who won something in 2019. reynor is carried by the region lock, rogue and soo never showed up again after their wins in 2019. but each of these 5 are constantly on top lvl for several years now, so theyre even successfull, when zerg isnt favoured. thats called not patch dependent. sure, mappool etc helped, but each of these 5 would have been successfull, even without the current strength of zerg.
thats the thing with winrates. theyre constantly changing. so you wont ever have a perfect 50% (hey, didnt i allrdy said that? kinda feel like im repeating myself). so a winrate between 47 and 53 is totally fine.
why are you that confident to say, that its because maps and not because units? wheres the difference to zerg atm? shouldnt you freak out about those win% too? why are you just yelling over the tvz winrates?
And now Zerg have been nerfed and Terran buffed. Notice how the QQ has not subsided? It really does feel like Terran players are chronic complainers who will not stop until Terran is insanely OP.
Terran players were complaining about Baneling busts in early WOL.
GM population is very weakly correlated with pro player population, especially in so Korea. This happens for 2 reasons. First reason is the skill gap between the top 50 (actual pro players) and the bottom 100 or so of GM (most of which are not pros) is pretty massive. Second reason is in Korea there are many pros who play on alt barcode accounts to conceal their identity and at least try to keep their builds secret. They often stay in Master 1 because they donât play enough games to be promoted, and most of their training they get from playing offline against their team members.
If you were to average it out over larger periods, it actually shrinks to 51/49%. If you hone-in on smaller time-periods you can find fluctuations of 60/40% and higher - all the way up to 100/0% if you shrink it to a game-per-game basis. This variability demonstrates the need to get a large sample to establish a stable average, and/or to filter the variables causing the instability such as (1) varying number of games per period, and (2) skill representation.
There can be periods with more games and periods with fewer. Imagine two periods. [A] with 75/25 for ZvT, and [B] with 49/51. This sounds bad on paper, because there were two periods and in one of them Terran had a 25% win-rate while the other had a 51% win-rate for terran. CLEARLY terran is underpowered! Actually, we have to consider the number of games played. If [A] had 100 games played, and [B] had 10,000 games played, the average win-rate for Terran would have been (100*0.25+10000*0.51) / 10100 = 50.7%. This actually shows terran is overpowered. This is REALLY important because the number of games played can shift RADICALLY when a large event is happens in one period and not the next.
You can filter the instability of the results by accounting for variables that cause instability, such as skill. Ranking algorithms can do this, like ELO and Glicko. If you did filter the results in such a manner, youâd see Terrans overperform and zergs underperform despite the seeming terran underperformance in win-rates. What this means is that if Zerg does have a higher win-rate than terran, this happens as a product of lopsided skill representation (Zerg on average had better players). This is already to be expected because the majority of professionals play Zerg, and the strongest professionals play Zerg. It would be expected to have lopsided skill representation.
We can combine #1 and #2 to create a third type of bias. Results can be biased by number of games played on a per-skill level basis. If a sample has equal numbers of high and low skilled players, the effect they have on a win-rate can be different because there can be more games played by one group or the other. This actually happens a lot in tournaments, because low-skilled players are eliminated thus we have more games with high skilled individuals. If the highest skilled individuals are Zerg, weâd have more games about those players and thus it skews the win-rates in favor of Zerg not because Zerg is strong but because the best players play Zerg.
Conclusion: Win-rates fluctuate too much, and this requires such a large sample that you donât have enough precision to measure balance over small time-periods (such as a new patch). To do that, you basically have no choice except to rely on a ranking algorithm like ELO. Even then itâs VERY difficult to measure balance.
They are both the least played race globally and the least represented currently in GM (a large dip occurred, likely meaning a spike in the GM skill floor).
You can check the site sc2unmaksed and verify everything I said. Iâm not saying GM race distribution has nothing to do with pro player distribution. Iâm just saying that itâs not the most accurate fit in some cases (e.g. Korean GM ladder).
Iâve got your problem right here. Lurker, Baneling, Swarm Host, Brood Lord, and to some extent Infestor. What do they all have in common? They are SIEGE units. I magine if terran had varying forms of the tank including one that could fly. LOL. Iâm not asking much Blizzard just take one or two of these zerg units out of the game. We BEGGED you to take the Swarm Host out in HOTS and you didnt do it. Also the baneling was supposed to be the replacement for the lurker. Now you have both in the game. Itâs kind of like they put campaign zerg into Versus mode. Thatâs why itâs not fair. If Blizzard is going to do that then give terran firebats, wraiths, and goliaths too. And give protoss arbiters, scouts, and dragoons. At least that would even things out a bit.