Creep kills...

[Edward]

Just wondering, on a balanced competitive match, how many CK should you have by, say, 5-10-15-20-30-40 min?

Sorry, didn't say it before: As a regular last hitting carry.

[ToxicHobo]

Depends on if you're support/carry/neutralling and what hero you are (gonna be harder to farm with chronos than tort for example)

[DreadTalon]

For a mid level player such as myself I consider that at 10 minutes 20 CS is a bad farm, 30 is medium and 40 is good, as a solo I will usually get between 40-50 at 10 mins. At 20 mins between 60 and 80, if your a carry you would probably have over 100.

That is for laning, jungling will give you more, if I jungle say legionnaire I will have 60 CS at 10 mins.

[thewishkah]

play zephyr and aim for 150-200 every 10 mins.

[N0th_]

srsly you guys are insane O.o having double cs than min time in pub game on side lane 4 me is not sucking that hard xD

[svirrz]

Shouldnt that depend on how good your enemy is when it comes to denying?

[balmain]

a carry is considered to be farming well should be at ~100 at 20 min

[balmain]

play zephyr and aim for 150-200 every 10 mins.

i dont think thats possible.. if you mean the first 10 minutes.

*sry for the double post, thought it would automatically merge

[Opax]

play zephyr and aim for 150-200 every 10 mins.

you only average 137 ck's pergame and your favorite hero is zephyr. Is that really realistic?

[WaRDeN]

80 creep kills at 20 minutes used to be a good benchmark for a carry, i think it still stands

[Opax]

lets count it then!
every 30 secs a creepwave spawns (5 creeps).
the first creep wave starts getting lasthitted/denied at 30 secs.
so after 5 minutes there will be about 9-10 creepwaves (~45-50 enemy creeps)
Well, i doubt that even the best dota player in the world would get everysingle one of the creeps because he has to be careful of ganks/harasses. even when babysitted, i think it would probably be on an average of only 25%(when playing against chain stuns) or 50% (when the enemy is playing passive) of the creeps getting last hitted.
You will have to compete for every single creep with 2 enemy heroes (and sometimes your lane mate too)
if 20 minute mark is around 200 creep spawns, I think it is reasonable to get around 80 ck if they don't gank you(or your teammates protecting you) and your lanemate is letting you have all of the creeps.

[FuzzyWuzzy]

You have to take into account that the CS should rise exponentially.

You should be 100+ in 20 minutes as a carry in the ideal case.

However, you should be 300+ in 45+ minutes, because the number of creeps and your farming speed increase a lot in that time.

[akitoes]

You have to take into account that the CS should rise exponentially.
.
Please no :(
http://www.mathwarehouse.com/exponential-growth/images/formula_exponential_growth.gif

All jokes aside I agree.

[FuzzyWuzzy]

Please no :(
http://www.mathwarehouse.com/exponential-growth/images/formula_exponential_growth.gif

All jokes aside I agree.

Yep.

A is a constant. B is increasing linearly due to more creeps. X is increasing linearly too due to more items.

Thus, A*(B^X) is exponential.

:D

[Prixm1]

It depends on the other team. The lanes etc. I get around 100-150 almoste every game. And they maybe last for 30mins or something donno.

[Sirroelivan]

Yep.

A is a constant. B is increasing linearly due to more creeps. X is increasing linearly too due to more items.

Thus, A*(B^X) is exponential.

:D

Except B is a constant as well, otherwise it wouldn't be an exponential function. ;)

[cvk]

if you are playing against equally skilled player at mid, you should average around 30-35 creeps at 10 minutes, and 50-70 at 15 minutes. After that you should start getting a lot more creeps since you usually have ability to travel between lanes and farm jungle easy, so 100 is pretty much a benchmark for a good farm game at 20-25 minutes. If you playing more ganks then ur cs will suffer but you will make up with kills, so you should have only around 50-80 creeps with ganks.

[FuzzyWuzzy]

Except B is a constant as well, otherwise it wouldn't be an exponential function. ;)

It's not a constant, it increases with time as more creeps start spawning.

I.e if you have 4 creeps in the beginning and an arbitrary farming speed, lets say X, you get 4^X.

After a while you get 8 creeps spawned with a farming speed of 3X -> 8^(3X), which is exponential.

[china]

TBH, I judge how well I'm farming by when I get blink, bkb, etc. It's easier for me cause it includes those kills and assist gold I get.

[evotech]

Depends on how good the guys you play against are, how goo lane you have, how good they are at harassing / denying etc

[KaWs]

play zephyr and aim for 150-200 every 10 mins.


i lol'd

[Extreme_Cake]

It's not a constant, it increases with time as more creeps start spawning.

I.e if you have 4 creeps in the beginning and an arbitrary farming speed, lets say X, you get 4^X.

After a while you get 8 creeps spawned with a farming speed of 3X -> 8^(3X), which is exponential.
Aren't you making a fairly unsound assumption with regards to the effect of your items on your farming speed? Exponential growth implies a regular progression, which isn't what you're going to get.

[tabako]

On most carry heroes, your creeps/min should normally increase as the game progresses. It is certainly not exponential though, and the people arguing that is exponential (mobzombie) are using bad assumptions to get such a model.

[Extreme_Cake]

It has some similarities to an exponential increase, to be fair. It just isn't strictly one.

[shivz]

what about denies? i denie about 15 per game (public)
After that it doesnt realy matter imo

[thewishkah]

you only average 137 ck's pergame and your favorite hero is zephyr. Is that really realistic?

i wasn't serious.. *whoooosh*

High CK's from zephyr is averaged with minimal CK's when i play support/gank heroes?

[oOKateOo]

deny count is less important. you should always go for last hits (LH) on enemy creeps instead of your own because denied creeps still give experience to heroes in range. denying becomes better when the enemies is oor. also, i distinguish between 'actual denies' and 'effective denies': an actual deny is a +1 on your CK label, but an effective deny is any attack that messes up the enemy's LH -- so long as they don't get the gold, you've done fine

it's possible to over-deny, too. what if you're trying to drop a tower? ya need those creeps. or soulstealer needs souls. or bloodhunter needs health. or demented shaman needs more bounce damage

and remember, everything is relative. 50 CS at 20 minutes is bad, but not if they only have 25.

a lil math for ya:

in one lane, 4 creeps (3 melee, 1 archer) spawn every 30 seconds from 0:00 to 16:00. at 16:00, 5 creeps (4 melee, 1 archer) every 30. at 40 minutes, add 1 archer and 1 melee for 7. 1 catapult will spawn every 4th wave after the first 6 minutes: 2 catapults will spawn every 4th wave past 40 min

if you have 100% last hit accuracy for a 30 minute game, you will have killed 279 creeps. there are variables, of course: are you soulstealer? did you nuke down every wave then kill the easy & normal camps during your free time? how many snotters did you kill? did you portal to far-pushed lanes with multiple waves? were you creep stacking?

anyway, hth :)~

[lumino`]

Aren't you making a fairly unsound assumption with regards to the effect of your items on your farming speed? Exponential growth implies a regular progression, which isn't what you're going to get.

What are you jibbering. Get some basic calculus lesson.

Exponential growth simply means a non-linear growth that increases over time.
A differential of an exponential growth is not linear perse.

Think about that!

The guy with the "function" is wrong too though, it misses a time component.

Best aproximation for a creeping function is.

(A * B^X)*T. Where a is a constant (gold from creeps), b is amount of creeps spawned, x is modifier based on items/level, and T is the time variable.
B and X are also variant and dependant on T.
Therefor the entire gold gain is an addition of exponential time functions, ranging from T0 to T1, ....., T(n-1) to Tn.

[FuzzyWuzzy]

What are you jibbering. Get some basic calculus lesson.

Exponential growth simply means a non-linear growth that increases over time.
A differential of an exponential growth is not linear perse.

Think about that!

The guy with the "function" is wrong too though, it misses a time component.

Best aproximation for a creeping function is.

(A * B^X)*T. Where a is a constant (gold from creeps), b is amount of creeps spawned, x is modifier based on items/level, and T is the time variable.
B and X are also variant and dependant on T.
Therefor the entire gold gain is an addition of exponential time functions, ranging from T0 to T1, ....., T(n-1) to Tn.

That's what I tried to explain, but I guess Calculus 101 was 5 years ago, so I can be forgiven :D

[ACatInTheHat]

Something has always bothered me. This something is like a hole in my soul, gaping like an open grave that I try to pretend is not there. But I can't pretend, not anymore. I have to ask: WHAT THE HELL DOES CS STAND FOR? I just can't make sense of it!

While we're at it, "SS" for someone missing has always eluded me with its true meaning as well. But that's for another story, under another moon.


VV cheers. :)

[Rocxx]

WHAT THE HELL DOES CS STAND FOR? I just can't make sense of it!


creep score

[akitoes]

While we're at it, "SS" for someone miSSing
___________

[tabako]

Exponential growth simply means a non-linear growth that increases over time.


No.

Any function that is increasing (positive first derivative), which also has increasing rate of increase (positive second derivative) will satisfy what you said - but that does not make a function exponential.

For example, f(x) = x^2 is not exponential despite satisfying what you said. It has non-linear growth, and the rate of that growth is increasing in time.



Best aproximation for a creeping function is.

(A * B^X)*T. Where a is a constant (gold from creeps), b is amount of creeps spawned, x is modifier based on items/level, and T is the time variable.
B and X are also variant and dependant on T.
Therefor the entire gold gain is an addition of exponential time functions, ranging from T0 to T1, ....., T(n-1) to Tn.


No, this is a terrible approximation.

The amount of creeps that spawn in the game closely resembles a linear function in time. Even if an increase in your "B" coefficient rapidly took you towards getting every creep that spawns (a bad assumption), the total amount of creep kills a player gets as a function of time is still so far from an exponential function its ridiculous.

[Proph3t1]

No.

Any function that is increasing (positive first derivative), which also has increasing rate of increase (positive second derivative) will satisfy what you said - but that does not make a function exponential.

For example, f(x) = x^2 is not exponential despite satisfying what you said. It has non-linear growth, and the rate of that growth is increasing in time.



No, this is a terrible approximation.

The amount of creeps that spawn in the game closely resembles a linear function in time. Even if an increase in your "B" coefficient rapidly took you towards getting every creep that spawns (a bad assumption), the total amount of creep kills a player gets as a function of time is still so far from an exponential function its ridiculous.



You know it's sad when people start using math to analyze HoN creep waves.

[lumino`]

No.

Any function that is increasing (positive first derivative), which also has increasing rate of increase (positive second derivative) will satisfy what you said - but that does not make a function exponential.

For example, f(x) = x^2 is not exponential despite satisfying what you said. It has non-linear growth, and the rate of that growth is increasing in time.


True my wording was improperly chosen.

An exponential function is a function where every derivative still is a non-linear function.



No, this is a terrible approximation.

The amount of creeps that spawn in the game closely resembles a linear function in time. Even if an increase in your "B" coefficient rapidly took you towards getting every creep that spawns (a bad assumption), the total amount of creep kills a player gets as a function of time is still so far from an exponential function its ridiculous.Why is it a terrible approximation. A general usage of exponential functions is when the base has a direct influence on later outcomes. Simply said, the more creeps that have spawned, and are killed, the more creeps you will kill in a shorter time period.

X obviously is capped, and does not have to take astronomical proportions. Also X is variant in time segments, but can be constant on various values of T. Hell X can even be 1, or less then one if it fits the equation. The fact is that creeping becomes much faster, the longer the game lasts.
For which a scalar would not do proper contribution to what it becomes.

Only T has normal intervals in this equation. X and B do not have this property but are still variant.

[LightRain]

Yep.

A is a constant. B is increasing linearly due to more creeps. X is increasing linearly too due to more items.

Thus, A*(B^X) is exponential.

:D
stop trolling the math majors

edit - just so we're all clear, f(x) = x^2 is a power function, or a polynomial, whichever you prefer. g(x) = a^x is an exponential function.
Your creep stat is a sum of a number of functions, each of which describes a certain segment of time, since the creeps spawn in discrete quantities that increase at fixed intervals (discontinuous!). Each function is your (creep kills/minute) * time farming. These functions can be approximated by a coefficient of farming (what ratio of available creeps you're likely to get) multiplied by the number of available creeps. Possibly the simplest way to model this is to graph your ck/min and take the integral.

Note that certain heroes see sharp jumps in their ck/min. Soulstealer, for example, reaches a coefficient of almost 1 when he has the mana to support farming via nuke and is in a single lane, and if you add Posthaste, you greatly increase the number of creeps available to him (albeit he cannot kill every single possible creep on the map)(inb4l2p). Each creep kill gives experience and gold, so we can see that ck/min depends in part on its own integral, and by FTC its derivative will depend in some part on itself. Giving the nature of what we are describing and the previous fact, we can conclude that part of ck/min is an exponential function (and creep stat, the integral of a partly exponential function, is also a partly exponential function, that is if ck/min is integrable at all).

As we continue to examine this function, we see that it has a linear part (passive gold gain) and a rather hard to predict part that adds sudden variance to the gold/exp we have available and thus can increase/decrease the rate at which we improve our farming coefficient (by this I refer to ganking and tower kills). Also, the level cap, the restriction of six item slots, and the cap on number of creeps together indicate that eventually all terms in the ck/min function become constant, and cs becomes a simple linear growth function. (alright, alright, human variance means that it will be a series of functions that average out to linear) Given this, and if not the eventual demise of one team's fountain, then certainly the eventual demise of the server hosting the game, we see that our time parameter is bounded, and thus creeps stat is, eventually, a bounded function.

This is not very rigorous but I'm going to get enough of rigorous over the next few weeks (exam on the first possible day and exam on the last possible day, exams spaced regularly in between. Better than all at once, I guess).

[akitoes]

stop trolling the math majors

edit - just so we're all clear, f(x) = x^2 is a power function, or a polynomial, whichever you prefer. g(x) = a^x is an exponential function.
Your creep stat is a sum of a number of functions, each of which describes a certain segment of time, since the creeps spawn in discrete quantities that increase at fixed intervals (discontinuous!). Each function is your (creep kills/minute) * time farming. These functions can be approximated by a coefficient of farming (what ratio of available creeps you're likely to get) multiplied by the number of available creeps. Possibly the simplest way to model this is to graph your ck/min and take the integral.

Note that certain heroes see sharp jumps in their ck/min. Soulstealer, for example, reaches a coefficient of almost 1 when he has the mana to support farming via nuke and is in a single lane, and if you add Posthaste, you greatly increase the number of creeps available to him (albeit he cannot kill every single possible creep on the map)(inb4l2p). Each creep kill gives experience and gold, so we can see that ck/min depends in part on its own integral, and by FTC its derivative will depend in some part on itself. Giving the nature of what we are describing and the previous fact, we can conclude that part of ck/min is an exponential function (and creep stat, the integral of a partly exponential function, is also a partly exponential function, that is if ck/min is integrable at all).

As we continue to examine this function, we see that it has a linear part (passive gold gain) and a rather hard to predict part that adds sudden variance to the gold/exp we have available and thus can increase/decrease the rate at which we improve our farming coefficient (by this I refer to ganking and tower kills). Also, the level cap, the restriction of six item slots, and the cap on number of creeps together indicate that eventually all terms in the ck/min function become constant, and cs becomes a simple linear growth function. (alright, alright, human variance means that it will be a series of functions that average out to linear) Given this, and if not the eventual demise of one team's fountain, then certainly the eventual demise of the server hosting the game, we see that our time parameter is bounded, and thus creeps stat is, eventually, a bounded function.

This is not very rigorous but I'm going to get enough of rigorous over the next few weeks (exam on the first possible day and exam on the last possible day, exams spaced regularly in between. Better than all at once, I guess).

Can I have your babies ? Especially seeming how I suck at math this year. I've always been a **** at following lectures scholarly >>

What got you interested in mathematics ? How do you approach scholar math ?

[LightRain]

Like many math majors, high school math was easy for me, so I went into uni for math :P
There's a bunch of stuff that's really intuitive for me (like limits, and induction) so I find it fun to do, and there's other stuff (like groups and rings) that really intrigues me and gives me something to aim for.
The best thing is to have good professors. I've been pretty lucky in that regard.

[akitoes]

Like many math majors, high school math was easy for me, so I went into uni for math :P
There's a bunch of stuff that's really intuitive for me (like limits, and induction) so I find it fun to do, and there's other stuff (like groups and rings) that really intrigues me and gives me something to aim for.
The best thing is to have good professors. I've been pretty lucky in that regard.
Oh I see :) Math can be stimulating when you're not force-fed I agree, but this year I feel like there's some kind of void in math classes. Recently I've been trying to seek out fun in maths by myself tho:)

Btw guys I'm stuck on these :

Question : Determine f''(x).

1)f(x) = 1/(x-1)
2)f(x) = 1 / (x?-4x+3)

I'm probably making a stupid mistake, For 1) I get -2x+2/(x-1)^3...
which is wrong >>

K FOUND OUT MISTAKE

PS:
What do you guys do in high school ?
Here in Paris we have mandatory Calculus I.
I heard you guys can do calculus 2 in HS ? :o

[LightRain]

Well, I'm from Ontario. We used to have a class that was "Advanced Functions + Relations, and Introductory Calculus", but I was in the last year to take that. Now it's just Advanced Functions and relations. Although, some people I talk to who TA for first year courses say that they prefer the students to come in knowing nothing, because they also don't have misconceptions or bad habits that way. And whether or not you did calc in high school, you start from scratch in first year anyway.
Also, I'm pretty sure your HS math courses will be better than mine were. Still, if you feel they're boring, by all means seek out some fun stuff on the side. What interests you? Calculus, geometry, algebra, number theory?
Oh yeah, I think you can still do calc in high school here if you take Advanced Placement classes (which are like intro to first year university stuff) but I never took those so I don't know what you study in them.

[schila]

I got an average of 4 a min, throughout all my games, and i usually play carry.
My usual rule of thumb is; 2 cs/min is horrible, 3 cs/min is okay, 4 cs/min is nice, 5 cs/min is ideal. You will usually never reach 5 unless you are a solo midder, or you get a supporter who gives you all the creeps. And even then its still hard to maintain. Imo aim for 4 cs/min for the first 30 min of the game, from there on your cs should rise much faster due to towers being pushed down and way more creeps spawning.

[HUSO1]

4-5/cs min

depends on lane enemy.

[Blet1]

4-5 cs minute really doesnt cut it as a carry :( however I cant deny that's what you usually end up with :(

[Evre1]

Like many math majors, high school math was easy for me, so I went into uni for math :P
There's a bunch of stuff that's really intuitive for me (like limits, and induction) so I find it fun to do, and there's other stuff (like groups and rings) that really intrigues me and gives me something to aim for.
The best thing is to have good professors. I've been pretty lucky in that regard.

AKA too lazy for an applied major :)

And I agree, being force-fed math kills the interest, especially if you already know what's being taught.

On topic: Instead of creep kills, whats a good gold/min for carry/support/etc?

[Zoomz1]

All you people who say ck are exponential are right in the sense taht you'll kill them faster but you wont be killing THAT much more creeps after the laning phase is gone and its mostly group fights.

With that said it depends entirely on your carry.

If you're a hard carry that needs items your creep score is either going to be boss or you better hope you were fed alot of kills/towers.

If you have a semi-carry (like puppet) you add versatility mid/early game so you should be helping out with ganks so your creep score will not be as great as it COULD be.

All that said 100 in 20 minutes is pretty unrealistic in a scrim setting if the other team is decent and goes around ganking.

I'd say 80-90 is much more realistic

[Proph3t1]

All you people who say ck are exponential are right in the sense taht you'll kill them faster but you wont be killing THAT much more creeps after the laning phase is gone and its mostly group fights.

With that said it depends entirely on your carry.

If you're a hard carry that needs items your creep score is either going to be boss or you better hope you were fed alot of kills/towers.

If you have a semi-carry (like puppet) you add versatility mid/early game so you should be helping out with ganks so your creep score will not be as great as it COULD be.

All that said 100 in 20 minutes is pretty unrealistic in a scrim setting if the other team is decent and goes around ganking.

I'd say 80-90 is much more realistic



100 cs is very realistic in 20 minutes IF you are the designated carry of the team (meaning you TOTALLY ignore any team fight for the first 20 minutes). Yamateh had 300 CS in like 35 minutes with 8 tower kills against XTC.NYC with QoP (Wretched Hag).



However, for a carry to get that farmed that quick there has to be alot of coordination of wards, tps, etc. to defend the carry whenever a gank is coming.

[LightRain]

AKA too lazy for an applied major :)
damn straight