Defense or Attack STATS?

[Elros]

So in MTW it would be impossible to kill a unit with a higher defence than your attack? Seems a bit odd.

[Ciaran]

No, it is possible, but it takes longer. And that means, the unit with the higher defense can fight on and inflict damage on its attacker. And if they can do damage faster than they receive, well, it´s obvious, isn´t it?
However, that´s the very basic, there´s also factors like morale, fatigure and terrain coming into play.

[player1]

On average attack is better, since sometimes, units can attack, but not be attacked, like at the begining of flanking, when attacked units doesn't face flankers. Or when two units attack one unit.

But, in prolonged fight, face to face, one to one, they should be equal.

[Elros]

Well, if you can make your attack significantly higher than the enemies defence, attack will prevent them attacking at all if you charge them. Or at least, it'll stop the first rank or two. If you manage to do that then most enemies would rout and you'd hopefully avoid a prolonged battle altogether.

[Puzz3D]

The (attack of the striker) - (defense of the target) is a number that appears in the exponent of a constant. I think that constant is 1.1 in Rome. That expression is then multiplied by another constant which was 1.9% in MTW. So the basic algorithm is something like:

chance to kill = 1.9% * 1.1 ^ (att - def + combat modifiers)

The exponent would be limited to a range of -41 to +41.
This would give chance to kill a range of .04% to 96%

[Elros]

Ah, yeah that makes much more sense. So a peasant against an elephant would have to attack 200 times. I'm guessing kill actually means remove a hit point? Then it would be 2000 times. On average of course.

[Mr Frost]

The (attack of the striker) - (defense of the target) is a number that appears in the exponent of a constant. I think that constant is 1.1 in Rome. That expression is then multiplied by another constant which was 1.9% in MTW. So the basic algorithm is something like:

chance to kill = 1.9% * 1.1 ^ (att - def + combat modifiers)

The exponent would be limited to a range of -41 to +41.
This would give chance to kill a range of .04% to 96%

The lethality number {which is my name for it , given C/A forgot to name it properly ;p} definitly effects the chance to kill .
I use it to not only facilitate longer fights {which is common} but also to better represent the relative lethality of a given weapon so that with modifyed attack rates they can allow attack factor/skill to represent purely skill level , which I find more intuitive and pleasing .

I'm not sure what its' effect is exactly {couldn't give you numbers with confidence} , but I did notice that it seemed to mesh with changes in attack rate {time between animation cycles} reasonably well atleast for between 0.1 - 0.5 lethality and 0 - 25 attack rate ; ie increasing .4 to .5 lethality with attack rate 25 was reasonably similar in effect to lowering 25 to 20 attack rate with .4 lethality .

[Puzz3D]

The lethality number {which is my name for it , given C/A forgot to name it properly ;p} definitly effects the chance to kill.

Yes I forgot about lethality. That's a new combat factor in Rome.

[econ21]

On average attack is better, since sometimes, units can attack, but not be attacked, like at the begining of flanking, when attacked units doesn't face flankers. Or when two units attack one unit.

This is a very good point, although I suppose the reverse sometimes applies - you can get attacked and can't attack back. This may be particularly important regarding the armour component of defence, which helps against missiles. However, given that the AI is not so good at flanking or using missiles, I agree with you that attack is better.