No way to tell what the probability is by just running tests, unless there is some very strict probability variator (as in the case of the if y different than a after n-1 attempts, then y=a).
You could, however, limit even more the factoes, and see if the probability increases, this tying another factor to its triggering.
You could do this with original HP difference; turn of occurence etc
From what i gather atm you have already seen no trigger the first turn, or if the attacking unit has less HP. You could examine the cases where the defend ini was triggered, and see if there is another set factor there.
Originally probability theory was developed to predict the outcome of games (eg throwing of the dice) and although in reality it is not very good at doing that (since there are physical forces at play there; how you threw the dice) it can predict the purely theoretical outcome for a near-infinite number of times the event takes place.
So having it trigger 10 times in 400 tests, does not mean that the probability is 2,5%. It could have triggered more times, or less. Even if you had run 1 billion tests you would still not have the exact probability (i do not remember, infact, if for so large numbers of tests there is a diagram that predicts the progression of the actual probability in relation to the one which is being falsely witnessed, but i suspect that there is).
Your best bet is to try to examine if there are even more set factors
) that i had been called "the virtuoso of probability theory". But i am not quite that nowdays...
so I don´t understand most of it.
