Unit Calculators

[RefluxSemantic]

Hmm, I'd think comparing Jans and Musks is going to be somewhat reasonable though. Huss vs coyotes is probably not accurate anymore. It's just an estimate though, to get some rough idea of what stats mean.

[Kaiserklein]

Not really. If you have two units with 100 hp 10 attack costing 100 res vs one unit with 200 hp 20 attack costing 200 res, the big unit wins with 25% hp left. Yet that's equal resources investment on both sides, and according to the "formula" the efficiency is the same for both units.
So it can be wrong by a margin of 25% which basically means it gives zero insight. 25% is huge, probs the gap between the best and the worst units in the game

Yes, but then you realize that you're usually comparing units who are only at worst ~20% apart in hp. So it's actually pretty reasonable.

Nah it's never reasonable, the formula is meaningless.

[RefluxSemantic]

Are you talking about whatever the OP is about? Or about just dividing the attack by villager seconds and the hp by villager seconds?

[RefluxSemantic]

I honestly don't think HP*Att is really true either.

It is a good theoretical approximation of the stength of a unit i think , a unit A beats a unit B if it kills it before being killed (obivous i know^^), so if:

(hp_A/dps_B) > ( hp_B/dps_A ) corresponding to "time to kill unit A by unit B" bigger than "time to kill unit B by A"
which is equivalent to:
hp_A*dps_A > hp_B*dps_B

The problem is that I think this assumed you fight units one by one. In the hypothetical scenario of a unit A that has twice the stats of unit B, it would imply that unit A is four times as strong as unit B. This is the case if unit A fights unit B for times, but not when unit A fights four unit B's. In that case, it's 'only' three times as strong. I'd definitely argue the latter scenario is more realistic.

I am honestly annoyed by the fact that I just can't figure out the mathematical formula for this. That'd make life a bit easier.

[louis293]

I honestly don't think HP*Att is really true either.

It is a good theoretical approximation of the stength of a unit i think , a unit A beats a unit B if it kills it before being killed (obivous i know^^), so if:

(hp_A/dps_B) > ( hp_B/dps_A ) corresponding to "time to kill unit A by unit B" bigger than "time to kill unit B by A"
which is equivalent to:
hp_A*dps_A > hp_B*dps_B

The problem is that I think this assumed you fight units one by one. In the hypothetical scenario of a unit A that has twice the stats of unit B, it would imply that unit A is four times as strong as unit B. This is the case if unit A fights unit B for times, but not when unit A fights four unit B's. In that case, it's 'only' three times as strong. I'd definitely argue the latter scenario is more realistic.

I am honestly annoyed by the fact that I just can't figure out the mathematical formula for this. That'd make life a bit easier.

yes exactly the part you quoted from my post is meaning ONE unit A figths ONE unit B, i tried then to make resonning to explain the general formula used in the software but the example of Kaiserklein shows that this formula ins't correct at all, so we simply shouldn't use this formula or software at all, especially if u compare very different units which will be similar to this example as said by Diarouga for example hussards/coyote so i agree now with them that this formula is clearly too far from reality and meaningless

[aligator92]

Not really. If you have two units with 100 hp 10 attack costing 100 res vs one unit with 200 hp 20 attack costing 200 res, the big unit wins with 25% hp left. Yet that's equal resources investment on both sides, and according to the "formula" the efficiency is the same for both units.
So it can be wrong by a margin of 25% which basically means it gives zero insight. 25% is huge, probs the gap between the best and the worst units in the game

Yes, but then you realize that you're usually comparing units who are only at worst ~20% apart in hp. So it's actually pretty reasonable.

One of us does not get the point.
Kaiser shows that comparing stats/cost can be off by 25 %. So it is entirely useless if you use it to compare units who are less than 25 % apart in strength, which is all of them. Because the error of the appraoch is bigger than the difference you are trying to detect with it.

[RefluxSemantic]

Not really. If you have two units with 100 hp 10 attack costing 100 res vs one unit with 200 hp 20 attack costing 200 res, the big unit wins with 25% hp left. Yet that's equal resources investment on both sides, and according to the "formula" the efficiency is the same for both units.
So it can be wrong by a margin of 25% which basically means it gives zero insight. 25% is huge, probs the gap between the best and the worst units in the game

Yes, but then you realize that you're usually comparing units who are only at worst ~20% apart in hp. So it's actually pretty reasonable.

One of us does not get the point.
Kaiser shows that comparing stats/cost can be off by 25 %. So it is entirely useless if you use it to compare units who are less than 25 % apart in strength, which is all of them. Because the error of the appraoch is bigger than the difference you are trying to detect with it.

I think you didn't get my point?

Being off 25% in this hypothetical scenario is caused by the fact that one unit has far more HP than the other. A bigger unit with the same stats per cost will indeed be stronger. My point is that in aoe3 units of the same type generally aren't that far apart in HP. The smaller the difference in HP, the smaller the effect that causes you to be 'off'. So yes, in a scenario were you are comparing a unit with twice the HP and attack, you'll be pretty far off. But this effect won't be too big when its a difference in hp of ~10%. That's my point. The effect kaiser describes will be 5 times smaller (assuming this is linear, I am not sure it is though), when you're comparing a unit that's 20% bigger in stats. So for 20% bigger in stats, you might be off by 5%, which is fine for a rough approximation. Also, one can just do this sort of back of the paper calculation and keep in the back of their mind that this sort of effect is present.

[duckzilla]

The calculator is crap. Use my approach and be op.

[aligator92]

I think you didn't get my point?

Being off 25% in this hypothetical scenario is caused by the fact that one unit has far more HP than the other. A bigger unit with the same stats per cost will indeed be stronger. My point is that in aoe3 units of the same type generally aren't that far apart in HP. The smaller the difference in HP, the smaller the effect that causes you to be 'off'. So yes, in a scenario were you are comparing a unit with twice the HP and attack, you'll be pretty far off. But this effect won't be too big when its a difference in hp of ~10%. That's my point. The effect kaiser describes will be 5 times smaller (assuming this is linear, I am not sure it is though), when you're comparing a unit that's 20% bigger in stats. So for 20% bigger in stats, you might be off by 5%, which is fine for a rough approximation. Also, one can just do this sort of back of the paper calculation and keep in the back of their mind that this sort of effect is present.

Yes, this was not clear from your other post, thanks for clearing that up.

[Kaiserklein]

Are you talking about whatever the OP is about? Or about just dividing the attack by villager seconds and the hp by villager seconds?

It's basically the same. You're talking about attack / cost and hp / cost. OP is talking about (attack / cost) x (hp / cost) = (attack x hp) / cost². So he really just multiplies both

So for 20% bigger in stats, you might be off by 5%, which is fine for a rough approximation.

Assuming this is true, it's still not a fine approximation at all. 5% "efficiency" difference is big. It's probably what makes the difference between an average unit and an OP unit, e.g a musk and a sepoy.
For example 5% should be roughly the difference between a musket and a rusket, since ruskets have 20% less stats but are 25% cheaper. Well, if you simulate it, you'll see ruskets theoretically end up shitting on muskets, for an equal investment.

[Kaiserklein]

I am honestly annoyed by the fact that I just can't figure out the mathematical formula for this. That'd make life a bit easier.

Well, it's more or less in that wikipedia link. It's a differential equation that takes into account some parameters that depend on the units stats. Not sure it's worth modelling because of overkill though.

[Dsy]

Here is the formula which fixes unit drop effect:

Cost effectivity=(dps x effHP x (n+1)x(n/2))/ cost)

n= number of units

Im lazy and dont really want to program it though.

[helln00]

spirit of the law has a modified lanchester square law equation for smaller scale, maybe that helps with reducing the error margins.

edit: in addition it uses the time to kill an unit to account for strength differences which helps to cover other errors as well