|
Love this chart, great data !
Could I suggest for one of you future analysis, you correlate the %win of a race in a given matchup with the research of a ceratin upgrade or creation of a certain building (notably, zealot charge, blink, termal lance, ghost academy, greater spire, etc).
Having, for example, the %winrate of a P in PvT with Charge researched and the %winrate without Charge researched could help understand the 'potential' late game imbalances in this matchup (this is by no mean a balance whine as the matchup seems more or less balanced over the whole course of a game).
|
Pretty nice analysis!
I think it all comes down to the defenders advantadge, which is basically nonexistent in both PvP and ZvZ, while it is very strong in TvT.
Here is hoping that Blizz is trying to improve the early game consistency of both P and Z in HotS, for the overall improvement of the game. Things like the no high-ground warp-in and the chronoboost-cannon might help with P, but with Z I have no idea if they're doing anything at all.
|
On March 07 2012 04:04 ZeroTalent wrote:Show nested quote +On March 07 2012 01:07 ChaosTerran wrote: edit: The numbers are actually really good for all matchups except PvP, when the better players wins around 8-9 out of 10 matches on average then the game is extremely skill-based, and luck is not that much of a factor like some people (namely idra) make it out to be. Sure there are all-ins and there is some luck factor, but when the better players wins 9 out of 10 games I think that you can't really argue that sc2 is very luck-based.
edit2: In most sports upsets happen and there is a chance the underdog wins, that's a good thing, as long as it is based on skill and there only is a small luck factor, which is exactly the case in SC2. It's mainly skill but there is a slight chance for an upset with a little bit of luck. You see that in european football all the time, where top teams sometimes lose to "worse" teams, it doesn't happen very often, but just like in SC2 it happens sometimes and makes everything more interesting. It might not even be luck anyway, just a top player or pro player having a bad game and the "scrub" playing extraordinarily well. I think this is right. Though, for the most we're talking about pro level players versus very serious casual players or up-and-coming pros here. It would be like taking the MLB All-Star team and having them play a AAA allstar team. Sure, the AAA team would win a few times (just look at the World Baseball Classic) but it's so rare that it's close to dumb luck. I may try to re-run these numbers using more buckets for players than "pros" and "scrubs" to see what happens when one player is only a slight favorite. Because being an 8-1 favorite is actually a very uninteresting game from a spectators perspective. Even in the NBA, the best team finishes the season at something like 65-17, which is a 4-1 favorite or so. The '97 Bulls were a 7-1 favorite. Great NFL teams can go 15-1 or 14-2, but that's rare; in the typical season the best team is 13-3 or 12-4.
Excellent point.
What definition of "pro" did you actually use when calculating these stats? I mean there are alot of players I could think of where I wouldn't know how to categorize them and if you rank low- mid grandmasters players and silver league players as "scrubs" then there will always be some sort of inaccuracy.
I think the best way for you to sort of re-run these numbers is to look at the leagues, sure leagues don't mean alot, but it is less arbitrary then "scrub", "pro", "semi-pro". You could look at how often it happens that a diamond player beats a master league player or a master league player beats a pro and if there are any pros who lose to some players that aren't high master, etc..
that would be interesting aswell.
edit: Here is what I would do:
Pros Grandmaster Master Diamond Platinum + Gold Silver + Bronze
this is how I would rank people, it's still not accurate, but if you really plan on re-running these numbers I think this would be a better system to use.
|
Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance.
|
On March 07 2012 06:40 ChaosTerran wrote: What definition of "pro" did you actually use when calculating these stats? I mean there are alot of players I could think of where I wouldn't know how to categorize them and if you rank low- mid grandmasters players and silver league players as "scrubs" then there will always be some sort of inaccuracy..
The OP says: "In addition, I strengthened the definition of pro to "any player who playes 7 games in the Ro8 (but ignoring Ro4, Ro2, etc.). In essence this means that during the even day tournaments, you must make the Ro8 at least three times. "
Seems like a perfectly good, measurable, objective criteria given his data set. It seems to go a long ways to identifying how much of a coin flip SC2 is which is what the post is about. Its not about some philosophical argument about the essence of "pro" or "scrub".
Anyway, great OP, well done, and very interesting.
|
On March 07 2012 06:47 BushidoSnipr wrote: Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance.
Honestly, it seems the data shows SC2 is in pretty darn good shape. PvP needs help, but the way people trash SC2, it all should have been much worse.
|
On March 07 2012 07:15 Smackzilla wrote:Show nested quote +On March 07 2012 06:47 BushidoSnipr wrote: Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance. Honestly, it seems the data shows SC2 is in pretty darn good shape. PvP needs help, but the way people trash SC2, it all should have been much worse. looks pretty bad to me, even the 10% in tvt is probably worse than you would see in any broodwar matchup maybe excluding zvz. I hope this will improve with heart of the swarm
|
On March 07 2012 07:21 nttea wrote:Show nested quote +On March 07 2012 07:15 Smackzilla wrote:On March 07 2012 06:47 BushidoSnipr wrote: Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance. Honestly, it seems the data shows SC2 is in pretty darn good shape. PvP needs help, but the way people trash SC2, it all should have been much worse. looks pretty bad to me, even the 10% in tvt is probably worse than you would see in any broodwar matchup maybe excluding zvz. I hope this will improve with heart of the swarm
10% is not very good. 5% would be better for all the match ups, since we are talking about players of drasticly different skill levels. I would like to see the more skilled player win 19 out of 20 games. Blizzard seems to be on that track, with providing more tools to deal with early game pressure. We will have to see when the beta hits and people get to mess with those new units.
|
Still, it is strange that some players can have insane winratios in PvP and ZvZ. Are they lucky? Or do they know more?
|
So your definition of pro is based on how many playhem tournament one plays? This seems like an extremely flawed analysis...
|
Can you also provide the number of data you have for each matchup? Permutations dictate that mirror matchups happen at most half as often as non-mirrors. This will change the error bars quite a bit to the probabilities.
|
On March 07 2012 07:21 nttea wrote:Show nested quote +On March 07 2012 07:15 Smackzilla wrote:On March 07 2012 06:47 BushidoSnipr wrote: Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance. Honestly, it seems the data shows SC2 is in pretty darn good shape. PvP needs help, but the way people trash SC2, it all should have been much worse. looks pretty bad to me, even the 10% in tvt is probably worse than you would see in any broodwar matchup maybe excluding zvz. I hope this will improve with heart of the swarm
A 9-1 favorite is bad? Anyway, this is borderline trolling without real data. Keep in mind, we're talking about a daily open where we compare guys who hit quarterfinals a few times vs. those who don't. It's not Flash vs. a good amateur.
|
On March 07 2012 07:35 ThE_OsToJiY wrote: So your definition of pro is based on how many playhem tournament one plays? This seems like an extremely flawed analysis...
Oh good grief. He simply has a data set where he's comparing players with consistent success vs. those who enter but are not consistently successful. He's not trying to tell you what a "pro" is.
|
Well, now we at least have a solid statistical verification that pvp and zvz are the most volatile matchups.
|
On March 07 2012 07:31 Plansix wrote:Show nested quote +On March 07 2012 07:21 nttea wrote:On March 07 2012 07:15 Smackzilla wrote:On March 07 2012 06:47 BushidoSnipr wrote: Lol this is great, showing the invalidities of each race and how the design needs to be fixed. It's clear that terran's design is sound, whereas someone like me, who doesn't even play protoss has a chance of beating SKMC. The game is still infantile I know, but Blizzard only cares about game balance, not design balance. Honestly, it seems the data shows SC2 is in pretty darn good shape. PvP needs help, but the way people trash SC2, it all should have been much worse. looks pretty bad to me, even the 10% in tvt is probably worse than you would see in any broodwar matchup maybe excluding zvz. I hope this will improve with heart of the swarm 10% is not very good. 5% would be better for all the match ups, since we are talking about players of drasticly different skill levels. I would like to see the more skilled player win 19 out of 20 games. Blizzard seems to be on that track, with providing more tools to deal with early game pressure. We will have to see when the beta hits and people get to mess with those new units.
I don't get the "drastically" different skills" comment. "Scrub" can include players who has made top 16 often but have only broke quarters once. Look at the actual criteria used to bin the players. Heck, you could have DRG play in playhem once, win the whole thing, never play again, and he's a "scrub".
At least that's how I'm interpretting the OP. Did I miss something?
|
|
On March 07 2012 07:59 Smackzilla wrote:Show nested quote +On March 07 2012 07:35 ThE_OsToJiY wrote: So your definition of pro is based on how many playhem tournament one plays? This seems like an extremely flawed analysis... Oh good grief. He simply has a data set where he's comparing players with consistent success vs. those who enter but are not consistently successful. He's not trying to tell you what a "pro" is.
Yea...he's only defining "pro" within the context of his dataset. He has to define it to establish the criteria that he is measuring. Show me a study where the researcher CAN avoid defining their terms. The important thing is that he is clear in his definition. And yes, given that his data is all based on Playhem, of course he is going to define "pro" in relation to the data he is analyzing. If you want to dismiss statistical studies, go for it...I'm always skeptical of statistics because of these built in assumptions. But don't fault the OP for being clear and self-reflexive about his assumptions--about his criteria for "pro."
|
Zerg and toss units are fast thus giving less reaction time to the opponent. Tvt is well balanced in skill cause the units dont bum rush you.
|
On March 07 2012 08:33 agahamsorr0w wrote: Zerg and toss units are fast thus giving less reaction time to the opponent. Tvt is well balanced in skill cause the units dont bum rush you.
Terran has the most allins and the most effective ones too.....I still love how people think zerg is all about rushing because its the exact opposite lol
|
On March 07 2012 08:40 BushidoSnipr wrote:Show nested quote +On March 07 2012 08:33 agahamsorr0w wrote: Zerg and toss units are fast thus giving less reaction time to the opponent. Tvt is well balanced in skill cause the units dont bum rush you. Terran has the most allins and the most effective ones too.....I still love how people think zerg is all about rushing because its the exact opposite lol
I think he means that the units like the zergling (and just morph banelings right outside his base) are fast as in they can cross the map quickly so there isn't alot of time to react. In PvP the rush distance is even lower thanks to warpgates.
|
|
|
|