|
Vatican City State733 Posts
On February 11 2012 22:47 ToastieNL wrote:
Lol at all those people stating that Terran loses early game by going for a risky all- in, and stays in the game to lose after the 20 minute mark to scew the graphs.
Anybody thought of WHY so many terrans go for those all- ins.
Hint; lategame is retarded.
Or their early game is extremely strong? There are any number of possible explanations and I doubt you could come up with quantitative data to back your claim
|
On February 11 2012 23:18 RDaneelOlivaw wrote:Show nested quote +On February 11 2012 22:47 ToastieNL wrote:
Lol at all those people stating that Terran loses early game by going for a risky all- in, and stays in the game to lose after the 20 minute mark to scew the graphs.
Anybody thought of WHY so many terrans go for those all- ins.
Hint; lategame is retarded. Or their early game is extremely strong? There are any number of possible explanations and I doubt you could come up with quantitative data to back your claim
Indeed there are different explanations and without more evidence we would be rational to accept most simple of them as most probable - Terrans wins more early and lose more later because their early game is strong and late game weak, any explanation postulating more then that without evidence backing them up would be less likely to be true.
|
On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason.
I extracted and recompiled the information in the OP for personal use, but you might find it good to know too.
games played and the respective winrates
(non-mirrors, taken by inspection, approximations, error small and almost negligable)
syntax:
games total: the total amount of games played during thedescribed timeframe, wins and losses
winrate: the winrate of the first mentioned race (for TvP, its the winrate for T, the P winrate is 100-T winrate)
games won: the amount of games won by the first mentioned race during the timeframe.
TvP:
+ Show Spoiler +
PvZ
+ Show Spoiler +
ZvT
+ Show Spoiler +
|
On February 11 2012 23:30 Roblin wrote:Show nested quote +On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason. I extracted and recompiled the information in the OP for personal use, but you might find it good to know too. games played and the respective winrates (non-mirrors, taken by inspection, approximations, error small and almost negligable) syntax: games total: the total amount of games played during thedescribed timeframe, wins and losses winrate: the winrate of the first mentioned race (for TvP, its the winrate for T, the P winrate is 100-T winrate) games won: the amount of games won by the first mentioned race during the timeframe. TvP: + Show Spoiler +PvZ + Show Spoiler +ZvT + Show Spoiler +
It should say TvZ and ZvP.
|
On February 11 2012 23:30 Roblin wrote:Show nested quote +On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason. I extracted and recompiled the information in the OP for personal use, but you might find it good to know too.
... the OP has a link to a spreadsheet with the raw data.
|
Dat negative correlation between TvZ
|
altho the numbers can be flawed. i wonder (highly doubt) that blizzard takes into account these things when balancing the game. I wonder if their internal data balances the game this way. its a sort of "relative balance". I doubt it though.
|
All the data really represents is the current metagame. Z almost always plays for the late game and T/P almost always does some sort of timing attack/push, and it's quite clearly reflected in the data.
If all races had aimed for late game every game then I suspect the results would be quite different.
|
On February 11 2012 23:30 Roblin wrote:Show nested quote +On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason. I extracted and recompiled the information in the OP for personal use, but you might find it good to know too. games played and the respective winrates (non-mirrors, taken by inspection, approximations, error small and almost negligable) syntax: games total: the total amount of games played during thedescribed timeframe, wins and losses winrate: the winrate of the first mentioned race (for TvP, its the winrate for T, the P winrate is 100-T winrate) games won: the amount of games won by the first mentioned race during the timeframe. TvP: + Show Spoiler +PvZ + Show Spoiler +ZvT + Show Spoiler +
man i love that tvz statistic for after the 15-20min marker. 35%! just brilliant, anyone want to make bets on how much lower it drops after the snipe nerf? i bet it will drop below 20%
|
Vatican City State733 Posts
On February 11 2012 23:29 Remi wrote:Show nested quote +On February 11 2012 23:18 RDaneelOlivaw wrote:On February 11 2012 22:47 ToastieNL wrote:
Lol at all those people stating that Terran loses early game by going for a risky all- in, and stays in the game to lose after the 20 minute mark to scew the graphs.
Anybody thought of WHY so many terrans go for those all- ins.
Hint; lategame is retarded. Or their early game is extremely strong? There are any number of possible explanations and I doubt you could come up with quantitative data to back your claim Indeed there are different explanations and without more evidence we would be rational to accept most simple of them as most probable - Terrans wins more early and lose more later because their early game is strong and late game weak, any explanation postulating more then that without evidence backing them up would be less likely to be true.
That's incredibly weak reasoning in this situation. This is a case where dismissing unknown details as unnecessary obscuring bits of information could very easily lead to incorrect conclusions.
|
I think the low win rates late game for tvz and high win rates in the beginning are linked. Terran has such a strong early game that most terrans can just win then and there if they apply alot of pressure early on. In other words alot of terrans currently disregard even trying to go for a macro game so if they don't kill the zerg early on they're in a really bad spot because they somewhat disregarded their economy to get a game ending army. If you watch some games by thorzain for example that guy sometimes maxes out as fast as the zerg does almost by going to three bases really fast. The only thing a terran really has to worry about doing this is some early 1base roach allin.
|
On February 12 2012 12:20 CptCutter wrote:Show nested quote +On February 11 2012 23:30 Roblin wrote:On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason. I extracted and recompiled the information in the OP for personal use, but you might find it good to know too. games played and the respective winrates (non-mirrors, taken by inspection, approximations, error small and almost negligable) syntax: games total: the total amount of games played during thedescribed timeframe, wins and losses winrate: the winrate of the first mentioned race (for TvP, its the winrate for T, the P winrate is 100-T winrate) games won: the amount of games won by the first mentioned race during the timeframe. TvP: + Show Spoiler +PvZ + Show Spoiler +ZvT + Show Spoiler + man i love that tvz statistic for after the 15-20min marker. 35%! just brilliant, anyone want to make bets on how much lower it drops after the snipe nerf? i bet it will drop below 20%
Unless I'm missing something, the statistics apply to the first race mentioned in the matchup (which would be Z in ZvT), meaning that Z is the one with the 35% winrate 20+ minutes in? I'm confused.
|
I would say that Zerg v Non-Zerg match ups take the longest because Zerg has to play the economic game (i.e. I'm getting more bases and more economy to beat you later) and if Protoss or Terran plays turtle defense, Zerg's option is usually to take the map than to try to bust in.
Also, a 6-pool is never going to result in a loss for Zerg in under 5 minutes unless they give up immediately. On a 6-pool Zerg gets to the base with the initial 6 Zerglings around 3:00 (longer on some maps) and the ensuing micro usually takes a while unless Zerg completely screws up. Even if the end result is the Protoss holds off the rush, unless the Zerg player leaves immediately it's going to result in a win for the Protoss longer than 5 minutes.
|
On February 11 2012 21:33 HaXXspetten wrote:Show nested quote +On February 11 2012 19:18 dde wrote:On February 11 2012 19:15 Cereb wrote:
Interesting.
One thing I'd like to add and put in question is that to me Terrans also seem to stay in games waaaay after the've already ,ore or less lost whereas zergs seem to throw in the (too) early GG more.
I guess this would makes sense seeing as Terran has pretty good defensive abilities and are therefore hard to kill whereas if you lose too much as Zerg you do not have anything to fall back on at your base.
lol so biased Very biased... but there is an element of truth to it actually >.>
Biased towards what? You think it's better to leave early than to stay in the game and defend? I'm just addressing that one race makes more sense to play one way over the other :/
This was just ment as an explanation for a PART of the graphs. It's not like I said that this was the only reason -_-;;
|
I would love to see this applied to the 10,000+ playhem pack.
|
It would be more useful with sample numbers.
|
On February 11 2012 23:43 foxmeep wrote:Show nested quote +On February 11 2012 23:30 Roblin wrote:On February 11 2012 21:34 david0925 wrote:
I really want to see the number of games involved in the game length winning percentage in each data point.
I'm not arguing for or against Terran late game, but as far as I know Terrans try to finish their games very early. So if the late game count is drastically lower than mid game it might be at least part of the reason. I extracted and recompiled the information in the OP for personal use, but you might find it good to know too. games played and the respective winrates (non-mirrors, taken by inspection, approximations, error small and almost negligable) syntax: games total: the total amount of games played during thedescribed timeframe, wins and losses winrate: the winrate of the first mentioned race (for TvP, its the winrate for T, the P winrate is 100-T winrate) games won: the amount of games won by the first mentioned race during the timeframe. TvP: + Show Spoiler +PvZ + Show Spoiler +ZvT + Show Spoiler + It should say TvZ and ZvP.
this is true, it was a mistake by me, the real stuff is the following:
TvP:
+ Show Spoiler +
ZvP
+ Show Spoiler +
TvZ
+ Show Spoiler +
edit: I would very much like the OP to extend the TvZ chart such that we can see the following timeframes:
0-5
5-10
10-15
15-20
20-25
25-30
30-35
35-40
40+
|
i think these stats illustrate an interesting aspect of the current meta-game. a lot of people complain about certain all-in timing attacks in some match-ups, but these attacks may just be a result of the need to end the game within a specific time period. if a timing attack is not strong enough (not all-in: often these timings are very precise and only work if executed perfectly and with enough force) and there is a transition to late game, it may just come to a point of no return where the game statistically favors one race over the other. in these cases, an all-in timing would be favored if not required to increase the chances of success in a certain match up on a certain map. yeah, it might seem like a better game if everyone played a straight-up match but what is the point of doing a middle-of-the-road build if it is easily defensible and leads to a late-game auto loss?
all skill being equal, a smart player would want to end the game in a certain timing and the other player would want to survive during that time. i would say don't blame the player blame the game.
|
Mirror match game length is interesting. All mirror matches have a low frequency of games lasting 20+ minutes. I think each of you should look at that information and draw a conclusion.
|
|
|
|
|
|
|
|
EPT
