|
|
On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level?
I run AD Reds Armor Yellows MR or Flat CDR Blues and AD Quints on Riven but the Flat CDR Blues are mainly for vs AD Mid, Top, and Jungle.
|
On January 05 2015 23:35 Uldridge wrote:
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level?
Some attack speed quints are actually the standard runing (on most adcs). Both 3 attack speed quints or 2 attack speed 1 ad quint are most popular.
Edit: you can also go attack speed blues for matchups in which you don't really need MR. For instance vs Janna + some adcs.
|
And vods too, I think!
I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea.
On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level?
What is the format you're trying to express with the Riven runes?
Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD.
|
the selection of words kind of confuses me, does this mean that slow is reduced to 20% from 50%?
Stalker's Blade
Passive - Chilling smite now steals 20% movement speed for 2 seconds rather than reducing enemy champions movement speed by 50% for 2 seconds.
|
On January 06 2015 00:20 M2 wrote:
the selection of words kind of confuses me, does this mean that slow is reduced to 20% from 50%?
Stalker's Blade
Passive - Chilling smite now steals 20% movement speed for 2 seconds rather than reducing enemy champions movement speed by 50% for 2 seconds.
Instead of it being a 50% slow it is now a 20% slow that gives you the speed it takes like BotRK's active does.
For instance, if the target has 400MS and you Smite it, it loses 80MS and you gain 80MS for the duration. Movement speed steals allow Riot to make chasing through slows remain strong while not crippling a targets ability to dodge skillshots. 50% slow on smite is stupid strong and I'm not surprised it's being changed now that Stalker's is seeing widespread play.
|
with Riven, I just wanna do as much damage as possible early. Snowball as hard and fast as I can.
with Caitlyn, same thing really, but I feel more attack speed runes than normal would benefit her more than ad. Or armor/mr/hp
|
Well that sucks my subscription to ogn already renewed for January
|
On January 06 2015 01:43 Slusher wrote:
Well that sucks my subscription to ogn already renewed for January
ayyy me too...
|
Whats the current consensus for which jungle item to pick for rengar? From what I see it's split pretty evenly among pros between blue and purple (don't know names).
|
On January 06 2015 03:21 SwizzY wrote:
Whats the current consensus for which jungle item to pick for rengar? From what I see it's split pretty evenly among pros between blue and purple (don't know names).
I play a lot of rengar and for me purple is superior. If you execute your ganks correct you just dont need blue. But Blue is legit and viable, so overall it should be a matter of taste :-)
|
|
|
On January 05 2015 23:53 Sponkz wrote:Show nested quote +On January 05 2015 07:27 Goumindong wrote:
Ok suppose we have three variables y,x and z. And we want to run a model to figure out y=xB+ e. Let y and e be a nx1 vector and x and z be nxk and B be kx1 vector. E is our error term. We assume x'e = 0 x'z and z'y =\= 0.
If we solve y=xB +e for B we multiply by x' and invert and get B=(x'x)^-1 x'y Ok pretty simple.
But if we have y= xB+zG+e instead (and G isn't zero by z'y is not zero). Then we do the operation and get B= (x'x)^-1 x'y + (x'x)^-1 x'zG. The last term of which is bias.
You may say "but I am not doing a regression" and he answer is that yes, you are, you just have a specific form of the x matrix which lets you separate everything. But it doesn't nullify the bias problem. Which will exist if dragons correlate with gold and if dragons correlate with winning. Now, you can definitely check to see if dragons correlate with gold and if it doesn't you're golden. But we know that they correlate with winning (first dragon has what kind of win %?)
*technically we are working on expectations but I don't want to bother writing that out over and over again. I know i only got a D in math, but holy shit that post is pure garbage. You're throwing equations here and there, but you aren't making ANY sense.
Sounds like standard economics
|
On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. Show nested quote +On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD.
not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live
|
On January 06 2015 05:16 IamPryda wrote:Show nested quote +On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD. not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live
Why wouldn't you be able to watch Riot's Twitch channel through the Twitch app?
|
On January 06 2015 05:17 red_ wrote:Show nested quote +On January 06 2015 05:16 IamPryda wrote:On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD. not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live Why wouldn't you be able to watch Riot's Twitch channel through the Twitch app?
twitch app doesn't do vods i don't think
|
On January 06 2015 05:41 Nos- wrote:Show nested quote +On January 06 2015 05:17 red_ wrote:On January 06 2015 05:16 IamPryda wrote:On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD. not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live Why wouldn't you be able to watch Riot's Twitch channel through the Twitch app? twitch app doesn't do vods i don't think
I think it's still getting rebroadcasted, just through Riot's channel and not OGNs.
|
On January 05 2015 23:53 Sponkz wrote:Show nested quote +On January 05 2015 07:27 Goumindong wrote:
Ok suppose we have three variables y,x and z. And we want to run a model to figure out y=xB+ e. Let y and e be a nx1 vector and x and z be nxk and B be kx1 vector. E is our error term. We assume x'e = 0 x'z and z'y =\= 0.
If we solve y=xB +e for B we multiply by x' and invert and get B=(x'x)^-1 x'y Ok pretty simple.
But if we have y= xB+zG+e instead (and G isn't zero by z'y is not zero). Then we do the operation and get B= (x'x)^-1 x'y + (x'x)^-1 x'zG. The last term of which is bias.
You may say "but I am not doing a regression" and he answer is that yes, you are, you just have a specific form of the x matrix which lets you separate everything. But it doesn't nullify the bias problem. Which will exist if dragons correlate with gold and if dragons correlate with winning. Now, you can definitely check to see if dragons correlate with gold and if it doesn't you're golden. But we know that they correlate with winning (first dragon has what kind of win %?)
*technically we are working on expectations but I don't want to bother writing that out over and over again. I know i only got a D in math, but holy shit that post is pure garbage. You're throwing equations here and there, but you aren't making ANY sense.
Sorry. Posting on a phone makes it hard to be precise. Are you familiar with linear algebra?
So to start off with we are talking about vectors or matricies. These are basically collections of numbers and we put them in a format that makes them easy to deal with. A matrix has rows and columns and so an n by k matrix has n rows and k columns. Each one has a number. " ' " is a symbol that is used for many things but the most common with regards to matrices is "transpose".
So if x is an n x 1 vector then its a list of numbers stacked atop each other . Then x' is a 1 x n vector, a list of numbers one after another, in the same order as x is stacked atop each other. x'x is the dot product of x and x. We have to take the transpose because we're working in a different notation (I.E. x*x would be different than x'x). Its essentially the sum of x^2. The assumption E(x'e)=0 is standard in regressions. Its equivalent to the assumption that E(e|x)=0 or that x is uncorrelated with the error term. For a geometric/visual example it is because the dot product of two vectors which are orthogonal is zero. Orthogonality is equivalent to no correlation.
To give a simple example we can give the above math an analogue to a simple average where y is what we want to average. B= (x'x)^-1 x'y. x is a n x 1 vector of 1s and y is an n x 1 vector of values. So B = Sum(y) / n. Which is the average.
The solution above where we look at y = xB + e is simply a general form of a linear regression, which is the general form of basically all summary statistics and most regression analysis before you start throwing things for loops. I use it because 1) most everyone who has statistics/regression/scientific regression training should be familiar with it* and 2) if you are then it quickly and clearly illustrates the intuition for problems in other models, even if the math is imprecise in those models 3) its not so complicated that its not hard to recall/reproduce from memory or to explain
*"Mathematical statistics" courses will focus on probabilities and distributions. But anyone in a scientific background will start with that for their regression analysis. Exceptions to the specific form may occur if people took courses which did not rely/require linear algebra. Linear algebra significantly reduces the amount of work required to write and analyze the math and so most places should require it before examining statistics in even a semi-serious way.
I am, more or less, assuming that sufficiency and the rest in this discussion are familiar with linear algebra. Sorry for talking over your head
|
On January 06 2015 05:53 Gahlo wrote:Show nested quote +On January 06 2015 05:41 Nos- wrote:On January 06 2015 05:17 red_ wrote:On January 06 2015 05:16 IamPryda wrote:On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD. not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live Why wouldn't you be able to watch Riot's Twitch channel through the Twitch app? twitch app doesn't do vods i don't think I think it's still getting rebroadcasted, just through Riot's channel and not OGNs.
He also specifically said watching the games live. I am genuinely confused what he thinks the issue is.
|
|
|
On January 06 2015 06:00 red_ wrote:Show nested quote +On January 06 2015 05:53 Gahlo wrote:On January 06 2015 05:41 Nos- wrote:On January 06 2015 05:17 red_ wrote:On January 06 2015 05:16 IamPryda wrote:On January 06 2015 00:07 Gahlo wrote:And vods too, I think! I know people have said Riot has their hands in too many pies with the pro scene, but I think they've done a good job when it comes to Korea. On January 05 2015 23:35 Uldridge wrote:
2 questions on runing:
Someone like Riven: flat ad mixed with arpen or flad ad + armor/mr + health or arpen + armor/mr + health?
On adc's with big range, so like only Caitlyn, maybe Varus, wouldn't it be more interesting to have a bit more attack speed than standard runing allows? Or is the most optimal just preference/matchup dependent from a certain level? What is the format you're trying to express with the Riven runes? Anyway, my generic page on her is AD/Scaling Health/Scaling CDR/AD. not a fan I usually watch the pro games on my ipad when I first get to my office through the twitch app, now I have no way of watching the games live Why wouldn't you be able to watch Riot's Twitch channel through the Twitch app? twitch app doesn't do vods i don't think I think it's still getting rebroadcasted, just through Riot's channel and not OGNs. He also specifically said watching the games live. I am genuinely confused what he thinks the issue is.
I misread and thought riot would not be broadcasting the games at all on twitch and just through there own website.
|
|
|
|
|
|
EPT
