|
On January 04 2017 02:50 Acrofales wrote:Show nested quote +On January 04 2017 00:51 Cascade wrote:On January 04 2017 00:32 Acrofales wrote:
You both seem to be missing the point of what it means to continue on until infinity. You're not stopping if you double up. You're stopping only if you go bust. It's like a Las Vegas algorithm. You will either terminate (you went bust) or keep on. The more you stack the odds towards profit, the smaller your chance becomes of actually going bust, and thus terminating in reasonable time. But the fact is still there: given an infinite amount of time it will actually happen.
Now if you want practical bounds, look at the literature on stochastic algorithms. yeah, that's the case with an expectation value of 0 or negative. Question is, with a positive expectation value (which you won't find in a casino), are you still certain that you will go broke eventually? If not, what is the risk? It either terminates (you go broke) or it doesn't. If you run for infinite time, you WILL go broke, even if there's only a one in a googolplex chance of that happening. It's what infinity is all about. So, let's say that on a loss you lose 1 dollar, and on a win you get 1000, then it's clear the expected value trends to infinity, and yet, because you don't give a shit about the expected value, but only about your termination condition (you're broke), you will literally only terminate when that happens. And in the infinite limit, that happens, because infinity is stupid like that. In PRACTICE, of course, that is entirely meaningless. But that wasn't the question.
That's such a dumb state mentioned Acrofales... Your statement is analogues to say in that the area under a normal distribution is infinite, because the curve has a positive non-zero value all the way up to infinity (which clearly it doesnt). I am certain the value is not infinity, I just don't know how to go about finding an answer.
|
On January 04 2017 07:08 FiWiFaKi wrote:Show nested quote +On January 04 2017 02:50 Acrofales wrote:On January 04 2017 00:51 Cascade wrote:On January 04 2017 00:32 Acrofales wrote:
You both seem to be missing the point of what it means to continue on until infinity. You're not stopping if you double up. You're stopping only if you go bust. It's like a Las Vegas algorithm. You will either terminate (you went bust) or keep on. The more you stack the odds towards profit, the smaller your chance becomes of actually going bust, and thus terminating in reasonable time. But the fact is still there: given an infinite amount of time it will actually happen.
Now if you want practical bounds, look at the literature on stochastic algorithms. yeah, that's the case with an expectation value of 0 or negative. Question is, with a positive expectation value (which you won't find in a casino), are you still certain that you will go broke eventually? If not, what is the risk? It either terminates (you go broke) or it doesn't. If you run for infinite time, you WILL go broke, even if there's only a one in a googolplex chance of that happening. It's what infinity is all about. So, let's say that on a loss you lose 1 dollar, and on a win you get 1000, then it's clear the expected value trends to infinity, and yet, because you don't give a shit about the expected value, but only about your termination condition (you're broke), you will literally only terminate when that happens. And in the infinite limit, that happens, because infinity is stupid like that. In PRACTICE, of course, that is entirely meaningless. But that wasn't the question. That's such a dumb state mentioned Acrofales... Your statement is analogues to say in that the area under a normal distribution is infinite, because the curve has a positive non-zero value all the way up to infinity (which clearly it doesnt). I am certain the value is not infinity, I just don't know how to go about finding an answer.
Philosophical Question for problems where you don't have a proof on: Which is more accurate, math that seems to give an answer you disagree with or intuiting the answer based on gut feeling and what your brain things is a better answer?
|
On January 04 2017 02:50 Acrofales wrote:Show nested quote +On January 04 2017 00:51 Cascade wrote:On January 04 2017 00:32 Acrofales wrote:
You both seem to be missing the point of what it means to continue on until infinity. You're not stopping if you double up. You're stopping only if you go bust. It's like a Las Vegas algorithm. You will either terminate (you went bust) or keep on. The more you stack the odds towards profit, the smaller your chance becomes of actually going bust, and thus terminating in reasonable time. But the fact is still there: given an infinite amount of time it will actually happen.
Now if you want practical bounds, look at the literature on stochastic algorithms. yeah, that's the case with an expectation value of 0 or negative. Question is, with a positive expectation value (which you won't find in a casino), are you still certain that you will go broke eventually? If not, what is the risk? It either terminates (you go broke) or it doesn't. If you run for infinite time, you WILL go broke, even if there's only a one in a googolplex chance of that happening. It's what infinity is all about. So, let's say that on a loss you lose 1 dollar, and on a win you get 1000, then it's clear the expected value trends to infinity, and yet, because you don't give a shit about the expected value, but only about your termination condition (you're broke), you will literally only terminate when that happens. And in the infinite limit, that happens, because infinity is stupid like that. In PRACTICE, of course, that is entirely meaningless. But that wasn't the question.
I don't think you are correct. I think for your argument to work you need the assumption that all infinite sums eventually converge, which very clearly is not the case.
It may make sense to look at it like someone above described. How high are the chances of going from 100 to 200 vs broke. How high are the chances of then going from 200 to 400 vs broke. Etc...
This clearly gives you an infinite sum of probabilities of going broke. If you add all of those up, you get the absolute probability of going broke. And if that sum converges to a value of less than 1, that means that there is a non-zero chance of never going broke.
I see your argument that from an algorithmical point of view, the algorithm only terminates once you are broke. But it is very possible to have an algorithm that simply never terminates. That is just a shitty algorithm, but it is easy to write one.
The problem in this case is that the probabilities are pretty annoying to calculate, as you have to deal with summed binomial distributions of non-constant length. You will probably have to very liberally work with > and < to get any result with this problem.
|
ZigguratOfUr
Iraq16955 Posts
|
Yes, that seems to be it. Thank you.
|
On January 04 2017 07:24 ZigguratOfUr wrote:Isn't this just a Gambler's Ruin problem? So your chance of getting infinitely rich (assuming the E.V is positive) is just 1-(q/p)^i with p = probability of winning q = 1-p and i=original capital. See: http://www.columbia.edu/~ks20/FE-Notes/4700-07-Notes-GR.pdfedit: Actually no, you're winning different amounts in your example so that makes it trickier. edit 2: For unequal payouts this paper details a general formula: https://arxiv.org/pdf/1209.4203.pdf
Ok skimming the second paper, that seems complicated enough for me not to feel bad to not have solved it myself.  thanks for finding.
On January 04 2017 02:50 Acrofales wrote:Show nested quote +On January 04 2017 00:51 Cascade wrote:On January 04 2017 00:32 Acrofales wrote:
You both seem to be missing the point of what it means to continue on until infinity. You're not stopping if you double up. You're stopping only if you go bust. It's like a Las Vegas algorithm. You will either terminate (you went bust) or keep on. The more you stack the odds towards profit, the smaller your chance becomes of actually going bust, and thus terminating in reasonable time. But the fact is still there: given an infinite amount of time it will actually happen.
Now if you want practical bounds, look at the literature on stochastic algorithms. yeah, that's the case with an expectation value of 0 or negative. Question is, with a positive expectation value (which you won't find in a casino), are you still certain that you will go broke eventually? If not, what is the risk? It either terminates (you go broke) or it doesn't. If you run for infinite time, you WILL go broke, even if there's only a one in a googolplex chance of that happening. It's what infinity is all about. So, let's say that on a loss you lose 1 dollar, and on a win you get 1000, then it's clear the expected value trends to infinity, and yet, because you don't give a shit about the expected value, but only about your termination condition (you're broke), you will literally only terminate when that happens. And in the infinite limit, that happens, because infinity is stupid like that. In PRACTICE, of course, that is entirely meaningless. But that wasn't the question.
Sorry for rubbing it in, but "because infinities are stupid" isn't a very solid argument. :D they can be counterintuitive, but that's even more reason to make careful arguments. yes, you do get infinitely many chances to go broke, but there is also a smaller chance to go broke each iteration as your expected wealth goes up. So not obvious that the sum of probability of going broke over the infinite iterations goes to 1. Indeed, the above paper calculates the sum and it's below 1.
|
|
|
On January 07 2017 01:15 JimmiC wrote:
I was thinking about comparing e-sports to "real" sports and this is what I came up with.
SC is like tennis, 1v1 game of strategy and skill.
League (mobas) are like Basketball 5v5 with each team having different rolls (pg,sg, sf pf and C in basketball AND top, mid, jungle, adc and support in League)
Cod is like football big number on numbers all guys have specific rolls.
Thoughts?
I've always narrowed it down to a more focused thing actually.
RTS (SC, BW, etc...) is like being the quarterback of a football team. You have to learn to use your pieces to maximize an overall goal, make adjustments on the fly, have your own mechanical input, but in the end its about getting all your pieces to move the way you want them to in real time.
MOBA (League, DOTA, etc...) is like the offensive/defensive linemen. Position based play based on territory control and making "breakthrough" plays to pierce defensive walls.
FPS (COD, Gears, etc...) is more like field goal kicks. More based on accuracy, zone positioning, and a "defend the bomb" type dynamic.
|
On January 07 2017 01:15 JimmiC wrote:
I was thinking about comparing e-sports to "real" sports and this is what I came up with.
SC is like tennis, 1v1 game of strategy and skill.
League (mobas) are like Basketball 5v5 with each team having different rolls (pg,sg, sf pf and C in basketball AND top, mid, jungle, adc and support in League)
Cod is like football big number on numbers all guys have specific rolls.
Thoughts?
well, they are similar in that it is 1on1 or 5on5. But not sure the comparison stretches much further. Why is sc like tennis more than badminton? Or squash? Or boxing? Or was tennis just the first 1on1 sport that came to mind? And similarly, why would a moba be more like basketball than hockey? Or is it more like soccer, even though the number of players isn't the same?
I mean... maybe you see something I missed, but if all you got here is "1on1 game of strategy and skill", or team sport with different roles, then you fit essentially all sports. So please carry on, but I think I want a bit more detail on why you picked out tennis and basketball out of all sports.
|
|
|
How is right guard different from left guard?...
soccer does not have positions that over lap...
|
|
|
Does it need to match the exact number of players? Does it even have to be competition sports?
|
|
|
On January 07 2017 09:49 JimmiC wrote:
Sport yes, numbers don't have to match, however you want to relate the esport to sport is fine
Then I would say an RTS is like football (assuming you only gave a fuck about the quarterback) or baseball.
Both sports are very pieces driven centering on a pivot point player or perspective. (The QB or the Pitcher)
In both sports the pivot point player dictates the direction and pace of the game, and the other "pieces/players" are used to maximize the pivot player's strength's and weaknesses.
A MOBA would be most akin to curling. A team of players with overspecialized roles have to develop a pattern based group task wherein the job of each member of the team are of equal importance at scoring the game.
FPS would be most like basketball. A bit chaotic, but very individualistic. Players are given general tasks, positions, or responsibilities--but often enough you kind of have to move based on the flow of the match as you jump between defense and offense on an individual level. Every person is both the support and the carry and changes depending on how the mood of the match changes.
Fighting games would be the one most akin to Tennis. Pattern based endurance runs where its about being the player who makes the least mistakes in long drawn out sessions.
|
Street fighter is like UFC 
|
|
|
On January 07 2017 01:15 JimmiC wrote:
I was thinking about comparing e-sports to "real" sports and this is what I came up with.
SC is like tennis, 1v1 game of strategy and skill.
League (mobas) are like Basketball 5v5 with each team having different rolls (pg,sg, sf pf and C in basketball AND top, mid, jungle, adc and support in League)
Cod is like football big number on numbers all guys have specific rolls.
Thoughts? I think Sports have much more simplified means to achieve objectives, and the objectives themselves are very simple in relation to eSport. This is why Sports is easier to follow.
The SC vs Tennis is the most far apart. The strategy involved in tennis is very, very, very limited in relation to SC. Chess vs SC might be more similar.
COD break the mold of the complexity/simplicity in relation to the first mentioned.
League vs Basketball is the only one that is somewhat comparable to your aim, and I suspect that this was the one you came up with initially.
|
what would be a good analogy for competitive starjewelled then ?
|
On January 07 2017 20:26 opisska wrote:
what would be a good analogy for competitive starjewelled then ?
+ Show Spoiler [twister] +
I has roughly the same amount of players. They pretend a bit to be real sports, but when you actually look at it, it turns out it's just a silly game, but if you look even closer you see that there is some strategy and skill to it. Most post people get bored after a few games.
|
|
|
|
|
|
EPT![[image loading]](http://sad.hasbro.com/db7312c97e69f2aa2a48e9c156bbc05885942775/e80625c59f576f511fb28aca83719ac3.png)
