Combat.... say WHAT!?

[Kuciwalker]

There's no magical bug that occurs at ~1% odds

[Virdrago]

I think the real reason most people get upset at losing at 95% is that they don't try to attack at 5% . Now, you're saying, why would I do that? Because if you decide to kamikaze all your units at something, be it 95 or 5, you'll notice good odds losing and some bad odds winning. Since it is extremely rare that anyone would bother trying to attack with more than 4 or 5 units who have a less than, say, 40% chance each time (most of us don't unless it's 70 or higher, or a siege unit), we only notice the high-percentage losses. How many times has anyone looked at the Combat Log and seen the AI losing to our units at 90% chances or higher? Generally, no-one bothers. But you can see it.

[Adagio]

Also note that when looking at the percentage it might not be correct. Often when I attack with my stack the winning percentage is based on unit X attacking unit Y, but the the actual attack is unit A against unit B
Too many times have I attacked where the odds says something like that:
+50% against melee (showing an axeman)
-----
+10% for combat I (showing a swordman)

That should give me some great odds... but as soon as the combat starts the units changes, so that I might end up with an axeman against an archer with that city protection promotion #3
So even though it still shows me having great chances of winning (it still shows the calculation for my unit against a swordman) the chances of winning are actually very low

[marvinkosh]

Originally posted by Adagio
Also note that when looking at the percentage it might not be correct. Often when I attack with my stack the winning percentage is based on unit X attacking unit Y, but the the actual attack is unit A against unit B

It's usually best to split up the stack before commencing an attack so that you can find the right unit to attack with. Otherwise it just throws them at the defenders, when what you actually want to do is very different.

For example, if you want to inflict a lot of collateral damage using your siege weapons to weaken all the defenders, it's better not to put them up against the best defender. So you might first send in a disposable unit with a city attack bonus, so the best defender loses at least some of his hit points and isn't picked for the next fight.

[Kuciwalker]

Collateral damage isn't correlated with the actual attack damage of the Cat, IIRC.

[Arrian]

I'll add some subjective experience of my own:

I've won at least one battle at about 3% odds of success.

It happens, it's just that smart human players generally do not even try such battles, as they look at the odds and (usually correctly) decide against attacking. In the case of my ~3% win, it was an old catapult I considered disposable, so I led with it (instead of my CRII Trebs) and, to my surprise, I won. Then I lost a treb

-Arrian

[Adagio]

Originally posted by marvinkosh
It's usually best to split up the stack before commencing an attack so that you can find the right unit to attack with. Otherwise it just throws them at the defenders, when what you actually want to do is very different.

Of course when I attacking I do split up my larger stack into smaller selections, so only the units I want to use to soften up their defense are selected, but sometimes it happens that I miss to deselect a catapult and instead of using my unit that has a 95% chance of winning against the single defender it uses the catapult which has like 5% chance of winning