shuri, there is a big difference in both what and how you stated your first post and your last
Where? I give two different examples with different kinds of units which only strengthens my point. There's a discrepancy no matter what you choose.
third, while i get your point that fake numbers shouldn't matter in a formula, they do matter in consistency of testing in realistic scenarios
if you put in something that can't possibly happen, it doesn't matter what values you calculate and the devs won't care
Fair enough - so from now on, I'll strive to keep the population below 4k with examples. I can assure you it makes absolutely no difference though.
fourth, i don't believe the formula is linear and this may be part of your extrapolation issue
fifth, "Notice that in each battle all grepo calculates is the proportion of units that are destroyed on either the attacking or defending side. So magnitude of units doesn't matter." - i believe this may be false
The examples I've given so far show it's probably true or at least close to true. If it's false, then a counterexample would be nice to see. In addition, what issue - I see none with what I've said?
second, there is also a difference in what you posted: slings vs arch, slings vs swords; losses can not be compared as they have different defensive strengths/ weaknesses
...
if you kill olu, it matters if it's slings/ hops (1 pop) vs horses (3) vs char (4)
It doesn't make a difference what particular def. units are being concerned if they're all killed. All that matters is the def. total. If I'm attacking only ranged units, I only need to check the ranged def. total. This is true no matter what multipliers are applied (including wall).
_
_
_
Same applies for mixed attack, but it's harder to create examples (below, sharp/ranged def totals only matter):
_
Similarly, it doesn't make a difference what off. units are concerned if they're all killed. Only the totals for each stat matter.
_
I can assure you similar examples can be constructed and I have yet to find something which counters this. Population/types of troops (except catapults if there is a wall) do not matter in calculation. Only atk/def totals matter. First the simulator determines which side wins. If it's the attacking side, then it calculates the proportion of troops killed for each of blunt, sharp, ranged. If it's the defending side, it just calculates the proportion of troops killed and distributes it evenly. How it deals with rounding - I'm not sure, but that's how it works as far as I can see. I've tried out many different numbers, and they all point to this conclusion.
sixth, if the example you gave with "real numbers", you changed both the wall and the defense so that tells me nothing
to compare, there have to be 4 sets for you to show me the calculations are off: low wall with high and low defense, high wall with high and low defense
I see you don't understand what the first example shows. The wall at lv 25 gives a buff of 141.9% to defensive troops (according to wiki). Supposing this was true, then 10000 swords would have the same strength behind a lv 25 wall as 21419 swords (ignoring extra base defence). However, the values are off by quite a margin which cannot be accounted for by base def/rounding.
i believe i've added enough to this project as i don't see it going anywhere
I don't see how you've added anything. If you believe what I've said isn't true - then fair enough if you can give a counterexample. However, all you've made is a bunch of unjustified claims which are contradicted by the examples i've shown.
I'm not saying the wiki values are
definitely wrong, but I'm saying they are
likely to be, given what I've shown. This is a request for the devs to check if they are wrong, and the wiki to be corrected if so. Of course this 'project' can't go anywhere without that happening since this is all just observation/speculation.