Combat Odds

[MondSemmel]

+-5% is more just the standard deviation for a test that is inconclusive. That is, if we do 100 or 200 trials and only get a 5% discrepency, it's probably just luck of the draw, not an actual problem with the odds calculation.

The only fudge that appears to be happening is with really large numbers of first strike... numbers that aren't actually doable in the game. 4-7 just isn't going to make almost any difference, though it is regrettable the calculator doesn't seem to take them into quite the proper effect.

But wouldn't a 5% discrepancy be between e.g. a displayed chance to lose of 4% (i.e. 96% win chance) and a real chance to lose of 4.2%?
The numbers you quoted (89% and 94%) are almost a 100% deviation...


That being said, I'd have to agree that pursuing the First Strike test in this form doesn't make any sense because nothing in the game has so many First Strikes.
What would I have to do to test whether the 4-7 First Strikes and things like that are calculated properly?

 

[Ecofarm]

For a simple look at the impact of first strikes:

1) The game assumes 50% of strikes hit, right?

2) WB. Create 2 units. Give one two strikes more than the other and create a target. Exit worldbuilder.

3) Check combat odds.

4) The difference between the two represents:

The increased percentage of winning assuming one extra strikes hit.

Now, the percentage on the guy with 2 less strikes are what your percentage is if you miss both of your extras.

It will give you an idea of the range we are talking about when the code gives you percentages for a striker. Sometimes, missing all your strikes can result in a percentage 15 or 20 less than displayed.

That's why strikers die so much. They flip coins that change percentages or, actually, move it deviations away from the percentage displayed - which is actually the median of a range. The percentage might say ">99%" but if you miss all your strikes it's more like 80%. There will be a few times you miss all your strikes, and you can't do 80% too often...

Thus, I don't promote Kithra drill4 anymore. He dies fast like that.

 

[Nicopol]

that sounds interesting.
is it possible to display a range (e.g. 80% - 97% chance) if u (or enemy) have first strikes?

edit:
nevermind - there must be a better solution

 

[xienwolf]

I still find myself always looking at the strength numbers instead of the percentages anyway. Takes a deliberate effort to look at the odds instead of guessing for myself that I am willing to risk a 3.8 vs 4.2, or a 10.0 vs 9.8

 

[MondSemmel]

I'm willing to conduct further tests, but could anyone post the code of how combat is calculated and how the odds are calculated beforehand/tell me where I can find those?
I imagine it would make it much easier to understand whether a discrepancy between odds and results is possible in general (it is, as shown in the extreme number of first strikes example)/can actually occur with normal values in a real FFH game.


On a different note, creating a promotion that makes units survive all fights above, say, 98% or 99% odds, but then with 1 (or 0, if that doesn't instantly kill them) HP (perhaps by giving them 100% withdrawal if and only if their combat odds are this high) - that way, heroes wouldn't be as dependent on luck/the odds anymore, but in the cases where you should have died but didn't die due to the promotion, your hero would still be weak until healed fully (which can take some time in an opponent's territory)...
This would be different to the immortal promotion in that it would really only help you in the cases where you should win in any case. Losing heroes at 99% odds is annoying though still possible - the whole odds system is fine for normal units (losing one of one hundred champions at 99% odds isn't nearly as bad as losing one hero in hundred 99% fights), but "super" units (heroes, world units, perhaps national units) should probably be able to manipulate odds from 98% or 99% to a 100% survival rate via promotions or something like that.

EDIT: If First Strike chances are a problem, and they probably are, why not get rid of these altogether (change them to normal First Strikes)? Perhaps Drill with 7 First Strikes would be a bit much, so one could reduce the number to 6 or so, but in general First Strike chances are more trouble than they are worth. They basically give you a chance to have a chance to deal damage...

 

[sixs_monkey]

I would certainly use an available promotion slot on "The James R. Kirk school of not losing" He may not be the best example, but being able to have my legendary units escape by the skin of their teeth with only the shirts on their back and those not quite intact would be much more palatable than "We had it all over those guys! What is it with the Yamato, anyway?!"

Having no idea if the combat code has a step where an auto-success retreat, conditional on the inital odds of battle would fit nicely in, I'd also buy one that depended on initial hits. Then you could have select units or even unit-types that could be beaten back but never killed on the first try.

For me, this option would remove the frustration of having no way to prevent my enemy from winning the lotto.