The Scavenging Guide - The Future of Early Game Farming

  • Thread starter DeletedUser121448
  • Start date

Deleted User - 11481044

Guest
Please find an analysis of the scavenge feature in the following jupyter notebook:
https://nbviewer.jupyter.org/github/mikenoethiger/tw-scavenge/blob/master/scavenge.ipynb

The notebook includes python code for computing return (resources), duration (seconds) and performance (resources/second) of a scavenge run. Moreover, a method for calculating optimal distribution of given capacity to level 1,2,3,4 scavenge in order to reach maximum performance is presented. Rather than solving the problem analytic, an algorithmic approach (using scipy.optimize.minimize function) was used to do the job.

The notebook is also available on GitHub.

Conclusion: Sending an equal amount of units in every level is almost always the most profitable approach.
 

JawJaw

Awesomest CM Ever
Reaction score
2,210
Please find an analysis of the scavenge feature in the following jupyter notebook:
https://nbviewer.jupyter.org/github/mikenoethiger/tw-scavenge/blob/master/scavenge.ipynb

The notebook includes python code for computing return (resources), duration (seconds) and performance (resources/second) of a scavenge run. Moreover, a method for calculating optimal distribution of given capacity to level 1,2,3,4 scavenge in order to reach maximum performance is presented. Rather than solving the problem analytic, an algorithmic approach (using scipy.optimize.minimize function) was used to do the job.

The notebook is also available on GitHub.

Conclusion: Sending an equal amount of units in every level is almost always the most profitable approach.

Love this. Great work!
 

QueenB

Member
Reaction score
11
Please find an analysis of the scavenge feature in the following jupyter notebook:
https://nbviewer.jupyter.org/github/mikenoethiger/tw-scavenge/blob/master/scavenge.ipynb

The notebook includes python code for computing return (resources), duration (seconds) and performance (resources/second) of a scavenge run. Moreover, a method for calculating optimal distribution of given capacity to level 1,2,3,4 scavenge in order to reach maximum performance is presented. Rather than solving the problem analytic, an algorithmic approach (using scipy.optimize.minimize function) was used to do the job.

The notebook is also available on GitHub.

Conclusion: Sending an equal amount of units in every level is almost always the most profitable approach.
Wow are you statistician lol nice work
 

Hobbes

New Member
Reaction score
0
scavengers 101 guide... each troop kind collects most efficiently at low numbers... !!!

only guide you need to know is to then divide each troop count into 3 or 4, depending on whether you have the 10k open yet, (why would you spent 30k ressources to open this, any kind of "early"game??), and then send each 3rd or 4th groups to scavenge as a team of the different troop kinds you plan to use :)

y voila, u scavenge most efficiently. (scavenge time is solely dependent on haul number of the unit, so using slow swords and archers and axes makes much sense if you are raiding everything on the map with 1k fast cavalery and 3k spears :).
 
Last edited:

Bobah

Well-Known Member
Reaction score
10
Line Graph to Distinguish Most Optimal Method at each amount of spears
09fcf1adbeaf0ca05a2cfbeb6a69ec6f.png

tl;dr until i have 300 spears its better to only use Tier 2 & 3 instead of all three, correct?
 
Top