11:33 The single strongest piece of evidence that Eclipse is not a terran-favored map, is that both Reddit AND Gypsy have claimed that it is. Reddit is nothing more than a place where people exactly like Gypsy go to be wrong together.
Reddit has gotten worse lately too since the api changes. I saw a thread on the front page about that poorly written math problem, you know the '2÷2(3+1)' stupid one that's designed to be ambiguous? And everyone was just arguing about it like it was 2010 facebook! Back in my day reddit used to know that was a trick! No longer. It's functionally indistinguishable from 9gag now
@@andrewferguson6901 It's never been perfect, I always thought it had a very good niche in smaller communities where genuine discussions and information sharing could take place (I've primarily used it for my industry, which is small, and a few smaller gaming communities). But the API changes wrecked whatever modding that was left, disinformation and botting is getting absurd, and the BW/SC subreddit is genuinely one of the most retarded groups I've ever seen in my life. Obviously every social media has it's problems, but whatever quality reddit has is gone.
lol your "strongest piece of evidence" is a stats drawn from 94 games on Eclipse, of which only... 25 are TvP. Meanwhile, eloboard - the Korean site recording modern pro BW games - has 1500 TvP and 1300 TvZ games on Eclipse, of which Terran won 50.3 and 52.5%. This analysis got many points wrong and Artosis just focus on those points to fool you guys.
I’ve always said, anybody can vomit out a bunch of numbers, a great data analyst will find the value in the numbers. Props to this guy for giving us insight on what the numbers mean AND putting it together in such an accessible and PRETTY presentation.
he seems to be furiosly enraged but i guess this has to be the most happy Artosis there has been for a long, long time ... which is still a lot of furios rage but also very happy :D
Why are you here watching him. He's a globalist shill. You need to boycott him. They made it loud n clear which side thry are on. Just absolutely trash
worth pointing out that artys meltdown of rightness showing protoss beat terran 53.3% of the time, same goes for terran beating zerg 53.3% of the time :D
So watching only artosis for my broodwar content, I was told that defilers were unbeatable, but it's only 50-50 in the late game and terran has an overall advantage.
Defilers are friggin OP and very hard to counter because they *almost* completely neutralize tanks with dark swarm which is a core unit in the terran army. Concussive damage is friggin BS and makes firebats almost useless in almost all scenarios outside of roasting zerglings, and they are the only “melee” unit that terrans have. Then you have plague which is basically irradiate but AOE, AND IT WORKS ON MECHANICAL UNITS even without the ability to eat zerglings for energy, defiler has some of the best abilities in the whole game.
That's not his point. People frequently say that while Protoss may have an advantage over Terran at lower skill levels, Terran ends up better at ASL levels. But these stats show even that isn't true, and when you exclude flash it's even more true. Additionally, he never claimed Terran wasn't favored over zerg.
@@abstractdaddy1384 But these stats are flawed. There are only over 400 TvP games at ASL which were played on like 40 different maps. Flip the results of 20 games and you have a totally different story. And guess what? There were seasons when they brought out very bad maps for Terran (ASL5, KSL3) that even Light decided to quit playing Terran for a while, and those are the game responsible for the 53% Protoss win rate in PvT. Go see the stats analyzed from 8 million replays with mmr analysis added, which Artosis totally ignored because of course, it doesn't go with his agenda.
So as it turns out, Terran beats zerg over 50%, zerg beats protoss over 50%, and protoss beats Terran over 50% And the percentages for win rates are bloody close, an incredible job old blizzard did
people should study it. He plays safe, never gambles, pre-build cannons vs drops and invis dudes.. he doesn´t care if he is scouted or not. I think he traps them to a big mid brawl where he outskills them.
@@chrisv384 I agree, stylistically he's like the complete opposite of Mini. Rather than relying on imperfect information he relies purely on positional play, mechanics, and smart engagements. A bit like Best but smarter. I really wish he would come back to ASL as Protoss right now lacks a solid and consistent player.
No margin of error, all races at 49%-51% win rate, P win rate 0.3% higher. Artosis: I KNEW IT!! Also shows T>Z, Z>P, and P>T Artosis: Yeah who cares I don't play zerg
@@Lykyk Hilarious to see the cope when the strongest player in the games history plays Terran, but it's not because Terran is strong! It's because he's just so special! Flash, Boxer, Nada, iloveoov, Light, Fantasy, all just overcoming the odds with the weakest race!
What about that one season of ASL where they rigged it against Flash and made the maps super imba. I know the author removed flash from the statistic analysis to be fair but if flash couldn't win, other terrans probably had like a 10% winrate that season. So that would kinda skews the data a little bit.
"You know what I like, that this information comes out before I die" LOOOOOOL NICE QUOTE WILL REMEMBER ALWAYS Arty your thinking and understanding was way ahead of its time but here we are now lucky all this came out to light while we can still appreciate your big brain thinking
...yeaaaaaaaaaaa, looking at this from a high level, and all the stats taken in agregate, these terran princesses arent at any significant disadvantage, generally speaking, the vast majority of the time, as is true with the other races, game is absurdly well balanced
Game is not well balanced. It's just really hard to play, hard to master, impossible to play perfectly. So people just lose to the game mechanics pretty often, and it's not dependent on a race.
That data suggest to me that ASL should be looking at having several 3P maps in the pool. I might even suggest 3 of 7 maps should be 3P. Sylphid is obviously quite balanced and had a ton of games. Looking back at prior seasons Ascension seemed to look pretty balanced as well. Season 15 and 16 both had had 4 4P, 1 3P and 2 2P.
Artosis saying that a bunch of dumbasses that don't play the game said that 2 player maps are T favored after immediately saying that Gypsy agreed Eclipse was T favored is one hell of an Arty moment
Sub 2000 games out of which 411 were tvps which included ASL 5 games on maps which were specifically made for Protoss to get advantage. Nicely done. Better disregard the 8 million games showing that t>p tho.
me: 0.3 is probably statistically insignificant artosis: lets skip over this bit me: wait a second what was that bit the bit: it is literally statistically insignificant
someone needs to do something like this for SC2...its pretty remarkable how balanced sc1 is. incredible map makers and incredible balancing. this is about as balanced you could possibly get.
No APM measurement? That would be one step towards seeing how much effort goes into a win, which is where a ton of the imbalance complaints come from. Equal winrate doesn't mean so much for balance when one player is putting in significantly more effort and has more knowledge than the other person. That's like saying RPG class A and RPG class B are balanced because they both do 5k DPS, when B is juggling 12 invisible internal cooldowns, has half the HP and mobility, and has triple as many skills to use as A. You can't determine balance just by the endpoints.
Apm is always zerg>terran>toss both bw and sc2. Zerg make unit though larva. It increases carzy amount of apm both games, toss have longest making time. So yeah always top 10 apm player was zerg even flash is not close when he was prime
@@kya6816And more units = more apm required But there is also diff apm required for diff unit comps, zvz muta ling vs zvt ultra ling, zvz much more apm intensive
T>Z>P>T. Objectively speaking - Broodwar's originally balance was close to perfection. Great analysis! It was very interesting to understand how much Flash actually contributes to Terran.. I expected to be a lot, but WOW..
geoff is looking down at dan here and clowning him even though he is right. fucken space marine. love you artosis. im so glad you keep the passion alive. thank you [+]
But… but arty. The data is said to indicate that on any other map than a 2-player map - pvt is Terran favored! You seemed to gloss over that bit for some reason?
Yeah but 2 player is f 67% (which is ridiculously high) and other were basically balanced. Anything under 5% is statistically not important. So 50-55% is pretty much balanced. And I am zerg player Btw.
The TvZ 3 player map is because Terran is closer to both other mains so you some advantage of 2 player maps and if you take your sector you can harass either other main and zergs can't get more than 2 other mains, they can't expand 10 times against terran's 5 like in that AMD map that is heavily zerg favored imo
Not sure about how useful the winrate over game duration charts are. Or at least it’s hard to say that a race is strong at X time because their winrate is high at that time. For example, Protoss can get away with cross-spawn nexus-first against Terran because of how strong they are in the early game, but they don’t win those game right away.
I mean, yes, that's a good point but on another hand you see that for those games that go into late game, just like it was said in the video, it's pretty much ambigous on who gets the advantage in the matchup because of that build - and in the early game you get to see who dies more due to performing nexus, cc or hatch first, and then by reverse engineering the stats seen you can see who gets better advanatge out of getting away with that build more
Artosis seems to have no understanding of statistics and completely ignores the most important conclusion of the paper, which is that the difference in winrate between the races is so incredibly small compared to the sample size that it is "statistically insignificant by any reasonable standards" e.g. the 3 races are as balanced as possible when played at this skill level on these specific maps. I'm pretty sure he actually did realize this though because he says "We'll just skim over this don't want to read that shit" like he knows if he reads that paragraph he will be blown the fuck out. This is like flipping a coin 100 times, getting 51 heads and 49 tails and concluding heads for sure has a higher winrate than tails. That MIGHT be the case for that specific coin / methodology of the person flipping the coin but you'd need a lot more flips and/or a much bigger difference to have any reasonable confidence in that fact.
It's interesting that PvT is the only matchup that remains unbalanced in the late-game. It seems like Terran have to have gained a significant lead from the upgrade push, or protoss will crawl back. In my experience, running a late game economy in this matchup is particularly awkward as Terran have to keep putting pressure while defending 3-4 bases.
LArger maps dimish the value of cheese and early aggression. This is also true of extra spawns because you don't know exactly where to attack and waste time getting there. There is no explanation why adjascent spawns helps terran.
"TvP in general might be slightly P favored but TvP on a 4 player map with adjacent spawns is strongly T-favored" >TvP 4 player map winrate is 53%, when Terran will spawn adjacent to P 66% of the time (Or am I being stupid?) I assume I'm just bad at statistics or misunderstanding the premise considering he's the guy making the charts, but either that doesn't seem very strongly favored or the winrate for TvP on a 4 player map is so obscenely high for Protoss WITH opposite spawns that it almost completely evens out an apparently strongly favored Terran matchup that happens far more often with adjacent spawns.
The winrate for Terran in TvP on 4 players maps with adjacent spawns is 57-43 (+14). With cross spawns, it's 46.5-53.5 (-7). (2x57 + 1x46.5) / 3 = 53.5, very close to 53.33.
Now i want to see flash as protoss an rain as terran.... Make everyone one race to their left then we do stats again lol flash still ontop but those number would be fun too GJ person who made this possible for arty to show me ^.^
I used to be able to go up with the SCV on the Zealot or anything that stood on the ramp if I did something with the Shift key, but I don't remember how to do it anymore. Can you help?
Interesting statistics, but it only includes games played in KSL and ASL (SC Remastered era). I wonder what the stats are since competitive broodwar started in Korea, back when in-house pro gaming houses was a thing.
not being sarcastic. ARTOSIS IS GOD!! Give this man all of the credit. He knows wtf he is talking about. I watch his casts every day and always entertained. This dude is my hero.
Thats 67% Protoss winrate vs Terran on 2p maps is a very significant outlier. Protoss winrates would have a notable drop if they don't play on 2p maps, and it also feels unfair to Terran for such a significant advantage to be given to Protoss on those maps.
On 4P maps, cross spawn vs adjacent is pretty huge too. It's pretty annoying that there are 2 big factors are map pick and RNG before the first mineral is gathered. The data appears to suggest we would find the most balance (across all matchups) if the map pool was tilted towards 3P maps.
We don't know exactly because they also stripped the top P and top Z, and then it was still +2.36%. My guess is it would be about +2.5%. Also Flash was not stripped from any of the individual categories, the author weirdly forgets about it after opening paragraph
@@SolarPlayer yeah I just included it for fun because I had noticed Artosis mentioning this a couple of times. I don't think it's really relevant because Flash is, after all, just another player, even if he happens to be the best one.
Yes - or 2,36pp. The other top players' results only affected the winrate of their own race in this calculation. Actually I just checked what happens if you only remove Flash and it turns out he's also bringing up the Zerg winrate by 0,05pp 😂
@@JackyVSO Oh are you the actual author? props to you for good work Flash is not at all just another player - he's an extreme outlier that warps the result of the questions you're trying to answer. It's totally fine and useful to have his TvT data in, or to have all his data in when you're trying to answer questions like "Is cross spawn better or close spawn?", but as soon as you're comparing races and matchups it's very dubious. It would be like leaving artosis in your dataset in an analysis of all C rank players, and then comcluding that terran is fine because of that. He actually would beat C rankers also with protoss, so the terran data you collect from him is worse than useless
@@SolarPlayer Thanks, glad you enjoyed it! I think it would be difficult to start setting up objective criteria for which players should count in the stats and which shouldn't count because they're too good (or too bad). At least I don't feel that kind of speculation is mine to make; but I guess that's why we have forums like this for discussion.
Is this a joke? A 2% difference on 1000 games is basically nothing. Statistically it's probably well within margin of error. 200 games... PvT wins 4 more times if cross-spawn. Arty is like "Omg I've been vindicated" ... but TvP is heavily favored if adjacent spawns on 4 player map... "Pffft... taking some liberties with your data"
About game length: there being relatively few long games should make it MORE spiky, not less, because there are so few data points. The fact that it's so even is actually pretty interesting.
Hey, I'm the guy that made these stats. Let me clarify those particular graphs: The way it's done is the data was divided into 10 intervals with the length of the intervals designed so that each interval would represent the same amount of games. So on each of the graphs, you can see a bunch of dots, placed right in the middle of their respective interval, which represent the actual data points, each being the average winrate of all games in the matchup that ended within that interval. The curve is then simply extrapolated from those. So each of the vertices on the graphs is based on the same amount of data (give or take a few games). This means that any fluctuations there might be in the super late game are not recorded because they are averaged out as part of the rightmost interval. I could have included data for each minute separately but then some minutes would have very few games or none at all, which would have caused a lot of random spikes on the graphs that I think would have been more confusing than helpful.
@@JackyVSO Thanks for the response. One thing you could do for this would be to use a spline to generate a local moving average, that would create the cleanest and most consistent graph that requires the least amount of decisionmaking on your part. I would also love to see measures of significance here. Confidence bands on the plots, and report estimated ranges rather than just the number.
@@michaellopate4969 I agree that would be the most correct way of presenting the data - and absolutely required if it was a scientific publication. The only reason I haven't done so is that it would be a lot of work and most people wouldn't be interested. Btw sorry if I was "talking down" to you before, you obviously know a lot about statistics based on your second comment :-)
@@JackyVSO I don't think it's so important as to actually presenting the data in the most scientifically accurate way, though that would be nice, but it affects the framing around the findings. Depending on uncertainty measures, it might go from "race X has a small advantage under these conditions" to "balance is not statistically different from zero." Especially in the race advantage in different times of the game, that can't be too many games in each bin! But the percentages here are displayed as if they are as accurate as the overall race win percentage. Alternatively, you could frame it more around how many games out of 100 are "decided" by these factors, or better yet, how often do these results meaningfully influence the outcome of a best-of-2, 5 or 7? That could be pretty easily tested through simulation. I know I'm asking for a lot here. If you don't have the time, software or inclination to do it, if you send me the raw data I can run through analysis and get back to you.
@@michaellopate4969 I'm not making any promises about changing the page but I'll gladly send over the database for you to analyze on your own. I'll also send the Python scripts I used to extract the insights if you like.
A 25 year old game that hasnt had a balance patch since it's release and has 50% winrate across all races, that's just genius game design, once in a millenia type of shit.
Interesting but the sample size is too small. I don’t question what artosis is saying. But some of these could be skewed a little or even a lot. Like if you flip a coin 100 times it could come tails 90 times. Unlikely. But reasonably possible. Too bad we don’t have like 100k games
absolutely insane the WR between all 3 races is that fucking close. I gotta give it to the folks that made BW. They ABSOLUTELY nailed it.
what about the mirror matchups?
To be fair, the mapmakers have to be included in that gratitude. BW is insanely unbalanced without proper maps keeping the balance in check..
Is it? If the win rates were not balanced, then everyone would just play the strong race.
@@solstice871 agreed
Statements uttered by the mentally deranged
We are in the greatest era of pre-AI Artosis
AlphaTosis
I should begin AI-tossis
this is indeed a wonderful moment in the Artosis beta.
11:33 The single strongest piece of evidence that Eclipse is not a terran-favored map, is that both Reddit AND Gypsy have claimed that it is.
Reddit is nothing more than a place where people exactly like Gypsy go to be wrong together.
Reddit is filled with this. Thanks for realizing.
Reddit is a place where people go to be wrong.
Reddit has gotten worse lately too since the api changes. I saw a thread on the front page about that poorly written math problem, you know the '2÷2(3+1)' stupid one that's designed to be ambiguous? And everyone was just arguing about it like it was 2010 facebook!
Back in my day reddit used to know that was a trick! No longer. It's functionally indistinguishable from 9gag now
@@andrewferguson6901 It's never been perfect, I always thought it had a very good niche in smaller communities where genuine discussions and information sharing could take place (I've primarily used it for my industry, which is small, and a few smaller gaming communities). But the API changes wrecked whatever modding that was left, disinformation and botting is getting absurd, and the BW/SC subreddit is genuinely one of the most retarded groups I've ever seen in my life.
Obviously every social media has it's problems, but whatever quality reddit has is gone.
lol your "strongest piece of evidence" is a stats drawn from 94 games on Eclipse, of which only... 25 are TvP.
Meanwhile, eloboard - the Korean site recording modern pro BW games - has 1500 TvP and 1300 TvZ games on Eclipse, of which Terran won 50.3 and 52.5%.
This analysis got many points wrong and Artosis just focus on those points to fool you guys.
Artosis being demonstrably correct is the best kind of Artosis
Its so gratifying to see our lord and savior educate us while owning us
I've only been wrong once in my life. It was that time that I thought I was wrong but turns out I was right.
Rain is just ridiculous good.
I’ve always said, anybody can vomit out a bunch of numbers, a great data analyst will find the value in the numbers.
Props to this guy for giving us insight on what the numbers mean AND putting it together in such an accessible and PRETTY presentation.
he seems to be furiosly enraged but i guess this has to be the most happy Artosis there has been for a long, long time ...
which is still a lot of furios rage but also very happy :D
The happiest since he got to be 2400
Why are you here watching him. He's a globalist shill. You need to boycott him. They made it loud n clear which side thry are on. Just absolutely trash
at this point I don't even know which one is a persona - this or The Normal Calm Caster Artosis.
@@trezenxcalm artosis just shows up when casting
😂
I love how professional Artosis can be during ASL and how off the rails he becomes while streaming 😂 Im glad he was alive to confirm his theories 😂
worth pointing out that artys meltdown of rightness showing protoss beat terran 53.3% of the time, same goes for terran beating zerg 53.3% of the time :D
He glanced over that one real quick
He caste dark swarm and jumped into a Nydus on that one.
It’s with flash tho
yeah but that but that 53.3 was with flash
Arty casually ignoring that Terran smashes Zerg by the same percentage that Protoss smashes Terran lol 5:30
So watching only artosis for my broodwar content, I was told that defilers were unbeatable, but it's only 50-50 in the late game and terran has an overall advantage.
can't make defilers with no gas
late game zerg starts losing geysers
😂
FIREBATS UNBEATABLE, STRONGEST UNIT
Defilers are friggin OP and very hard to counter because they *almost* completely neutralize tanks with dark swarm which is a core unit in the terran army.
Concussive damage is friggin BS and makes firebats almost useless in almost all scenarios outside of roasting zerglings, and they are the only “melee” unit that terrans have.
Then you have plague which is basically irradiate but AOE, AND IT WORKS ON MECHANICAL UNITS
even without the ability to eat zerglings for energy, defiler has some of the best abilities in the whole game.
@@Larsgman I agree that it's way easier for protoss to play against it than terran.
Protoss is 0.1% better than Terran on one stat: "PROTOSS IS OP!!! LOOK!!"
Terran is 0.1% better than Zerg on one stat: .......silence......
That's not his point. People frequently say that while Protoss may have an advantage over Terran at lower skill levels, Terran ends up better at ASL levels. But these stats show even that isn't true, and when you exclude flash it's even more true.
Additionally, he never claimed Terran wasn't favored over zerg.
@@abstractdaddy1384 But these stats are flawed. There are only over 400 TvP games at ASL which were played on like 40 different maps. Flip the results of 20 games and you have a totally different story. And guess what? There were seasons when they brought out very bad maps for Terran (ASL5, KSL3) that even Light decided to quit playing Terran for a while, and those are the game responsible for the 53% Protoss win rate in PvT.
Go see the stats analyzed from 8 million replays with mmr analysis added, which Artosis totally ignored because of course, it doesn't go with his agenda.
So as it turns out, Terran beats zerg over 50%, zerg beats protoss over 50%, and protoss beats Terran over 50%
And the percentages for win rates are bloody close, an incredible job old blizzard did
That rotational balance is why 3 is a pretty ideal number of races for an RTS. Get to 4, and what happens if a race has 2 good or bad match ups?
😂😂and you cant blame the matchup. Its all about player skill and matchup understanding... so people who cry all the time end up stuck in 2k!😅😅😅
@@Zraknulyou get stuff like Orc vs Undead in wc3/tft
Rain's PvP win rate is absolutely ridiculous, and is even more impressive when you realize he has the highest number of PvPs played.
Rain is on another level, best PvP by far
people should study it. He plays safe, never gambles, pre-build cannons vs drops and invis dudes.. he doesn´t care if he is scouted or not. I think he traps them to a big mid brawl where he outskills them.
@@chrisv384 I agree, stylistically he's like the complete opposite of Mini. Rather than relying on imperfect information he relies purely on positional play, mechanics, and smart engagements. A bit like Best but smarter. I really wish he would come back to ASL as Protoss right now lacks a solid and consistent player.
No margin of error, all races at 49%-51% win rate, P win rate 0.3% higher.
Artosis: I KNEW IT!!
Also shows T>Z, Z>P, and P>T
Artosis: Yeah who cares I don't play zerg
Only if you ignore that Flash stands out, son.
@@Lykykstfu we had a TvT asl15 finals, flash is not responsible for Terran winrate
@@Lykyk Hilarious to see the cope when the strongest player in the games history plays Terran, but it's not because Terran is strong! It's because he's just so special! Flash, Boxer, Nada, iloveoov, Light, Fantasy, all just overcoming the odds with the weakest race!
@@user-sk6nm1wr2j You're basically just typed out the Terrans are the best at the top argument.
What about that one season of ASL where they rigged it against Flash and made the maps super imba. I know the author removed flash from the statistic analysis to be fair but if flash couldn't win, other terrans probably had like a 10% winrate that season. So that would kinda skews the data a little bit.
"You know what I like, that this information comes out before I die" LOOOOOOL NICE QUOTE WILL REMEMBER ALWAYS Arty your thinking and understanding was way ahead of its time but here we are now lucky all this came out to light while we can still appreciate your big brain thinking
CROSS SPAWNS BAD FOR TERRAN?! WHO SAID THAT?! 😂😂😂
...yeaaaaaaaaaaa, looking at this from a high level, and all the stats taken in agregate, these terran princesses arent at any significant disadvantage, generally speaking, the vast majority of the time, as is true with the other races, game is absurdly well balanced
Game is not well balanced. It's just really hard to play, hard to master, impossible to play perfectly. So people just lose to the game mechanics pretty often, and it's not dependent on a race.
@@Askhat08 That's the dumbest thing I've ever heard. They balance it for human players.
Thank God FBI didn't got it's hands on this important study. The world needs to see this
At least this time...
That data suggest to me that ASL should be looking at having several 3P maps in the pool. I might even suggest 3 of 7 maps should be 3P.
Sylphid is obviously quite balanced and had a ton of games. Looking back at prior seasons Ascension seemed to look pretty balanced as well.
Season 15 and 16 both had had 4 4P, 1 3P and 2 2P.
It looks like my PvT bias was not counting 2 player maps enough. 67% win rate is crazy.
The virgin 2000 game effort post vs the Chad believing whatever Artosis said to be law
Would be cool if you and Tasteless looked at this together as some sort of ASL extra next season =)
Artosis: "I was right all along"
*proceeds to skip over all the parts in the video that show he was wrong
All this hate at protoss being "stronger" when there has never been a protoss bonjwa. I don't wanna hear it. We don't have a protoss era in brood war.
seeing Flash's stats displayed like this is genuinely jaw dropping holy shit
Artosis saying that a bunch of dumbasses that don't play the game said that 2 player maps are T favored after immediately saying that Gypsy agreed Eclipse was T favored is one hell of an Arty moment
Someone could play the game and still be wrong 😂
The Old Man Yells at Cloud is strong in this one.
"TvP winrates over time"
WHAAAAAAAAT IIIIIIIIS THAAAAAAAT, DOOOOOOO YOOOOOUUUUUUUUU SSSSSSSSSEEEEEEEEE HOOOOOOOOWWWWWW OP PROOOOOOOTTTTOOOOSSSS IIIIIISSSS?????
"TvZ winrates over time"
*silently scrolls past it*
Sub 2000 games out of which 411 were tvps which included ASL 5 games on maps which were specifically made for Protoss to get advantage. Nicely done. Better disregard the 8 million games showing that t>p tho.
me: 0.3 is probably statistically insignificant
artosis: lets skip over this bit
me: wait a second what was that bit
the bit: it is literally statistically insignificant
As a Terran player. I always knew you were right.
Artosis is so gracious in his vindication 😛
Great stats though.
Everyone knows that Flash is insane, but Last's stats are almost as crazy.
Guys in the chat. We will never live this down. IN RT WE TRUST
So Terran beats Zerg just as bad as Protoss beats Terran but you don’t see Zergs crying like Artosis lol
Larva: Aside from Flash and Last I feel like I can beat any Terran player. It's something about those players, not about race as a whole.
someone needs to do something like this for SC2...its pretty remarkable how balanced sc1 is. incredible map makers and incredible balancing. this is about as balanced you could possibly get.
No APM measurement? That would be one step towards seeing how much effort goes into a win, which is where a ton of the imbalance complaints come from.
Equal winrate doesn't mean so much for balance when one player is putting in significantly more effort and has more knowledge than the other person.
That's like saying RPG class A and RPG class B are balanced because they both do 5k DPS, when B is juggling 12 invisible internal cooldowns, has half the HP and mobility, and has triple as many skills to use as A. You can't determine balance just by the endpoints.
Apm is always zerg>terran>toss both bw and sc2.
Zerg make unit though larva. It increases carzy amount of apm both games, toss have longest making time.
So yeah always top 10 apm player was zerg even flash is not close when he was prime
+ highest apm terran player was ty when he was 10 yo kid 😂
He could be one of best bw terran if league live longer
@@kya6816And more units = more apm required
But there is also diff apm required for diff unit comps, zvz muta ling vs zvt ultra ling, zvz much more apm intensive
Sane ones actually knew you were right all along
And props to the stats maker
T>Z>P>T.
Objectively speaking - Broodwar's originally balance was close to perfection.
Great analysis! It was very interesting to understand how much Flash actually contributes to Terran.. I expected to be a lot, but WOW..
Arty really chewing the digital scenery here
2000 episode cast celebration when?
Lovely, interesting stats, with a great sense of humor! SOLID work!
I go crazy for these stats!
Quality content
geoff is looking down at dan here and clowning him even though he is right. fucken space marine. love you artosis. im so glad you keep the passion alive. thank you [+]
Oh man those win rate deltas over time are insane!
Some chatter: Artosis bible should be called "Cross Spawn".
But… but arty. The data is said to indicate that on any other map than a 2-player map - pvt is Terran favored! You seemed to gloss over that bit for some reason?
Yeah but 2 player is f 67% (which is ridiculously high) and other were basically balanced. Anything under 5% is statistically not important. So 50-55% is pretty much balanced. And I am zerg player Btw.
I fucking knew it!!!!!
After all these years. VINDICATION!
The TvZ 3 player map is because Terran is closer to both other mains so you some advantage of 2 player maps and if you take your sector you can harass either other main and zergs can't get more than 2 other mains, they can't expand 10 times against terran's 5 like in that AMD map that is heavily zerg favored imo
excellent point
Awesome video. Great material for the anti-protoss propaganda.
Artosis really showing he doesn't really understand how statistics works here.
Not sure about how useful the winrate over game duration charts are. Or at least it’s hard to say that a race is strong at X time because their winrate is high at that time. For example, Protoss can get away with cross-spawn nexus-first against Terran because of how strong they are in the early game, but they don’t win those game right away.
I mean, yes, that's a good point but on another hand you see that for those games that go into late game, just like it was said in the video, it's pretty much ambigous on who gets the advantage in the matchup because of that build - and in the early game you get to see who dies more due to performing nexus, cc or hatch first, and then by reverse engineering the stats seen you can see who gets better advanatge out of getting away with that build more
Artosis seems to have no understanding of statistics and completely ignores the most important conclusion of the paper, which is that the difference in winrate between the races is so incredibly small compared to the sample size that it is "statistically insignificant by any reasonable standards"
e.g. the 3 races are as balanced as possible when played at this skill level on these specific maps.
I'm pretty sure he actually did realize this though because he says "We'll just skim over this don't want to read that shit" like he knows if he reads that paragraph he will be blown the fuck out. This is like flipping a coin 100 times, getting 51 heads and 49 tails and concluding heads for sure has a higher winrate than tails. That MIGHT be the case for that specific coin / methodology of the person flipping the coin but you'd need a lot more flips and/or a much bigger difference to have any reasonable confidence in that fact.
The pie chart is stunningly even
artosis_ we know you were correct (at least me) but we been trolling you cuz we love you
It's interesting that PvT is the only matchup that remains unbalanced in the late-game. It seems like Terran have to have gained a significant lead from the upgrade push, or protoss will crawl back. In my experience, running a late game economy in this matchup is particularly awkward as Terran have to keep putting pressure while defending 3-4 bases.
Rain PvP 🤯🤯🤯
I think TvZ 3 player gives zerg strength because they usually get the 2nd main and Terran does not have a default 2nd main as with a 4 player map.
Taking a few wings of liberty with your data
Those were some fantastic stats
LArger maps dimish the value of cheese and early aggression. This is also true of extra spawns because you don't know exactly where to attack and waste time getting there. There is no explanation why adjascent spawns helps terran.
Cant exactly remember who maybe firebathero, he explained that it's weird when the tank's range is diagonal.
Because the Protoss can't go cross spawn Nexus first - they're close enough to actually get punished
"TvP in general might be slightly P favored but TvP on a 4 player map with adjacent spawns is strongly T-favored"
>TvP 4 player map winrate is 53%, when Terran will spawn adjacent to P 66% of the time (Or am I being stupid?)
I assume I'm just bad at statistics or misunderstanding the premise considering he's the guy making the charts, but either that doesn't seem very strongly favored or the winrate for TvP on a 4 player map is so obscenely high for Protoss WITH opposite spawns that it almost completely evens out an apparently strongly favored Terran matchup that happens far more often with adjacent spawns.
The winrate for Terran in TvP on 4 players maps with adjacent spawns is 57-43 (+14). With cross spawns, it's 46.5-53.5 (-7). (2x57 + 1x46.5) / 3 = 53.5, very close to 53.33.
Now i want to see flash as protoss an rain as terran....
Make everyone one race to their left then we do stats again lol flash still ontop but those number would be fun too GJ person who made this possible for arty to show me ^.^
"ash world" is actually char, the zerg homeworld
Makes sense
Zerglings attack faster
27:50 why tf did this make me laugh so hard 🤣🤣🤣🤣💀💀💀💀
The best kind of correct.
This is moment of glory, and he deserved it ^^
The only universal truth is terran players will always find a way to complain....even after seeing they beat zerg just as hard as protoss beats terran
Artosis gloating is so fun
I used to be able to go up with the SCV on the Zealot or anything that stood on the ramp if I did something with the Shift key, but I don't remember how to do it anymore. Can you help?
Thats a bug and its illegal.
Interesting statistics, but it only includes games played in KSL and ASL (SC Remastered era). I wonder what the stats are since competitive broodwar started in Korea, back when in-house pro gaming houses was a thing.
not being sarcastic. ARTOSIS IS GOD!! Give this man all of the credit. He knows wtf he is talking about. I watch his casts every day and always entertained. This dude is my hero.
Thats 67% Protoss winrate vs Terran on 2p maps is a very significant outlier. Protoss winrates would have a notable drop if they don't play on 2p maps, and it also feels unfair to Terran for such a significant advantage to be given to Protoss on those maps.
On 4P maps, cross spawn vs adjacent is pretty huge too. It's pretty annoying that there are 2 big factors are map pick and RNG before the first mineral is gathered.
The data appears to suggest we would find the most balance (across all matchups) if the map pool was tilted towards 3P maps.
that only shows Flash is GOAT, nothing more, 0.xx % is nothing, can change with next tournament - lol
I still play in the way of an Eddy this makes me cool now
Does that mean Flash is bringing up the Terran win rate by 3% by himself?
We don't know exactly because they also stripped the top P and top Z, and then it was still +2.36%. My guess is it would be about +2.5%. Also Flash was not stripped from any of the individual categories, the author weirdly forgets about it after opening paragraph
@@SolarPlayer yeah I just included it for fun because I had noticed Artosis mentioning this a couple of times. I don't think it's really relevant because Flash is, after all, just another player, even if he happens to be the best one.
Yes - or 2,36pp. The other top players' results only affected the winrate of their own race in this calculation. Actually I just checked what happens if you only remove Flash and it turns out he's also bringing up the Zerg winrate by 0,05pp 😂
@@JackyVSO Oh are you the actual author? props to you for good work
Flash is not at all just another player - he's an extreme outlier that warps the result of the questions you're trying to answer. It's totally fine and useful to have his TvT data in, or to have all his data in when you're trying to answer questions like "Is cross spawn better or close spawn?", but as soon as you're comparing races and matchups it's very dubious. It would be like leaving artosis in your dataset in an analysis of all C rank players, and then comcluding that terran is fine because of that. He actually would beat C rankers also with protoss, so the terran data you collect from him is worse than useless
@@SolarPlayer Thanks, glad you enjoyed it! I think it would be difficult to start setting up objective criteria for which players should count in the stats and which shouldn't count because they're too good (or too bad). At least I don't feel that kind of speculation is mine to make; but I guess that's why we have forums like this for discussion.
I am so glad I watched this.
Is this a joke? A 2% difference on 1000 games is basically nothing. Statistically it's probably well within margin of error.
200 games... PvT wins 4 more times if cross-spawn. Arty is like "Omg I've been vindicated"
... but TvP is heavily favored if adjacent spawns on 4 player map... "Pffft... taking some liberties with your data"
First couple of ASL seasons were before remasted came out
About game length: there being relatively few long games should make it MORE spiky, not less, because there are so few data points. The fact that it's so even is actually pretty interesting.
Hey, I'm the guy that made these stats. Let me clarify those particular graphs: The way it's done is the data was divided into 10 intervals with the length of the intervals designed so that each interval would represent the same amount of games. So on each of the graphs, you can see a bunch of dots, placed right in the middle of their respective interval, which represent the actual data points, each being the average winrate of all games in the matchup that ended within that interval. The curve is then simply extrapolated from those. So each of the vertices on the graphs is based on the same amount of data (give or take a few games). This means that any fluctuations there might be in the super late game are not recorded because they are averaged out as part of the rightmost interval. I could have included data for each minute separately but then some minutes would have very few games or none at all, which would have caused a lot of random spikes on the graphs that I think would have been more confusing than helpful.
@@JackyVSO Thanks for the response. One thing you could do for this would be to use a spline to generate a local moving average, that would create the cleanest and most consistent graph that requires the least amount of decisionmaking on your part. I would also love to see measures of significance here. Confidence bands on the plots, and report estimated ranges rather than just the number.
@@michaellopate4969 I agree that would be the most correct way of presenting the data - and absolutely required if it was a scientific publication. The only reason I haven't done so is that it would be a lot of work and most people wouldn't be interested.
Btw sorry if I was "talking down" to you before, you obviously know a lot about statistics based on your second comment :-)
@@JackyVSO I don't think it's so important as to actually presenting the data in the most scientifically accurate way, though that would be nice, but it affects the framing around the findings. Depending on uncertainty measures, it might go from "race X has a small advantage under these conditions" to "balance is not statistically different from zero." Especially in the race advantage in different times of the game, that can't be too many games in each bin! But the percentages here are displayed as if they are as accurate as the overall race win percentage.
Alternatively, you could frame it more around how many games out of 100 are "decided" by these factors, or better yet, how often do these results meaningfully influence the outcome of a best-of-2, 5 or 7? That could be pretty easily tested through simulation.
I know I'm asking for a lot here. If you don't have the time, software or inclination to do it, if you send me the raw data I can run through analysis and get back to you.
@@michaellopate4969 I'm not making any promises about changing the page but I'll gladly send over the database for you to analyze on your own. I'll also send the Python scripts I used to extract the insights if you like.
Never doubt Papa Tosis...
didn't the ladder statistics show terran at the highest winrate ?
Best video on the whole entire internet
3:35 says it all hahaha
I love how terran ego has no way to account for Flash's success so they always asspull some reason why he doesn't count.
It's more that he's so good he skews the statistics
if anything he counts too much
kind of like serral in current sc2
flash could also play off races at a top level because of his understanding of the game, flash is an anomaly
I love you RT don't change
If flash is so much better, removing the top 1 player from each race is making your case better! So definitely not a fuckup.
Rain is a fucking god man . Wish he would play more. 2nd best player ever in bw
Can it be that the map pool was alterd to stop flash from winning and its not the races but maps?
jungle is BGH, got it :D
This was fun
A 25 year old game that hasnt had a balance patch since it's release and has 50% winrate across all races, that's just genius game design, once in a millenia type of shit.
It actually had balance patches.
@@danielfcastro true no patches for 23 years then though
Wait is it April 1st already. Was I sleeping, had I slept?
Interesting but the sample size is too small. I don’t question what artosis is saying. But some of these could be skewed a little or even a lot.
Like if you flip a coin 100 times it could come tails 90 times. Unlikely. But reasonably possible.
Too bad we don’t have like 100k games
how'd banelings do? Oh they won? Amazing. Who would've thunk? The poor children.
oh yeah BW...lets fix SC2 now.
God damnit. You're too funny bro!
Interestingly is that the win rates would be almost dead even if all maps were 3 player maps
In 4 player maps Terran is actually favored in tvp. Lol.