|
On May 23 2014 08:00 zergnewb wrote:
It is not about if they are able to tell the risk, it is about if they want to take the risk. And then how good of a player they are would just affect how good they are at getting around those worse odds.
If a decision is favorable for you, why wouldn't you take it? You can make the argument of minimizing variance, to get away from all the hypotheticals, it would mean that a challenger player does not go all in at lv1 vs. a bronze, even though the challenger player can win, and instead would rather outfarm and gain an advantage that way. But in the long run the second "style" is incorrect.
|
On May 23 2014 08:00 zulu_nation8 wrote:Show nested quote +On May 23 2014 07:54 xes wrote:On May 23 2014 07:31 zulu_nation8 wrote:On May 23 2014 07:26 Goumindong wrote:
Your argument is circular; "its not risky because he did it and he is a pro so its not risky".
Professionals and Amateurs can have different risk tolerances with regards to play. Some will make plays which are more risky, some will not make those plays. We are talking about risk, we are talking about these plays going south with some non-zero probability whether that is in terms of CS, towers, or kills. Clearly it must be the case that some risk must be involved somewhere, if it were not then lanes wouldn't be determined at level 2 they would be determined at champion select and the game would not have a potentially varying outcome.
The people who take more risks, we call "more aggressive". There is probability in every decision obviously because it's a game of incomplete information. The ones who look for an edge in favorable situations are better players. The ones who don't take those favorable risks are not as good. In poker would you fold your 51/49 hand because you wanna minimize variance? You obviously look for an edge everywhere but you may judge incorrectly or the edge isn't there. You guys are making it seem like there always a bunch of decisions which lead to different outcomes but are all equally good. Do you feel that at the highest level of play, decision making can be reduced to the "online poker" model where the only parameters are easily estimated, or more like live poker where there are estimations based on your estimation of the other actor's estimations? Or do you feel that live poker itself can be reduced to the online poker model, since the online poker model when playing extended hands against the same table already factors in "playstyle." I think in general you are correct that there is a correct play to squeeze out the edge if you can asses it and it is cut and dry, but if decision making is based on a stochastic model where the parameters are also probabilistic, then you very much still have a sliding scale of risk tolerance that determines play. Your counterargument seems to be that the better player will have the more accurate estimation and thus there was a correct and incorrect play independent of variance and thus risk tolerance, but in models like this there is inherent error where sometimes the error of the estimation is greater than the estimation itself. The second The second paragraph, please explain more, I'm not good at stats, what's an example of that?
For statistics in the real world, there are a lot of times where the notion "expected value" (i.e. "value at risk" lolol) is less meaningful than textbooks say because there is uncertainty in both the probability of something happening, uncertainty in the model used to derive the probability of something happening, and uncertainty in the valuation of the outcome (and the parameters in the model of deriving valuation).
An example would be if you were playing blackjack and you didn't know how many decks were being cycled and also instead of showing you the house cards the dealer told you and may or may not be lying.
The armchair example for league might be in your calculations on what to do bot lane there is a "Jungler" parameter, but you only have a rough idea where their jungler is and if the jungler was playing perfectly (lets assume here that perfect play is a given thing for the jungler regardless of the rest of the example) he would be X but maybe he's not playing perfectly which would ruin your plans.
However, this is a totally uninformed example. It could be possible that while theoretically this is a problem, but in practice these sorts of uncertainties are almost never present in League and that a solid correct/incorrect decision can be reached all the time. I feel that you are arguing
1) Casters use "aggressive" so much its meaningless
2) "Aggressive" itself is meaningless because if you played the game as a turn based strategy game with full map vision and knowledge of skill levels there is always a correct/incorrect play which is when "aggression" is a good or bad play
3) In an actual game when you don't have perfect information, the nature of League as an artificial game means your risks in decision making are so limited that it might as well be situation (2)
|
The commentators do not know what they are talking about. What a surprise.
Remember the majority of players are silver or lower and don't really know anything either.
|
So basically there are two arguments here as far as I can tell;
That a player should avoid certain situations, and do, favorable situations I'm assuming, to minimize variance; aka they don't want games to be decided at lv2, etc.
To that, I'm arguing no one passes up favorable situations or plays passive for its own sake, they only pass up a trade or all-in because they judge it to be unfavorable.
This debate I'm 99% sure I'm right, I don't wanna dwell on it further because it's stupid to try to argue against it.
The second argument is people may judge a situation to be favorable but then actually have it be unfavorable, in the long run that is. Because of errors in input, etc. I would argue that this judgement depends on skill, and is not a relevant factor in deciding whether or not to take the risk, because if the situation is judged to be favorable, everyone would take it regardless.
And there's a third, kind of dumb argument, that I hope no one is making; which is that when someone like vasilli goes all in lv1 and loses he's actually knowingly making an unfavorable decision because he plays reginaldaggressivestyle. The counterargument would be that in the long run, he judges that particular trade as favorable. The short term outcome does not matter. This applies to all league decisions.
|
Judging situations isn't so cut and dry. There are often too many factors involved, especially if they involve multiple players. What skill shots are going to land? How will everyone else involved react? Exactly how much damage output does everyone have, how much can everyone take, and how much damage will the minions do? Yes, of course the more skilled player is better at judging situations, but people aren't robots. Nobody can calculate everything, especially when there are unknown variables involved.
|
On May 23 2014 08:13 xes wrote:Show nested quote +On May 23 2014 08:00 zulu_nation8 wrote:On May 23 2014 07:54 xes wrote:On May 23 2014 07:31 zulu_nation8 wrote:On May 23 2014 07:26 Goumindong wrote:
Your argument is circular; "its not risky because he did it and he is a pro so its not risky".
Professionals and Amateurs can have different risk tolerances with regards to play. Some will make plays which are more risky, some will not make those plays. We are talking about risk, we are talking about these plays going south with some non-zero probability whether that is in terms of CS, towers, or kills. Clearly it must be the case that some risk must be involved somewhere, if it were not then lanes wouldn't be determined at level 2 they would be determined at champion select and the game would not have a potentially varying outcome.
The people who take more risks, we call "more aggressive". There is probability in every decision obviously because it's a game of incomplete information. The ones who look for an edge in favorable situations are better players. The ones who don't take those favorable risks are not as good. In poker would you fold your 51/49 hand because you wanna minimize variance? You obviously look for an edge everywhere but you may judge incorrectly or the edge isn't there. You guys are making it seem like there always a bunch of decisions which lead to different outcomes but are all equally good. Do you feel that at the highest level of play, decision making can be reduced to the "online poker" model where the only parameters are easily estimated, or more like live poker where there are estimations based on your estimation of the other actor's estimations? Or do you feel that live poker itself can be reduced to the online poker model, since the online poker model when playing extended hands against the same table already factors in "playstyle." I think in general you are correct that there is a correct play to squeeze out the edge if you can asses it and it is cut and dry, but if decision making is based on a stochastic model where the parameters are also probabilistic, then you very much still have a sliding scale of risk tolerance that determines play. Your counterargument seems to be that the better player will have the more accurate estimation and thus there was a correct and incorrect play independent of variance and thus risk tolerance, but in models like this there is inherent error where sometimes the error of the estimation is greater than the estimation itself. The second The second paragraph, please explain more, I'm not good at stats, what's an example of that? For statistics in the real world, there are a lot of times where the notion "expected value" (i.e. "value at risk" lolol) is less meaningful than textbooks say because there is uncertainty in both the probability of something happening, uncertainty in the model used to derive the probability of something happening, and uncertainty in the valuation of the outcome (and the parameters in the model of deriving valuation). An example would be if you were playing blackjack and you didn't know how many decks were being cycled and also instead of showing you the house cards the dealer told you and may or may not be lying. The armchair example for league might be in your calculations on what to do bot lane there is a "Jungler" parameter, but you only have a rough idea where their jungler is and if the jungler was playing perfectly (lets assume here that perfect play is a given thing for the jungler regardless of the rest of the example) he would be X but maybe he's not playing perfectly which would ruin your plans. However, this is a totally uninformed example. It could be possible that while theoretically this is a problem, but in practice these sorts of uncertainties are almost never present in League and that a solid correct/incorrect decision can be reached all the time. I feel that you are arguing 1) Casters use "aggressive" so much its meaningless 2) "Aggressive" itself is meaningless because if you played the game as a turn based strategy game with full map vision and knowledge of skill levels there is always a correct/incorrect play which is when "aggression" is a good or bad play 3) In an actual game when you don't have perfect information, the nature of League as an artificial game means your risks in decision making are so limited that it might as well be situation (2)
I think I understand your example but I don' think there's any situation where there's so much uncertainty that probability doesn't matter, or if that's what you're suggesting.
For example with the jungler positioning, if say junglers have 0 CD TPs and can be anywhere at any given time, then it would mean no one should trade ever, and in that case none of the bot 2v2 trade probabilities matter because there's a huge unknown factor that you can't account for.
In reality, the enemy jungler can be correctly assessed to be at a certain half of the map nearly all the time. I'm trying to think of when there's an unknown that's so big that you it messes up everything. Baron crits may be one, or the biggest one I can think of. People often misjudge its damage and misjudge their own ability to kill it.
The first two arguments, yes, I'm arguing for those.
The third, at very high level of play, pretty much, not always but waaaaaaaaaaaay more often than people think is the case.
|
On May 23 2014 08:20 GolemMadness wrote:
Judging situations isn't so cut and dry. There are often too many factors involved, especially if they involve multiple players. What skill shots are going to land? How will everyone else involved react? Exactly how much damage output does everyone have, how much can everyone take, and how much damage will the minions do? Yes, of course the more skilled player is better at judging situations, but people aren't robots. Nobody can calculate everything, especially when there are unknown variables involved.
People can calculate -- and when I say calculate I mean make decisions based on instincts and experience, aka being a good player -- to an accurate enough degree that they feel confident taking any risks at all. If so much is uncertain then the game would truly be random, or at least the majority of decisions, and none of us assumes that is the case. For the sake of argument we have to all assume that when someone takes a risk, they feel like it's a favorable one in the long run, and, that he is able to estimate the parameters with enough accuracy compared to the "true," replay parameters, which are still estimates but are close enough that it doesn't matter.
To go back to the original example, Vasilii gets baited and caught a lot because he takes those marginal outcomes, so that he will fail a lot but should, and does succeed more. It's a sign of skill, that he recognizes and capitalizes on the small edges, rather than a display of playstyle.
I admit this isn't simple (it should be though) for commentators or anyone who haven't played a lot of games to understand. But you don't have to know any stats or math to arrive at the same conclusion. In BW I used to think "playing it safe" is often the right thing to do, what is actually happening is I'm just passing up on opportunities because I can't recognize them. I would imagine that's how people interpret the "passivity" in pro games, like literally projecting their own badness onto others and making narratives out of it.
|
Suppose I am playing Katarina vs an Ahri in the mid lane. The Ahri has only 300 health left and has no flash or ignite, so I know for sure that if I get a full EQWR combo on her I will kill her 100% in less than 1 second.
But, I also know that if she lands an E on me in less than 1 second after I E in, I will be dead because I only got 300 health left so she can kill me in an E(QWR) combo.
Do I go in or not?
|
I think one of the biggest factors into how convoluted decision making is, and how hard it is to assess 'correctness' is the idea that you as a player have to assume the other player is indeed attempting to do their best to win, whether that is what their actions seem to represent or not. You see someone flash forward, and you instantly have to assess whether this is a mistake, or them representing something terrible for you that for whatever reason you had not considered. This is where I think the poker analogy comes into play. The fact that a bluff can ever be said to be 'a good bluff' is representative enough of the idea that the 'wrong' play can actually be the 'right' play if represented properly, but it is still a risk because at its core it is the 'wrong' play.
You're talking about taking 60/40's and 80/20s, but the nature of that math is that sometimes the 20/80 is actually correct. An 'aggressive' player may fish for the 20/80, perhaps even knowing it could be a losing play, because the game is played against other human beings that think and react, and you affect how they think and react with your actions. Forcing any action is the only way possible to force the wrong action.
I can't argue that in a utopic environment it is impossible to refute your statement that 'the better player makes the correct play more often.' I will say though, that I think the ability to judge that correct play with hindsight is both unfair(because alternative play would fundamentally change every action thereafter), and essentially impossible to do with accuracy(partly because of the previously noted convolution).
There will always be debate about what the correct play is in all competitions between people, except maybe in a game like chess(I don't follow it).
|
If you're challenger and he's gold, meaning he won't be able to land the E most of the time and you may even flash or dodge it some of the time, it's like a 80/20 for you so obv do it. If you're both bronze then it's like a 60/40 because he's bronze and won't hit the charm most of the time but you may still fuck it up but your micro is easier. If you're bronze and he's challenger it becomes 30/70, etc, etc. You'd have to decide for yourself
|
On May 23 2014 08:42 red_ wrote:
I think one of the biggest factors into how convoluted decision making is, and how hard it is to assess 'correctness' is the idea that you as a player have to assume the other player is indeed attempting to do their best to win, whether that is what their actions seem to represent or not. You see someone flash forward, and you instantly have to assess whether this is a mistake, or them representing something terrible for you that for whatever reason you had not considered. This is where I think the poker analogy comes into play. The fact that a bluff can ever be said to be 'a good bluff' is representative enough of the idea that the 'wrong' play can actually be the 'right' play if represented properly, but it is still a risk because at its core it is the 'wrong' play.
You're talking about taking 60/40's and 80/20s, but the nature of that math is that sometimes the 20/80 is actually correct. An 'aggressive' player may fish for the 20/80, perhaps even knowing it could be a losing play, because the game is played against other human beings that think and react, and you effect how they think and react with your actions. Forcing any action is the only way possible to force the wrong action.
I can't refute that in a utopic environment it is impossible to refute your statement that 'the better player makes the correct play more often.' I will say though, that I think the ability to judge that correct play with hindsight is both unfair(because alternative play would fundamentally change every action thereafter), and essentially impossible to do with accuracy(partly because of the previously noted convolution).
There will always be debate about what the correct play is in all competitions between people, except maybe in a game like chess(I don't follow it).
In the 1029382138129 games of LoL I've played I can't think of one example where after I've seen the replay that I'm still not sure if it was the correct or incorrect decision. Can you think of an example in a pro game where people argue over whether it was the right decision or not? I'm interested to see. When I'm reflecting in the middle of the game I never think about the outcome, because it's results-oriented. It's about making the correct decision the majority of the time regardless of the outcome. If I flash for a kill expecting my mid to follow up and he doesn't, and I die; it may mean it was a bad flash or it may mean it's the right play given a better midlaner. By the same logic the same decision can be incorrect at lower levels but correct at higher levels, but in both cases, there is a correct and incorrect decision.
The probabilities are very rough statistical representations of instinctive decisions in LoL. The very fact that there are good players and bad players, and that there are good players who remain good consistently, supports the fact that this game is not random, and more importantly that good players can estimate well enough true parameters and all the necessary factors to consider in any given situation. Good players make accurate estimates, and hence correct decisions. This shouldn't be an argument.
|
On May 23 2014 08:50 zulu_nation8 wrote:Show nested quote +On May 23 2014 08:42 red_ wrote:
I think one of the biggest factors into how convoluted decision making is, and how hard it is to assess 'correctness' is the idea that you as a player have to assume the other player is indeed attempting to do their best to win, whether that is what their actions seem to represent or not. You see someone flash forward, and you instantly have to assess whether this is a mistake, or them representing something terrible for you that for whatever reason you had not considered. This is where I think the poker analogy comes into play. The fact that a bluff can ever be said to be 'a good bluff' is representative enough of the idea that the 'wrong' play can actually be the 'right' play if represented properly, but it is still a risk because at its core it is the 'wrong' play.
You're talking about taking 60/40's and 80/20s, but the nature of that math is that sometimes the 20/80 is actually correct. An 'aggressive' player may fish for the 20/80, perhaps even knowing it could be a losing play, because the game is played against other human beings that think and react, and you effect how they think and react with your actions. Forcing any action is the only way possible to force the wrong action.
I can't refute that in a utopic environment it is impossible to refute your statement that 'the better player makes the correct play more often.' I will say though, that I think the ability to judge that correct play with hindsight is both unfair(because alternative play would fundamentally change every action thereafter), and essentially impossible to do with accuracy(partly because of the previously noted convolution).
There will always be debate about what the correct play is in all competitions between people, except maybe in a game like chess(I don't follow it). In the 1029382138129 games of LoL I've played I can't think of one example where after I've seen the replay that I'm still not sure if it was the correct or incorrect decision. Can you think of an example in a pro game where people argue over whether it was the right decision or not? I'm interested to see. When I'm reflecting in the middle of the game I never think about the outcome, because it's results-oriented. It's about making the correct decision the majority of the time regardless of the outcome. If I flash for a kill expecting my mid to follow up and he doesn't, and I die; it may mean it was a bad flash or it may mean it's the right play given a better midlaner. By the same logic the same decision can be incorrect at lower levels but correct at higher levels, but in both cases, there is a correct and incorrect decision. The probabilities are very rough statistical representations of instinctive decisions in LoL. The very fact that there are good players and bad players, and that there are good players who remain good consistently, supports the fact that this game is not random. Good players make accurate estimates, and hence correct decisions. This shouldn't be an argument.
That's not the argument though, or at least it wasn't, and you've only made it the argument by bringing large amounts of semantics about the terminology into play.
A play cannot be both incorrect and correct, that's my point. But the idea that different people(say, the mid laner and yourself in your own example) can conceive of a play being both is evidence enough of incomplete knowledge hampering the decision making process to an extent that makes manipulating it to your favor something players can and will do in different ways, which observers will call 'playstyle.'
|
On May 23 2014 08:26 zulu_nation8 wrote:
I think I understand your example but I don' think there's any situation where there's so much uncertainty that probability doesn't matter, or if that's what you're suggesting.
Maybe this will make sense. In a perfect world we can, instantaneously, perfectly evaluate all of the information before us, and use that to process a function which produces the optimal action for the given situation.
People who are "closer" at doing this could be seen as "better at the game". But that might not be the case. It could be the case that such a function does not exist that there isn't a "best move always"
A good example of this is Rock Paper Scissors. The "optimal play" is to play randomly every time. The problem is that the strategy space in RPS isn't transitive and playing randomly is hard. That is If S1>S2 and S2>S3 there is no guarantee that S1>S3. And since i know that other people won't necessarily be playing perfectly randomly, given their imperfect strategy there exists a strategy which beats them (but in turns is beaten by something else)
So what is the best play in RPS? Well its a bit complicated, the best play in RPS realizes that playing perfectly randomly is difficult, and attempts to determine a pattern in the other player while also realizing that if the other player is beating you, you have to start playing randomly. If playing randomly was hard then doing all that is really hard
League is potentially the same way. It could be the case that in game strategy is transitive, but it might not be.
I actually have an example of this from a game i played a long time ago. I was Ryze mid vs Ahri and at level 6 Ahri all-in's me and is about to win. This is a positive EV play for her. On her third R i am pretty sure i know which way she is going so i flashed forward towards her rather than away from her. She presses her R at the same time in the wrong direction and since her skills are on CD and mine are not, i kill her.
She says "who flashes forward?". This makes sense, flashing forward is, in all other situations a losing play, it takes your escape away and opens you up to be combo'd. Given that i knew how to dodge her R this no longer makes it a positive EV play for her. Had she known that my strategy was stronger she probably wouldn't have played, but there is no way to know that my strategy because until the strategy is tested it is unknown.
|
On May 23 2014 08:57 red_ wrote:
A play cannot be both incorrect and correct, that's my point.
That's my point from the start... A passive and aggressive "style" that both lead to the same outcome in the long run is nonsense.
On May 23 2014 08:57 red_ wrote:
But the idea that different people(say, the mid laner and yourself in your own example) can conceive of a play being both is evidence enough of incomplete knowledge hampering the decision making process to an extent that makes manipulating it to your favor something players can and will do in different ways, which observers will call 'playstyle.'
So you agree there is one correct play? Or are you saying that the individual decisions can be conflicting but both be correct from their own perspectives? That's a whole different argument. And if that's what you're saying, it can't happen. If there is one correct play, say in that case the midlaner is right and I'm wrong, then he made the better decision. After watching the replay and talking it out there should be no disagreement. There is "true" probability. Just because a low ezreal farming alone looks like a 100% kill doesn't mean it's actually 100%. The 100% you judged for yourself is incorrect. The "true" probability might be 0 if his whole team is nearby waiting. Like in all cases, the ability to determine the true probability is part of skill. It sounds complicated but really isn't. Good players will recognize a bait if it is one. There will always be signs.
What would playstyle be in either of these cases?
|
I knew buying Phantom Karthus would pay off, now to get that VU on Wonder Woman Sivir and my shitty skin collection upgrade is complete
|
On May 23 2014 09:06 Goumindong wrote:Show nested quote +On May 23 2014 08:26 zulu_nation8 wrote:
I think I understand your example but I don' think there's any situation where there's so much uncertainty that probability doesn't matter, or if that's what you're suggesting. Maybe this will make sense. In a perfect world we can, instantaneously, perfectly evaluate all of the information before us, and use that to process a function which produces the optimal action for the given situation. People who are "closer" at doing this could be seen as "better at the game". But that might not be the case. It could be the case that such a function does not exist that there isn't a "best move always" A good example of this is Rock Paper Scissors. The "optimal play" is to play randomly every time. The problem is that the strategy space in RPS isn't transitive and playing randomly is hard. That is If S1>S2 and S2>S3 there is no guarantee that S1>S3. And since i know that other people won't necessarily be playing perfectly randomly, given their imperfect strategy there exists a strategy which beats them (but in turns is beaten by something else) So what is the best play in RPS? Well its a bit complicated, the best play in RPS realizes that playing perfectly randomly is difficult, and attempts to determine a pattern in the other player while also realizing that if the other player is beating you, you have to start playing randomly. If playing randomly was hard then doing all that is really hardLeague is potentially the same way. It could be the case that in game strategy is transitive, but it might not be. I actually have an example of this from a game i played a long time ago. I was Ryze mid vs Ahri and at level 6 Ahri all-in's me and is about to win. This is a positive EV play for her. On her third R i am pretty sure i know which way she is going so i flashed forward towards her rather than away from her. She presses her R at the same time in the wrong direction and since her skills are on CD and mine are not, i kill her. She says "who flashes forward?". This makes sense, flashing forward is, in all other situations a losing play, it takes your escape away and opens you up to be combo'd. Given that i knew how to dodge her R this no longer makes it a positive EV play for her. Had she known that my strategy was stronger she probably wouldn't have played, but there is no way to know that my strategy because until the strategy is tested it is unknown.
How is ahri failing to recognize the only possible good play for you a good strategy?That just makes her dumb.
|
On May 23 2014 07:49 zulu_nation8 wrote:Show nested quote +On May 23 2014 07:35 Goumindong wrote:On May 23 2014 07:23 zulu_nation8 wrote:On May 23 2014 07:17 Legitimacy wrote:On May 23 2014 07:02 zulu_nation8 wrote:
If you watch the replay there's one correct decision in 99% if not 100% of all situations. When having to decide with incomplete information, the better player will make the correct decision more often.
You edited in a a priori after, I'm not sure how that applies.
What people see when they think of "hyperaggression" is a certain player trying to maximum or obtain an advantage. For example, if you are a caitlyn vs. lucian, "aggression" may look like the caitlyn missing cs at lv1 on purpose to try to autoattack the Lucian. The purpose is to establish lane dominance and make the Lucian blow his pot which will let you gain an advantage later on in the lane, which outweighs the gold you lose by missing those CS. A "passive" player will simply farm and try to get lv2 early, or try to have a safe laning phase of w/e. But it's the incorrect play as there's an advantage you're missing by either not having the mechanics or awareness to. In this case there's one right way to play, and that's to establish yourself as the alpha AD, aka "aggresiveness," the other way, by playing like a pussy, means ur bad. Those are not arbitrary qualities, they are extensions of skill. And now assume the Lucian fights back agains the Caitlyn level 1. Both are taking creep aggro and it's really a toss-up as to who's going to come up with the kill. What do you classify that as? Because it's the wrong play for both parties because it's really a 50/50 coin flip as to who's going to come out on top with the uncertainty that is creep aggro. Once you reach that point it's no longer skill-based but rather luck-based. Pros a lot of the time won't do this. Does that make them bad? It's not 50/50 unless it's the same AD playing against his clone. The better player with the better micro, reactions, decision making, etc, will win. If both players believe they can win an all in, which is the only reason they would and should go all in in the first place, then one of them is obviously wrong. I don't understand how any situation in league can be 50/50 let alone a lv2 all in. Pros don't do this in LCS because they're not playing against scrubs in solo queue who will lose the all in most of the time. Ok, so think about a person. They put an input into the computer. This input has some amount of precision associated with it (precision in terms of how close to the desired input was given). Additionally players are making decisions swiftly, these micro decisions (which way do i flash, how do i manage creep aggro, will I b in tower range after a flash, etc etc) also have errors because some portion of the time that same decision made again would have a negative rather than positive outcome, just because the situation was misjudged. So we have two types of error which do not have to deal at all, with the macro decision of 'do i go in or not' We have input error and we have micro decision error. If we have these two types of error, then the outcome of a macro decision is uncertain to some degree even when we are talking about small things like 'do i go in on this or not'. And this was before we get into game state randomness like "is the jungler there or not?", or "is that bush warded?" The 50/50 example is just a hypothetical to get you thinking. 90/10 examples happen all the time, 80/20 examples. Do you go in 90/10 to win a kill with no other repercussions other than your summoner? Do you go in 60/40 for a kill/be killed situation? One of these is not particularly aggressive, one of these is particularly aggressive. It comes down to the value placed in summoner skills, and the level of risk that the player is willing to absorb. These aspects of players are different, some take more risks, some take less risks. Some don't go HAM, some go HAM. Some are passive players, some are aggressive players Second part - I'll even go as far as to say the 60/40s don't happen that often either. The 60/40 example, I can think of one situation where player trait matters, but I'm not confident, that is if you die, you lose, and if the other player dies he loses.
There should never be a 60/40 situation for a good player, least of all lvl 1. Good players on confident champions should know when they have an advantage that is at least 90% success rate, not counting outside circumstances like a jungler sitting in lane brush.
|
What if you're TRYNDAMERE?!
|
On May 23 2014 09:16 GolemMadness wrote:
What if you're TRYNDAMERE?!
The soloq gods favour the confident pawns.All situations are favourable.
|
On May 23 2014 09:06 Goumindong wrote:Show nested quote +On May 23 2014 08:26 zulu_nation8 wrote:
I think I understand your example but I don' think there's any situation where there's so much uncertainty that probability doesn't matter, or if that's what you're suggesting. Maybe this will make sense. In a perfect world we can, instantaneously, perfectly evaluate all of the information before us, and use that to process a function which produces the optimal action for the given situation. People who are "closer" at doing this could be seen as "better at the game". But that might not be the case. It could be the case that such a function does not exist that there isn't a "best move always" A good example of this is Rock Paper Scissors. The "optimal play" is to play randomly every time. The problem is that the strategy space in RPS isn't transitive and playing randomly is hard. That is If S1>S2 and S2>S3 there is no guarantee that S1>S3. And since i know that other people won't necessarily be playing perfectly randomly, given their imperfect strategy there exists a strategy which beats them (but in turns is beaten by something else) So what is the best play in RPS? Well its a bit complicated, the best play in RPS realizes that playing perfectly randomly is difficult, and attempts to determine a pattern in the other player while also realizing that if the other player is beating you, you have to start playing randomly. If playing randomly was hard then doing all that is really hardLeague is potentially the same way. It could be the case that in game strategy is transitive, but it might not be. I actually have an example of this from a game i played a long time ago. I was Ryze mid vs Ahri and at level 6 Ahri all-in's me and is about to win. This is a positive EV play for her. On her third R i am pretty sure i know which way she is going so i flashed forward towards her rather than away from her. She presses her R at the same time in the wrong direction and since her skills are on CD and mine are not, i kill her. She says "who flashes forward?". This makes sense, flashing forward is, in all other situations a losing play, it takes your escape away and opens you up to be combo'd. Given that i knew how to dodge her R this no longer makes it a positive EV play for her. Had she known that my strategy was stronger she probably wouldn't have played, but there is no way to know that my strategy because until the strategy is tested it is unknown.
the optimal play is what works the majority of the time in the long run. In rock paper scissors that would be, from your example, going random. In league there is no such thing. If say her play was 80/20, vs. 80% of Ryzes who displayed the micro you displayed, is your elo, etc, it would've worked. Then you're the 20. Doesn't mean there is no optimal play because he failed in that particular example. The EV doesn't change mid decision. If she played the same situation 100 times vs. Ryzes of the same skill level then she would get 80 kills and die 20 times. If you think every other ryze at your elo would make the same play, then she miscalculated, but it doesn't seem to be the case.
Whatever the case is, there is an optimal play when all, and realistically, most of the factors are considered. So your example means either she got a bad outcome because you had some divine Faker like inspiration that would never happen in a million years, or she underestimated how easy it is to dodge her skillshots so that most would've made the same play as you,
|
|
|
|
|
|
EPT
