Not logged in
FAQ  •  Advanced search  •  Login

Optimal income curve in stats?

<<

olethuora

Fast Nova
Fast Nova

Posts: 31

Joined: 12.04.2012, 13:50

Post 13.04.2012, 12:36

Optimal income curve in stats?

It would be interesting to see the optimal income curve, so you can compare how good the players were to the best possible income. That curve could be pre-calculated somehow, in a manner that you first send mantas all the time, then toucans, etc, and never send anything bigger unless you spend the money just before round ends :)

This is how it would look like:
optimal1.png
optimal1.png (6.53 KiB) Viewed 2641 times


Where blue is the optimal income.

Sponsor

Post 13.04.2012, 12:36

Re: Optimal income curve in stats?

<<

bimm

Large Manta
Large Manta

Posts: 70

Joined: 13.01.2012, 14:37

Post 21.04.2012, 05:27

Re: Optimal income curve in stats?

..
Last edited by bimm on 28.04.2012, 02:41, edited 1 time in total.
<<

olethuora

Fast Nova
Fast Nova

Posts: 31

Joined: 12.04.2012, 13:50

Post 21.04.2012, 14:09

Re: Optimal income curve in stats?

Yeah we only need one curve calculated. It's not that hard to calculate, could do it "brute force" way too, so you just leave the game running with simple AI which predicts how many creeps of each class it can send at which income, and then just run it as long as the AI can't spend the money anymore: then the rest of the curve can be calculated easily to infinity amount of time.

Income from kills isn't necessary, and I don't think it would even make the curve much more higher, compared to real player curves, because players need to spend a lot money for defense, while the optimal curve doesn't. So I think without creep kills income add, the optimal curve would still always be higher than any player.

I wanted to see this curve for a reference point to see how efficient every player was; if the best player curve is a lot lower compared to optimal curve, it means it was very nooby game, so you can quickly judge how good players there were.
<<

bimm

Large Manta
Large Manta

Posts: 70

Joined: 13.01.2012, 14:37

Post 22.04.2012, 10:05

Re: Optimal income curve in stats?

..
Last edited by bimm on 28.04.2012, 02:41, edited 1 time in total.
<<

olethuora

Fast Nova
Fast Nova

Posts: 31

Joined: 12.04.2012, 13:50

Post 22.04.2012, 13:01

Re: Optimal income curve in stats?

You could then for example end up with only toucans as the best result, but it is still possible that a single zeus and the rest spent on mantas is better.

Interesting, didn't think about that at all. But by quick calculations, it won't work. Here is why: you can send around 80 creeps each round, 80*500 = 40k. If you send one zeus and rest on mantas, it means you have only ~140k income, and if you can send 80 creeps per round, 80*8000 = 640000, a lot higher than your income, so zeus isn't better.

I think the easiest way to calculate this, is to calculate how many of which class creeps you can send per round, including the pause coming from keyboard waiting, which is around 1 second. So, you can send 40k for mantas, 640k for toucans, 4.8M for titans, and rest for motherships.

If you have 6.2M income, you would simply calculate: 6.2 - 4.8 = 1.4M for class 4 creeps, which means: 1.4 - 1.25 = 1 mothership, 1 zeus (and 2 sharks if time for it), and 4.8M for titans.
<<

bimm

Large Manta
Large Manta

Posts: 70

Joined: 13.01.2012, 14:37

Post 23.04.2012, 07:34

Re: Optimal income curve in stats?

..
Last edited by bimm on 28.04.2012, 02:41, edited 1 time in total.
<<

olethuora

Fast Nova
Fast Nova

Posts: 31

Joined: 12.04.2012, 13:50

Post 23.04.2012, 13:14

Re: Optimal income curve in stats?

bimm wrote:We would have to know the exact number of creeps that can be sent each round beforehand, though. I think it's closer to 100, spending 50k credits on mantas seems to be possible.

I think the maximum is around 90 (after some tests). With 100 you can indeed spend the money, but you don't have any time to switch to the other class. You could spend 100 for mantas only, though!

bimm wrote:Hmm, am I dumb or are there really only 16*80=1280 ways to divide 80 creeps to the 16 creep-type-slots?

I don't know, but I thought to make this same way as cashier returns specific amount of change for you: first he subtracts the largest piece of money, then second largest, and so on. You don't need to know how many ways you can do that, you simply just subtract first the largest chunks off as long as you can until you can't anymore, then you move to smaller pieces until there is no money to subtract from. This could be calculated in a split of a microsecond, and if it notices the delays from key switching are too high, it will calculate it again, until it finds the optimal sequence. So, we don't really need it to be one math problem, it could be just solvable by simple AI like that. Even if it was super slow to do, it wouldn't matter because we only calculate it once after all ;)
<<

bimm

Large Manta
Large Manta

Posts: 70

Joined: 13.01.2012, 14:37

Post 24.04.2012, 00:55

Re: Optimal income curve in stats?

..
Last edited by bimm on 28.04.2012, 02:41, edited 1 time in total.
<<

florian

Mako
Mako

Posts: 14

Joined: 17.07.2011, 14:06

Post 24.04.2012, 12:32

Re: Optimal income curve in stats?

take a sheet of paper and a pen.... you can easily write that down if ur interestet in it..
1:200 2:240 3:260 4:310 5:340 6:370 7:405 8:445 9:495 10:540 11:590 12:645 13:705 14:775 15:850

and so on...its simply income divided by 50 in the beginning.
later u can divide by 500 or something but thats not really important I think.
<<

krunx

User avatar

Mothership
Mothership

Posts: 2172

Joined: 13.07.2011, 07:25

Location: Cologne

Post 25.04.2012, 08:51

Re: Optimal income curve in stats?

I think how close you can be in the optimal income curve, depends very much on how balanced the skill level of players is. And not just how well the individual players are.
<<

dcode

User avatar

Mothership
Mothership

Posts: 2200

Joined: 09.07.2011, 00:59

Post 25.04.2012, 14:49

Re: Optimal income curve in stats?

The optimal income curve will of course be always that of sending the most and weakest creeps you can of all the money you have and furthermore getting the bounty of all the creeps you receive instantly. Unfortunately, I assume this is so unrealistic and above a real income curve that it's not valuable.
Think it, design it, build it, run it. That's what I do.
<<

olethuora

Fast Nova
Fast Nova

Posts: 31

Joined: 12.04.2012, 13:50

Post 29.04.2012, 11:18

Re: Optimal income curve in stats?

dcode wrote:Unfortunately, I assume this is so unrealistic and above a real income curve that it's not valuable.

You can always scale it lower, you know.

Return to Suggestions

Who is online

Users browsing this forum: No registered users and 51 guests

© CreepTD.com · Powered by phpBB · Style by ST Software