EPTSo I suggest that map pools should be focused/centered around the Zerg where the list should include a zerg-favoured map (against Terran) such as Aztec (not Neo Aztec), and a Protoss favoured map (against Zerg) such as Sparkle to even out the odds.
Does Tesagi Exist? - Page 7
| Forum Index > BW General |
|
William Blue
42 Posts
So I suggest that map pools should be focused/centered around the Zerg where the list should include a zerg-favoured map (against Terran) such as Aztec (not Neo Aztec), and a Protoss favoured map (against Zerg) such as Sparkle to even out the odds. | ||
|
TT1
Canada9968 Posts
On January 10 2019 12:27 William Blue wrote: In my opinion, the imbalances lie on the Zerg rather than Terran. Zerg has a low ZvT winrate and a high ZvP winrate, whereas TvP is balanced (around 50-51% average according to the analysis.) So I suggest that map pools should be focused/centered around the Zerg where the list should include a zerg-favoured map (against Terran) such as Aztec (not Neo Aztec), and a Protoss favoured map (against Zerg) such as Sparkle to even out the odds. right..  | ||
|
DropBear
Australia4261 Posts
Zerg basic combat unit is melee. Protoss basic combat unit is melee. Terran basic combat unit is ranged. A skilled player operating ranged units should beat a skilled player operating melee units in many situations. Also a strong Terran army is far less resource intensive than a strong zerg/toss army. Terrans are able to take less bases and mineral only expos while Protoss/Zergs cannot and have to spread wider to have enough gas to compete. It is far easier to protect a small area than a large area (or several small areas). | ||
|
Luddite
United States2315 Posts
To pick an extreme example, on Fastest Protoss has a clear, obvious advantage. Obviously that's not a "pro" map, but it does at least show that the Protoss units aren't inherently crippled. But a typical pro macro-style map limits their ability to do shuttle raids on workers. Looking at these stats (excluding flash and last): http://sponbbang.com/race/?month1=2017-05&month2=2019-01&bj=이영호, 김성현 You can see the map effect really clearly. Just TvZ win rates: All maps combined (42361 games): 53.5% Circuit Breaker (12837 games): 54.9% Fighting Spirit( 9476 games): 56.2% Gladiator (2981 games): 53.1% Transistor (1706 games): 46.1% Crossing Fields 1189 games): 44.5% The rest that they list have under 1k games played, but show the same sort of variance between maps. So the answer is pretty simple: if you want Terran to win more, sponsor games on Circuit Breaker and Fighting Spirit. If you want them to win less, sponsor games on Transistor and Crossing Fields. The overall imbalance is just there there have been a lot more games on Circuit Breaker and Fighting Spirit than any other map. I don't think Terran is inherently better. But, as long as Circuit Breaker and Fighting Spirit are the "standard" maps (both to play on, and for map makers to copy their design) then Terran will have an advantage. | ||
|
Dazed.
Canada3301 Posts
Look at all the map features we DONT use because of terran -Cliffs -small maps [t>>>z, hell even big maps are used to help give p some strength in pvt] -lots of maps have unbuildable middles or other locations to limit turret crawling etc -one of those ASL maps recently even nerfed the gas at the natural third location in a tvp, which would really only impact terran, showing korean map makers are trying to weaken the terran turtle even further | ||
|
Luddite
United States2315 Posts
On January 10 2019 14:42 Dazed. wrote: if maps are used to correct balance, reference to map stats isnt answering the question of balance but talking around it. I think its clear if we are to discuss the issue that terran has the most versatile set of options, the best player has traditionally been terran, that says something to me. Terran is a race with a wide variety of options which means they tend to take the initiatve in dominating the meta [i.e have the most bonjwas] but balance in broodwar is a feature of maps, so maps are often tilted against terran to make up for this inherent top heavy strength. Look at all the map features we DONT use because of terran -Cliffs -small maps [t>>>z, hell even big maps are used to help give p some strength in pvt] -lots of maps have unbuildable middles or other locations to limit turret crawling etc -one of those ASL maps recently even nerfed the gas at the natural third location in a tvp, which would really only impact terran, showing korean map makers are trying to weaken the terran turtle even further Well, you can also come up with map features that would disadvantage terrans: -island maps -multiple gasses in the main -unbuildable/unlandable high ground behind minerals I guess if they patched the race balance then we might see a lot of maps that look really crazy, but it's hard to speculate. I don't think it's talking around the balance issue to focus on the maps that we actually use and have data from, rather than hypothetical maps that don't exist or never get used. | ||
|
RWLabs
Korea (South)273 Posts
On January 10 2019 14:21 Luddite wrote: You have to look at the maps, to do this kind of analysis. The maps are just as important as the players. To pick an extreme example, on Fastest Protoss has a clear, obvious advantage. Obviously that's not a "pro" map, but it does at least show that the Protoss units aren't inherently crippled. But a typical pro macro-style map limits their ability to do shuttle raids on workers. Looking at these stats (excluding flash and last): http://sponbbang.com/race/?month1=2017-05&month2=2019-01&bj=이영호, 김성현 You can see the map effect really clearly. Just TvZ win rates: All maps combined (42361 games): 53.5% Circuit Breaker (12837 games): 54.9% Fighting Spirit( 9476 games): 56.2% Gladiator (2981 games): 53.1% Transistor (1706 games): 46.1% Crossing Fields 1189 games): 44.5% The rest that they list have under 1k games played, but show the same sort of variance between maps. So the answer is pretty simple: if you want Terran to win more, sponsor games on Circuit Breaker and Fighting Spirit. If you want them to win less, sponsor games on Transistor and Crossing Fields. The overall imbalance is just there there have been a lot more games on Circuit Breaker and Fighting Spirit than any other map. I don't think Terran is inherently better. But, as long as Circuit Breaker and Fighting Spirit are the "standard" maps (both to play on, and for map makers to copy their design) then Terran will have an advantage. I think you're talking around the issue for a different reason. This may account for the success of Terrans in Sponbbangs (though I'd argue BJs have less incentive to win than progamers- just look at Calm), but what about the overall historical Terran dominance during Kespa days? If you want the numbers I made a post on the second page, but to chalk up the relative success of Terrans to maps alone seems unlikely. Mapmakers have little reason to intentionally create Terran favoured maps, and in fact may shift map designs to help an underperforming race out; like in Savior's dominance the island map Desert Fox was thrown in to help Protosses (though Bisu won without it). Over the 12 years we had progamers, we saw the cycling of hundreds of maps and horrifically unbalanced maps were usually removed quickly. Taking all this into account and sticking to the "Terrans aren't OP the maps used favour Terran" explanation, it seems like the only way to account for Terran dominance is to say that Terrans are dominant on most maps- which is just tesagi with extra steps. Or maybe maps aren't the only reason (if one at all) that Terrans were so dominant. For full disclosure, I'm a Pregi/Tossregi man myself. | ||
|
MarcoJ
Germany146 Posts
This is very difficult and goes way deeper than looking at winrates. There are just to many factors that can influence the data in lead you to a wrong conclusion. | ||
|
RWLabs
Korea (South)273 Posts
On January 10 2019 16:57 MarcoJ wrote: I think to find a proper answer its far more complex than how we are looking at it. You cant just look at the overall winrates and discuss shifts in the 5% region and not take the players into account. [...] This is very difficult and goes way deeper than looking at winrates. There are just to many factors that can influence the data in lead you to a wrong conclusion. That's exactly the reason we take twelve years of data instead of a single year or tournament. The hundreds of maps, hundreds of players, and tens of thousands of games that comprise the sample size reduces the impact of extraneous variables because statistics is a still thing. I don't talk about post-Kespa data if I can help it because unlike progamers, BJs are incentivized to get star balloons and not necessarily win (if anything, people like Calm can actually earn more by losing and reacting), and the player pool is far smaller. Dealing with unknown variables is par for the course if you're planning to do any analytical study. Would you say what you've said about any another study? "Well this heart medication reduced blood pressure by 20% in 47500 participants out of 50000 in comparison to the placebo, but there are too many variables so we can't make a conclusion." Yeah there are extraneous variables in any study. That's why you address it and discuss its significance, not discredit the conclusion outright. Obviously it is impossible to quantify the exact skill of a player (I'd think that's exactly what winrates are but whatever) and compare it against the exact impact of maps. This is what's called the nirvana fallacy, and also why proving anything does not exist in science. To simplify: my argument can boil down to: - If the three races were equal in skill ceiling and average performance we would see similar rates of success for whatever metric is chosen to measure it, with some variation expected (H0). - We can assume a similar induction of skilled players in all races. - Most progamers can be expected to have practiced hard, and there is nothing to suggest one race worked harder than any other (Sea, Hiya, Stork being rare cases where little effort yielded good results). - Maps can be expected to have had an impact, but likely not significant enough to account for the historical dominance for multiple reasons, but most easily seen by the fact that Terran average winrate was never last place despite long periods of non-Terran favoured maps. - Historically using almost any metric of success, Terrans excelled. - Even after the Kespa days Terrans still dominate, if you care about post-Kespa rankings. From this there are a few possible reasons for the Terran dominance: - We know TvZ favours Terran, ZvP favours Zerg, but PvT only slightly favours Protoss, which would obviously advantage Terrans. In other words, Protoss is crap. - Players with the best knowledge of the game just happened to pick Terran (ie: first assumption is wrong). Possible, but what does it say when players who understand the game best all chose the same race? Also if anything shouldn't Protoss see the most success as it's the most popular one and has the largest pool of players to draw from? - Terran is stronger than the other races. The first explanation makes the most sense to me as following that logic Terrans would be the most successful, Zergs would be moderately successful, and Protosses would be the least successful- and that's exactly what we see. If you have any concrete data or argument to suggest why I'm wrong, I'm all ears. | ||
|
vOdToasT
Sweden2870 Posts
At very high levels, Protoss is slightly weaker than Zerg and Terran. We should solve this with maps. Blue Storm is an example of a Protoss favoured PvZ map, and La Mancha is an example of a Protoss favoured PvT map. I'm not saying that we should make it imbalanced in the other direction, but I am saying that we should adopt maps that are slightly more Protoss favoured than the norm is today (No more FS and CB). What would a map that is good for Protoss in both PvZ and PvT look like? The third base for Terran would need to be difficult to take, or be a mineral only, or both. Taking a fourth base should also be made difficult for mech. For PvZ, we should avoid free main bases behind natural expos. Make expansions very open and hard to defend with sunken spore lurker. If Terran only gets a mineral only, but Protoss gets a full expansion, that's good for Protoss. An additional possibility is to go with very open, hard to defend expansions, but to also give players an expansion with two geysers that is not viable for mech as a third base, but which is viable for P and Z. Giving this one base a choke point might be better for P in PvZ. Edit: Statistics from various time periods, as well as strategic analysis, prove that my claims about La Mancha and Blue Storm are correct. Thus, it is possible to make maps that are better for Protoss. We should admit that Protoss is slightly weak on FS and CB, and move on to more Protoss favoured (thus more balanced) maps. I am not advocating for going so far that it actually becomes imbalanced. I only advocate moving slightly in the direction of what is Protoss favoured. Other examples of Protoss favoured maps include Outlier and Central Plains. These maps, especially Central Plains, went too far and were actually significantly imbalanced in favour of Protoss in both match ups, but they prove that balance is all about maps. That Protoss is slightly underpowered in the normal ecosystem is a result of maps. | ||
|
Dazed.
Canada3301 Posts
On January 10 2019 15:12 Luddite wrote: Island maps favour terran over zerg, and are fairly only slightly disadvantaged against p i think, whereas in a semi island map t is favoured in all matchups decisively. Anyway, sure, there are maps that are anti terran, but I think if you go through map history you will find a lot more terran imbalanced maps. Frankly, boxers entire era were just maps that were impossible to stop any decent terran on, especially before the muta glitch was found out. Maybe all these terran maps were a matter of contrivance- terran was popular, they rigged the leagues a little bit over the years. Or, terran is just strong on a lot of map concepts because its the swiss army knife of broodwar, which is why its imbalanced. Well, you can also come up with map features that would disadvantage terrans: -island maps -multiple gasses in the main -unbuildable/unlandable high ground behind minerals I guess if they patched the race balance then we might see a lot of maps that look really crazy, but it's hard to speculate. I don't think it's talking around the balance issue to focus on the maps that we actually use and have data from, rather than hypothetical maps that don't exist or never get used. | ||
|
Alpha-NP-
United States1242 Posts
| ||
|
kidcrash
United States616 Posts
On January 10 2019 21:50 vOdToasT wrote: At lower levels, Protoss is the strongest, but that doesn't matter; just git gud. At very high levels, Protoss is slightly weaker than Zerg and Terran. We should solve this with maps. Blue Storm is an example of a Protoss favoured PvZ map, and La Mancha is an example of a Protoss favoured PvT map. I'm not saying that we should make it imbalanced in the other direction, but I am saying that we should adopt maps that are slightly more Protoss favoured than the norm is today (No more FS and CB). What would a map that is good for Protoss in both PvZ and PvT look like? The third base for Terran would need to be difficult to take, or be a mineral only, or both. Taking a fourth base should also be made difficult for mech. For PvZ, we should avoid free main bases behind natural expos. Make expansions very open and hard to defend with sunken spore lurker. If Terran only gets a mineral only, but Protoss gets a full expansion, that's good for Protoss. An additional possibility is to go with very open, hard to defend expansions, but to also give players an expansion with two geysers that is not viable for mech as a third base, but which is viable for P and Z. Giving this one base a choke point might be better for P in PvZ. Edit: Statistics from various time periods, as well as strategic analysis, prove that my claims about La Mancha and Blue Storm are correct. Thus, it is possible to make maps that are better for Protoss. We should admit that Protoss is slightly weak on FS and CB, and move on to more Protoss favoured (thus more balanced) maps. I am not advocating for going so far that it actually becomes imbalanced. I only advocate moving slightly in the direction of what is Protoss favoured. Other examples of Protoss favoured maps include Outlier and Central Plains. These maps, especially Central Plains, went too far and were actually significantly imbalanced in favour of Protoss in both match ups, but they prove that balance is all about maps. That Protoss is slightly underpowered in the normal ecosystem is a result of maps. Tau cross was always statistically a very balanced map as well. I'd like to see them run a tournament with La Mancha, Tau Cross and Blue Storm. I do think most balance problems can be ironed out with a decent map pool. | ||
|
Jealous
10076 Posts
On January 10 2019 19:58 RWLabs wrote: That's exactly the reason we take twelve years of data instead of a single year or tournament. And you would be making a huge mistake right off the bat if you do so. Are 15 minute 1 hatch ZvP games from 1999 relevant at all to balance in 2019? | ||
|
vOdToasT
Sweden2870 Posts
On January 11 2019 01:37 Alpha-NP- wrote: What’s a map that is good for Z in ZvT? Blue Storm? I can't think of any imbalanced map in favour of Z in ZvT, but there are plenty of maps at 50% (Not CB or FS). | ||
|
Luddite
United States2315 Posts
On January 10 2019 16:15 RWLabs wrote: I think you're talking around the issue for a different reason. This may account for the success of Terrans in Sponbbangs (though I'd argue BJs have less incentive to win than progamers- just look at Calm), but what about the overall historical Terran dominance during Kespa days? If you want the numbers I made a post on the second page, but to chalk up the relative success of Terrans to maps alone seems unlikely. Mapmakers have little reason to intentionally create Terran favoured maps, and in fact may shift map designs to help an underperforming race out; like in Savior's dominance the island map Desert Fox was thrown in to help Protosses (though Bisu won without it). Over the 12 years we had progamers, we saw the cycling of hundreds of maps and horrifically unbalanced maps were usually removed quickly. Taking all this into account and sticking to the "Terrans aren't OP the maps used favour Terran" explanation, it seems like the only way to account for Terran dominance is to say that Terrans are dominant on most maps- which is just tesagi with extra steps. Or maybe maps aren't the only reason (if one at all) that Terrans were so dominant. For full disclosure, I'm a Pregi/Tossregi man myself. I read your post from page 2. It was interesting, but I'd like to see more detailed stats from the Kespa days. Do you have any stats that break it down by map? Also like Jealous said, you can kinda ignore the earliest years of Kespa when everything was so new. (although Boxer being such a fan favorite from that era might have caused some lingering bias in favor of Terran) You said "Mapmakers have little reason to intentionally create Terran favoured maps, and in fact may shift map designs to help an underperforming race out". I agree they don't want to let any one race dominate, but I wonder if they do have an incentive to help Terran a little? Kespa wanted to make money, so they needed maps that would give entertaining games and pull in fans. That usually means either weird gimmick maps that only last one season, or macro maps where players can quickly build up to max supply and slug it out all over the map. So we get maps like Python/Andromeda/Fighting Spirit/Circuit Breaker, basically the same map being played forever. (edit- i guess Andromeda actually has a losing record for Terrans, but it was also played the least and is the most unusual of those maps) The hardcore fans can geek out over how good Jaedong is at zergling rushes, but most casual fans aren't interested. So we get maps that it hard to do early zergling (or zealot) aggression, and it makes it too easy for Terran to get their perfect 3/3 200 supply steamroller. There might be some conflict here between what makes for good TV, and what makes for a well-balanced esport. To test this, I wonder if we could group up the maps by type? I think Terran has an advantage on maps that are 4 players, symmetric, easy to take a third and no obvious gimmicks. It's not so much that they have an advantage over hundreds of different maps, it's this one particular map type. It started with Lost Temple and it never stopped. I'll bet if you divided the maps between "4 player macro maps" and everything else, you'd see a big difference in race balance. | ||
|
Sr18
Netherlands1141 Posts
| ||
|
SilentchiLL
Germany1405 Posts
If we remove the best player by far from every race Protoss would lose Bisu and I think one of the guys here who said that without Bisu we might have actually gotten a balance patch wasn't too far off the mark. On January 11 2019 04:31 Sr18 wrote: Here is a thought. If we need to remove the data from the earliest years of professional starcraft because the level of play back then wasn't developed enough, how do we know that the current level of play is? As sports go, starcraft is still very young. Look at chess, that game has been played for much longer and the level of play is still rising. Who's to say that in starcraft the level of play can't reach much higher levels. At which of these levels are we determining whether or not the game is balanced? Because it's been 20 years and Starcraft isn't chess, chess is perfectly balanced in nearly all aspects because the players have the same "units" and Broodwar might be great but it will not last centuries. To be honest the comparison is terrible and you should feel terrible for making it in the first place. | ||
|
Dazed.
Canada3301 Posts
Like @ the guy mentioning maps which would help protoss out, im totally in favour of that in terms of the pro circuit. But the type of maps he was describing are kind of....boring maps, and we are already playing on fairly bland maps, as well. Balance through maps might not be good enough anymore. To test this, I wonder if we could group up the maps by type? I think Terran has an advantage on maps that are 4 players, symmetric, easy to take a third and no obvious gimmicks. It's not so much that they have an advantage over hundreds of different maps, it's this one particular map type. It started with Lost Temple and it never stopped. I'll bet if you divided the maps between "4 player macro maps" and everything else, you'd see a big difference in race balance. If you constrained it one step further and kept the 4 player macro maps to be on the slightly smaller end of the scale [like fs compared to electric circuit] i think you would see the tilt favour terran all the more. Easily splittable macro maps with small to medium size are the best for terran imo. | ||
|
Anc13nt
1557 Posts
On January 11 2019 04:31 Sr18 wrote: Here is a thought. If we need to remove the data from the earliest years of professional starcraft because the level of play back then wasn't developed enough, how do we know that the current level of play is? As sports go, starcraft is still very young. Look at chess, that game has been played for much longer and the level of play is still rising. Who's to say that in starcraft the level of play can't reach much higher levels. At which of these levels are we determining whether or not the game is balanced? tbh with all the pros going to military soon and getting older, unfortunately I think the peak level of starcraft play may already be behind us. Perhaps one could look at data after 2007 ( which I guess some would consider this near the beginning of the "modern" era of broodwar), in which zerg won 11 out of 26 starleagues, terran won 9 out of 26 and protoss won 6 out of 26. I'm not talking about current broodwar since I don't have much info regarding it but I do think based on those stats alone, tesagi during the days of pro broodwar is a bit overhyped while tossregi seems pretty convincing, even if it's pretty minor imbalance in my opinion. Also, I wonder what the korean pros think about tesagi and protoss. | ||
| ||
